US20210216246A1 - Results processing circuits and methods associated with computational memory cells - Google Patents
Results processing circuits and methods associated with computational memory cells Download PDFInfo
- Publication number
- US20210216246A1 US20210216246A1 US17/081,869 US202017081869A US2021216246A1 US 20210216246 A1 US20210216246 A1 US 20210216246A1 US 202017081869 A US202017081869 A US 202017081869A US 2021216246 A1 US2021216246 A1 US 2021216246A1
- Authority
- US
- United States
- Prior art keywords
- rsp
- data line
- stage
- stages
- causing
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Abandoned
Links
Images
Classifications
-
- G—PHYSICS
- G11—INFORMATION STORAGE
- G11C—STATIC STORES
- G11C11/00—Digital stores characterised by the use of particular electric or magnetic storage elements; Storage elements therefor
- G11C11/21—Digital stores characterised by the use of particular electric or magnetic storage elements; Storage elements therefor using electric elements
- G11C11/34—Digital stores characterised by the use of particular electric or magnetic storage elements; Storage elements therefor using electric elements using semiconductor devices
- G11C11/40—Digital stores characterised by the use of particular electric or magnetic storage elements; Storage elements therefor using electric elements using semiconductor devices using transistors
- G11C11/41—Digital stores characterised by the use of particular electric or magnetic storage elements; Storage elements therefor using electric elements using semiconductor devices using transistors forming static cells with positive feedback, i.e. cells not needing refreshing or charge regeneration, e.g. bistable multivibrator or Schmitt trigger
- G11C11/413—Auxiliary circuits, e.g. for addressing, decoding, driving, writing, sensing, timing or power reduction
- G11C11/417—Auxiliary circuits, e.g. for addressing, decoding, driving, writing, sensing, timing or power reduction for memory cells of the field-effect type
- G11C11/419—Read-write [R-W] circuits
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F3/00—Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
- G06F3/06—Digital input from, or digital output to, record carriers, e.g. RAID, emulated record carriers or networked record carriers
- G06F3/0601—Interfaces specially adapted for storage systems
- G06F3/0628—Interfaces specially adapted for storage systems making use of a particular technique
- G06F3/0655—Vertical data movement, i.e. input-output transfer; data movement between one or more hosts and one or more storage devices
- G06F3/0659—Command handling arrangements, e.g. command buffers, queues, command scheduling
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F3/00—Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
- G06F3/06—Digital input from, or digital output to, record carriers, e.g. RAID, emulated record carriers or networked record carriers
- G06F3/0601—Interfaces specially adapted for storage systems
- G06F3/0602—Interfaces specially adapted for storage systems specifically adapted to achieve a particular effect
- G06F3/0604—Improving or facilitating administration, e.g. storage management
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F3/00—Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
- G06F3/06—Digital input from, or digital output to, record carriers, e.g. RAID, emulated record carriers or networked record carriers
- G06F3/0601—Interfaces specially adapted for storage systems
- G06F3/0628—Interfaces specially adapted for storage systems making use of a particular technique
- G06F3/0638—Organizing or formatting or addressing of data
- G06F3/0644—Management of space entities, e.g. partitions, extents, pools
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F3/00—Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
- G06F3/06—Digital input from, or digital output to, record carriers, e.g. RAID, emulated record carriers or networked record carriers
- G06F3/0601—Interfaces specially adapted for storage systems
- G06F3/0668—Interfaces specially adapted for storage systems adopting a particular infrastructure
- G06F3/0671—In-line storage system
- G06F3/0683—Plurality of storage devices
-
- G—PHYSICS
- G11—INFORMATION STORAGE
- G11C—STATIC STORES
- G11C7/00—Arrangements for writing information into, or reading information out from, a digital store
- G11C7/10—Input/output [I/O] data interface arrangements, e.g. I/O data control circuits, I/O data buffers
- G11C7/1006—Data managing, e.g. manipulating data before writing or reading out, data bus switches or control circuits therefor
-
- G—PHYSICS
- G11—INFORMATION STORAGE
- G11C—STATIC STORES
- G11C7/00—Arrangements for writing information into, or reading information out from, a digital store
- G11C7/18—Bit line organisation; Bit line lay-out
-
- G—PHYSICS
- G11—INFORMATION STORAGE
- G11C—STATIC STORES
- G11C8/00—Arrangements for selecting an address in a digital store
- G11C8/16—Multiple access memory array, e.g. addressing one storage element via at least two independent addressing line groups
Definitions
- the disclosure relates generally to a computational memory element and in particular to a computational memory element array having results processing circuitry.
- Memory cells have traditionally been used to store bits of data. It is also possible to architect a memory cell so that the memory cell is able to perform some simple logical functions when multiple memory cells are connected to the same read bit line. For example, when memory cells A, B, and C are connected to a particular read bit line and are read simultaneously, and the memory cells and read bit line circuitry are designed to produce a logical AND result, then the result that appears on the read bit line is AND (a,b,c) (i.e. “a AND b AND c”), where a, b, and c represent the binary data values stored in memory cells A, B, and C respectively.
- the read bit line result is the logical AND of the data stored in those memory cells.
- the memory cells may be subdivided into a plurality of sections and each section may have a plurality of bit line sections that perform a logical function and their logical functions, may be used to execute a wide variety of computational algorithms.
- bit line sections in a section execute the same computational algorithm on their respective data because the bit line sections share the same read and write control signals.
- FIG. 1 illustrates an example of a semiconductor memory that may include a plurality of computation memory cells and RSP circuitry
- FIG. 2 illustrates an example of a computer system that may include a plurality of computation memory cells and RSP circuitry
- FIG. 3A illustrates an example of a processing array with computational memory cells that may be incorporated into a semiconductor memory or computer system
- FIG. 3B illustrates the processing array with computational memory cells having one section and multiple bit line sections
- FIG. 3C illustrates the processing array with computational memory cells having multiple sections and multiple bit line sections
- FIGS. 4A and 4B illustrate examples of two different types of computational memory cells that may be used in the semiconductor memory of FIG. 1 , the computer system of FIG. 2 or the processing array of FIGS. 3A-3C ;
- FIG. 5 illustrates read/write logic including read logic, read data storage, and write logic associated with each bit line section in the processing array device depicted in FIG. 3C ;
- FIG. 6 illustrates the read/write logic that includes a results processing (RSP) data line and RSP circuitry in each bit line section that is used to produce combined computational results across all bit line sections on the RSP data line;
- RSS results processing
- FIG. 7 illustrates a single bit line section with the RSP circuitry including RSP logic that produces the combined computational results on the RSP line and an additional line to the write multiplexer that provides the ability to store the logic state of the RSP data line;
- FIG. 8 illustrates an example of the signal timing associated with the RSP logic shown in FIG. 7 ;
- FIG. 9 illustrates high level RSP functionality when “K” sections with “n” bit lines in the processing array including the generation of the RSP result from each section;
- FIG. 10 illustrates a processing array with 2048 bit lines and four RSP data lines
- FIG. 11 illustrates a single bit line section from the processing array in FIG. 10 ;
- FIG. 12 illustrates the detailed RSP logic implemented in FIG. 11 ;
- FIG. 13 illustrates an example of the signal timing for generating an rsp2K result and an resp32K result and transmitting those results out of the processing array
- FIG. 14 illustrates an example of the signal timing for generating an rsp16 result and transmitting that result to the write multiplexer in each bit line section;
- FIGS. 15A and 15B illustrate an example of the signal timing for generating an rsp256 result, transmitting the result back to rsp16 and transmitting that result to the write multiplexer in each bit line section;
- FIGS. 16A and 16B illustrate an example of the signal timing for generating an rsp2K result, transmitting the result back to rsp256 and then to rsp16 and transmitting that result to the write multiplexer in each bit line section;
- FIGS. 17A and 17B illustrate an example of the signal timing for generating an rsp32K result, transmitting the result back to rsp2K, to rsp256 and then to rsp16 and transmitting that result to the write multiplexer in each bit line section;
- FIGS. 18A and 18B illustrate the high level RSP functionality for sixteen sections with 2K bit lines.
- the disclosure is particularly applicable to a processing array, semiconductor memory or computer that utilizes a plurality of computational memory cells (with each cell being formed with a static random access memory (SRAM) cell) and additional response (RSP) circuitry to provide a mechanism to logically combine the computation results across multiple bit line sections in a section and across multiple sections, and transmit the combined result as an output of the processing array and/or store the combined result into one or more of those multiple bit line sections.
- SRAM static random access memory
- RSP additional response
- the disclosure is directed to a memory/processing array that has a plurality of computing memory cells in an array with additional RSP circuitry.
- Each computing memory cell in a column in the array may have a read bit line and the read bit line for each of the computing memory cells in the column may be tied together as a single read bit line.
- the memory/processing array may be subdivided into one or more sections (an example of which is shown in FIGS. 3B and 3C ) wherein each section has a unique set of “n” bit lines (each bit line being part of a bit line section) where each bit line section (bl-sect) comprises a single read bit line and a pair of positive and negative write bit lines, with each bit line connected to “m” computational memory cells.
- Each bit line section also may have a read data storage that is used to capture and store the read result from the read bit line during read operations (so a read data storage is implemented per read bit line) and read circuitry for routing the read data or the selected write data for performing logical operations.
- BL-Sect[x,y] is a shorthand notation indicating a bit line section with bit line “y” in section “x” and “bl-sect” means bit line section.
- FIG. 1 illustrates an example of a semiconductor memory 10 that may include a plurality of computation memory cells and circuitry that provides an RSP capability that is described below in more detail.
- the below disclosed plurality of computation memory cells and RSP circuitry allow the semiconductor memory 10 to logically combine computation results across bit line sections of the plurality of computation memory cells.
- FIG. 2 illustrates an example of a computer system 20 that may include a plurality of computation memory cells and the RSP circuitry that are described below in more detail.
- the below disclosed plurality of computation memory cells and RSP circuitry allow the semiconductor memory 10 or computer system 20 and memory 24 to logically combine computation results across bit line sections of the plurality of computation memory cells.
- the computer system 20 in FIG. 2 may have at least one processor 22 and a memory 24 that may include the plurality of computation memory cells and read circuitry for selecting read or write data.
- FIG. 3A illustrates an example of a processing array 30 with computational memory cells in an array that may be incorporated into a semiconductor memory or computer system and may include RSP circuitry.
- the processing array 30 may include an array of computational memory cells (cell 00 , . . . , cell 0 n and cell m 0 , . . . , cell mn).
- the array of computational memory cells may be rectangular as shown in FIG. 3A and may have a plurality of columns and a plurality of rows wherein the computational memory cells in a particular column may also be connected to the same read bit line (RBL 0 , . . . , RBLn).
- the processing array 30 may further include a wordline (WL) generator and read/write logic control circuit 32 that may be connected to and generate signals for the read word line (RE) and write word line (WE) for each memory cell (such as RE 0 , . . . , REn and WE 0 , . . . , WEn) to control the read and write operations is well known and one or more read/write circuitry 34 that are connected to the read and write bit lines of the computational memory cells.
- the processing array may have read/write circuitry 34 for each set of bit line signals of the computational memory cells (e.g., for each column of the computational memory cells whose read bit lines are connected to each other).
- BL 0 read/write logic 340 may be coupled to the read and write bit lines (WBLb 0 , WBL 0 and RBL 0 ) for the computational memory cells in column 0 of the array and BLn read/write logic 34 n may be coupled to the read and write bit lines (WBLbn, WBLn and RBLn) for the computational memory cells in column n of the array as shown in FIG. 3A .
- the wordline (WL) generator and read/write logic control circuit 32 may also generate one or more control signals that control each read/write circuitry 34 .
- the one or more control signals may include a Read_Done control signal, an XORacc_En control signal, an ANDacc_En control signal and an ORacc_En control signal whose operation and details are described in the above incorporated by reference application.
- control signals are used so that the wordline (WL) generator and read/write logic control circuit 32 may generate different control signals for each embodiment or the wordline (WL) generator and read/write logic control circuit 32 may generate each of the control signals, but then only certain of the control signals or all of the control signals may be utilized as described in the above incorporated by reference co-pending patent application.
- the wordline (WL) generator and read/write logic control circuit 32 may activate one or more word lines that activate one or more computational memory cells so that the read bit lines of those one or more computational memory cells may be read out. Further details of the read operation are not provided here since the read operation is well known.
- FIGS. 3B and 3C illustrate the processing array 30 with computational memory cells having sections having the same elements as shown in FIG. 3A .
- the array 30 in FIG. 3B has one section (Section 0 ) with “n” bit lines (bit line 0 (BL 0 ), . . . , bit line n (BLn)) in different bit line sections (bl-sect), where each bit line connects to “m” computational memory cells (cell 00 , . . . , cell m 0 for bit line 0 , for example).
- the m cells may be the plurality of computational memory cells that are part of each column of the array 30 .
- FIG. 3C illustrates the processing array 30 with computational memory cells having multiple sections.
- the processing array device 30 comprises “k” sections with “n” bit lines each, where each bit line within each section connects to “m” computational memory cells. Note that the other elements of the processing array 30 are present in FIG. 3C , but not shown for clarity.
- the BL-Sect(0,0) block shown corresponds to the BL-Sect(0,0) shown in FIG. 3B with the plurality of computational memory cells and the read/write logic 340 and each other block shown in FIG. 3C corresponds to a separate portion of the processing array.
- the set of control signals, generated by the wordline generator and read/write logic controller 32 , for each section may include one or more read enable control signals (for example S[ 0 ]_RE[m: 0 ] for section 0 ), one or more write enable control signals (for example S[ 0 ]_WE[m: 0 ] for section 0 ) and one or more read/write control signals (for example S[ 0 ]_RW_Ctrl[p: 0 ] for section 0 ).
- the array 30 may have a plurality of sections ( 0 , . . . , kin the example in FIG. 3C ) and each section may have multiple bit line sections ( 0 , n per section, in the example in FIG. 3C ).
- FIGS. 4A and 4B illustrate examples of two different types of computational memory cells that may be used in the semiconductor memory of FIG. 1 , the computer system of FIG. 2 or the processing array of FIGS. 3A-C .
- the computational memory cell are based on an SRAM memory cell.
- FIG. 4A illustrates an example of a dual port SRAM cell 20 that may be used for computation.
- the dual port SRAM cell may include two cross coupled inverters I 21 , I 22 and two access transistors M 23 and M 24 that interconnected together to form a 6T SRAM cell.
- the SRAM may be operated as storage latch and may have a write port.
- the two inverters are cross coupled since the input of the first inverter is connected to the output of the second inverter and the output of the first inverter is coupled to the input of the second inverter as shown in FIG. 4A .
- a Write Word line carries a signal and is called WE and a write bit line and its complement are called WBL and WBLb, respectively.
- the Write word line WE is coupled to the gates of the two access transistors M 23 , M 24 that are part of the SRAM cell.
- the write bit line and its complement (WBL and WBLb) are each coupled to one side of the respective access transistors M 23 , M 24 as shown in FIG. 4A while the other side of each of those access transistors M 23 , M 24 are coupled to each side of the cross coupled inverters (labeled D and Db in FIG. 4A .)
- the circuit in FIG. 4A may also have a read word line RE, a read bit line RBL and a read port formed by transistors M 21 , M 22 coupled together to form as isolation circuit as shown.
- the read word line RE may be coupled to the gate of transistor M 21 that forms part of the read port while the read bit line is coupled to the source terminal of transistor M 21 .
- the gate of transistor M 22 may be coupled to the Db output from the cross coupled inverters I 21 , I 22 .
- multiple cells can turn on to perform an AND function.
- RBL is pre-charged high and if the Db signal of all cells that are turned on by RE is “0”, then RBL stays high since, although the gate of transistor M 21 is turned on by the RE signal, the gate of M 22 is not turned on and the RBL line is not connected to the ground to which the drain of transistor M 22 is connected. If the Db signal of any or all of the cells is “1” then RBL is discharged to 0 since the gate of M 22 is turned on and the RBL line is connected to ground.
- RBL NOR (Db0, Db1, etc.) where Db0, Db1, etc. are the complementary data of the SRAM cells that have been turned on by the RE signal.
- the Db signal of the cell 20 may be coupled to a gate of transistor M 22 to drive the RBL.
- the Db signal is isolated from the RBL line and its signal/voltage level by the transistors M 21 , M 22 . Because the Db signal/value is isolated from the RBL line and signal/voltage level, the Db signal is not susceptive to the lower bit line level caused by multiple “0” data stored in multiple cells in contrast to the typical SRAM cell. Therefore, for the cell in FIG. 4A , there is no limitation of how many cells can be turned on to drive RBL.
- the cell offers more operands for the AND function since there is no limit of how many cells can be turned on to drive RBL.
- the RBL line is pre-charged (not a static pull up transistor as with the typical 6T cell) so this cell can provide much faster sensing because the current generated by the cell is all be used to discharge the bit line capacitance with no current being consumed by a static pull up transistor so that the bit line discharging rate can be faster by more than 2 times.
- the sensing for the disclosed cell is also lower power without the extra current consumed by a static pull up transistor and the discharging current is reduced by more than half.
- the write port of the cell in FIG. 4A is operated in the same manner as the 6T typical SRAM cell.
- the write cycle and Selective Write cycle for the cell have the same limitation as the typical 6T cell.
- FIG. 4B illustrates an implementation of a dual port SRAM cell 100 with an XOR function.
- the dual port SRAM cell 100 may include two cross coupled inverters I 31 , I 32 and two access transistors M 33 and M 34 that are interconnected together as shown in FIG. 4B to form the basic SRAM cell.
- the SRAM may be operated as storage latch and may have a write port.
- the two inverters I 31 , I 32 are cross coupled since the input of the first inverter is connected to the output of the second inverter (labeled D) and the output of the first inverter (labeled Db) is coupled to the input of the second inverter as shown in FIG. 4B .
- the cross coupled inverters I 31 , I 32 form the latch of the SRAM cell.
- the access transistor M 33 and M 34 may have their respective gates connected to write bit line and its complement (WBL, WBLb) respectively.
- a Write Word line carries a signal WE.
- the Write word line WE is coupled to the gate of a transistor M 35 that is part of the access circuitry for the SRAM cell.
- the circuit in FIG. 4B may also have a read word line RE, a read bit line RBL and a read port formed by transistors M 31 , M 32 coupled together to form as isolation circuit as shown.
- the read word line RE may be coupled to the gate of transistor M 31 that forms part of the read port while the read bit line RBL is coupled to the drain terminal of transistor M 31 .
- the gate of transistor M 32 may be coupled to the Db output from the cross coupled inverters I 31 , I 32 .
- the isolation circuit isolates the latch output Db (in the example in FIG. 4B ) from the read bit line and signal/voltage level so that the Db signal is not susceptive to the lower bit line level caused by multiple “0” data stored in multiple cells in contrast to the typical SRAM cell.
- the cell 100 may further include two more read word line transistors M 36 , M 37 and one extra complementary read word line, REb.
- REb When the read port is active, either RE or REb is high and the REb signal/voltage level is the complement of RE signal/voltage level.
- RBL is pre-charged high, and if one of (M 31 , M 32 ) or (M 36 , M 37 ) series transistors is on, RBL is discharged to 0. If none of (M 31 , M 32 ) or (M 36 , M 37 ) series transistors is on, then RBL stay high as 1 since it was precharged high.
- D is the data stored in the cell and Db is the complement data stored in the cell, describes the functioning/operation of the cell:
- the word size is 8, then it needs to be stored in 8 cells (with one cell being shown in FIG. 4B ) on the same bit line.
- the following equation for the multiple bits on the bit line may describe the operation of the cells as:
- RBL AND(XNOR( RE 1, D 1),XNOR( RE 2, D 2), . . . ,XNOR( REi,Di )), where i is the number of active cell. (EQ2)
- the circuit 100 may also be used to do logic operations mixing true and complement data as shown below:
- RBL AND( D 1, D 2, . . . Dn,Dbn+ 1, Dbn+ 2, . . . Dbm ) (EQ3)
- D1, D2, Dn are “n” number of data with RE on and Dbn+1, Dbn+2, Dbm are m-n number of data with REb on.
- RBL XOR( RE,D ) (EQ4)
- RBL NOR( D 1, D 2, . . . , Dn,Dbn+ 1, Dbn+ 2, . . . Dbm ) (EQ5)
- D1, D2, . . . Dn are n number of data with RE on and Dbn+1, Dbn+2, . . . Dbm are m-n number of data with REb on.
- the read port of the circuit 100 is FIG. 4B may be reconfigured differently to achieve different Boolean equation.
- transistors M 31 , M 32 , M 36 and M 37 may be changed to PMOS and the source of M 32 and M 37 is VDD instead of VSS, the bit line is pre-charged to 0 instead of 1 and the word line RE active state is 0.
- the logic equations EQ1 is inverted so that RBL is an XOR function of RE and D (EQ6).
- EQ3 is rewritten as an OR function (EQ7) as follows:
- RBL XOR( RE,D ) (EQ6)
- RBL OR( D 1, D 2, . . . , Dn,Dbn+ 1, Dbn+ 2, . . . Dbm ) (EQ7)
- D1, D2, . . . Dn are n number of data with RE on and Dbn+1, Dbn+2, Dbm are m-n number of data with REb on.
- RBL XNOR( RE,D ) (EQ8)
- RBL NAND( D 1, D 2, . . . , Dn,Dbn+ 1, Dbn+ 2, . . . Dbm ) (EQ9)
- D1, D2, . . . Dn are n number of data with RE on and Dbn+1, Dbn+2, . . . Dbm are m-n number of data with REb on.
- EQ1 can be written as below:
- RBL XNOR(Key, D ) EQ10
- RBL AND(XNOR(Key[0], D [0]),XNOR(Key[1], D [1], . . . ,Key[7], D [7]) EQ11
- FIG. 5 illustrates more details of the read/write circuitry 34 including read logic, read data storage, and write logic for each bl-sect in the processing array device depicted in FIG. 3C .
- the read/write circuitry 34 for each bit line section may include read circuitry 50 , a read storage 52 , implemented as a register, and write circuitry 54 .
- the read/write circuitry 34 may also implement one embodiment of the RSP circuitry as described below.
- the read circuitry 50 and read storage 52 allows the data on the read bit lines connected to the particular read circuitry and read storage to accumulate so that more complex Boolean logic operations may be performed.
- Various implementations of the read circuitry 50 and read storage 52 may be found in Ser. No. 16/111,178 filed Aug.
- the write circuitry 54 manages the writing of data from each section.
- Each of the read circuitry 50 , read storage 52 and write circuitry 54 may be connected to one or more control signals (S[x]_RW_Ctrl[p: 0 ] in the example implementation shown in FIG. 5 ) that control the operation of each of the circuits.
- the control signals may include the read control signals that are described above in the incorporated by reference patent application.
- the read circuitry 50 may receive inputs from the read bit line of the computing memory cells of the section (S[x]_RBL[y]) and the write circuitry 54 may receive an input from the read data storage 52 and output data to the write bit lines of the computing memory cells of the section (S[x]_WBL[y] and S[x]_WBLb[y] in the example in FIG. 5 ).
- the algorithm e.g. a search algorithm
- the RSP data line remains in its “0” default state only if the computation result captured in all bl-sect Read Registers is “0”; otherwise, the RSP data line is pulled to “1”, thereby producing a logical OR of all bl-sect Read Register states on the RSP data line.
- this scheme only requires one pull-down transistor on the RSP data line. However, several pull-down transistors (controlled by the same control signal) may be implemented on the RSP data line if needed (e.g. to decrease the amount of time required to discharge the RSP data line to “0”).
- RSP circuitry is implemented in each bl-sect such that the logical OR result produced on the RSP data line can be stored in a computational memory cell in the bl-sect.
- RSP circuitry is implemented in the section such that the logical OR result produced on the RSP data line is driven to circuitry outside the processing array.
- FIG. 6 illustrates the read/write logic 34 that includes a results processing (RSP) data line and RSP circuitry in each bit line section that is used to produce combined computational results across all bit line sections on the RSP data line.
- the read/write logic 34 has the same read logic 50 and read register 52 as the circuitry in FIG. 5 that operates in the same manner. As shown in FIG.
- the write logic 54 may further have a write multiplexer 62 , such as a 6:1 multiplexer) that selects, based on the RW_Ctrl[p: 0 ] control signal, the data to be written into the memory cells for this bit line section wherein the data is selected from the read register 52 output (S[x]_RBL[y]_Reg_Out signal for example) as well as the read register output data from the neighboring bit line sections (S[x ⁇ 1]_RBL[y]_Reg_Out, S[x+1]_RBL[y]_Reg_Out, S[x]_RBL[y ⁇ 1]_Reg_Out and S[x]_RBL[y+1]_Reg_Out).
- a write multiplexer 62 such as a 6:1 multiplexer
- the read/write circuitry 34 may further include RSP Logic 64 that is used to produce the combined computation result on the RSP data line using the “n” bl-sect Read Register outputs as the data sources.
- the RSP data line data may be input to the write MUX 62 as shown that provides the ability to store the logic state of the RSP data line in a bl-sect memory cell.
- the RSP logic is controlled by the RSPsel and RSPend control signals (generated by the read/write logic control 32 in FIG. 3A ) and has an output that drives the RSP data line.
- FIG. 7 illustrates a single bit line section with the RSP circuitry 70 including RSP logic that produces the combined computational results on the RSP line and an additional line to the write multiplexer that provides the ability to store the logic state of the RSP data line.
- the circuitry may include a pull down enable signal (S[x]_RSP_PDen for example) control signal that defaults to “1”, and is only “0” from the rising edge of the RSPsel control signal to the falling edge of the RSPend control signal.
- S[x]_RSP_PDen pull down enable signal
- the circuitry may further include a latch 72 (RSP latch) whose data input is the “RBL_Reg_Out” output (S[x]_RBL[y]_Reg_out for example) of the Read Register 52 of the bit line section, whose clock input is the RSPsel control signal, whose reset input is the RSP_PDen control signal, and whose data output is an RSP Pull-Up Enable (“RSP_PUen”) control signal.
- the RSP circuitry 70 may further include an RSP Pull-Down Transistor 74 (“RSP PD”) implemented on the RSP data line, whose enable input is the RSP_PDen control signal. In the depicted implementation, there is one RSP PD per bl-sect.
- the circuitry 70 may further include an RSP Pull-Up Transistor 76 (“RSP PU”) implemented on the RSP data line, whose enable input is the RSP_PUen control signal. In the depicted implementation, there is one RSP PU per bl-sect.
- the circuitry 70 may further include a first Buffer 78 (“BUF 1 ”) whose input is the RSP data line, and whose output “RSP in” is driven to the bl-sect Write Mux shown in FIG. 6 . In the depicted implementation, there is one such buffer per bl-sect.
- the circuitry 70 may further include a second Buffer 79 (“BUF 2 ”) whose input is the RSP data line, and whose output “RSP_out” is driven out of the processing array. In the depicted implementation, there is one such buffer for the entire section.
- FIG. 8 illustrates an example of the signal timing associated with the RSP logic shown in FIG. 7 when RSP is engaged to store the RSP result in the bl-sect and to drive the RSP result out of the processing array.
- the timing diagram illustrates the minimum delay required between the assertion of RSPsel and the assertions of RSPend and GWE. If needed, this delay can be increased without affecting the depicted RSP functionality.
- the RSP implementation in FIG. 7 allows for an RSP read operation (i.e. a read operation that generates a computation result in the bl-sect Read Register that is subsequently used for RSP engagement) to be initiated as often as every 3 cycles.
- the RSP implementation in FIG. 7 allows for a non-RSP read operation (i.e. a read operation that generates a computation result in the bl-sect Read Register that is not subsequently used for RSP engagement) to be initiated in any cycle(s) after an RSP read operation is initiated, without affecting the RSP result associated with that RSP read operation.
- FIG. 9 illustrates high level RSP functionality for a processing array having “K” sections (Section 0 , Section 1 , . . . , Section K) with each section having “n” bit line sections (BL-Sect[ 0 , 0 ], . . . , BL-Sect[ 0 ,n], etc.) with “n” bit lines in the processing array.
- each section of the processing array has RSP circuitry and logic in each bit line section and a buffer that generate the RSP result (S[ 0 ]_RSP_out, etc.) from each section.
- a mechanism to logically combine the computation results across multiple bit line sections (bl-sects) in a section and across multiple sections where each bl-sect produces a computation result that is ultimately captured in its Read Register is provided.
- the circuitry for the second embodiment may include:
- the implementation is such that the outbound RSP flow can be halted at any RSP stage before the inbound RSP flow (see below) is engaged to return the OR result from the RSP stage at which the outbound RSP flow was halted back to all stage 1 RSP data lines, and from there store it in the bl-sects. In this way any RSP stage result, from stage 1 to stage “s”, can be stored in the bl-sects.
- the purpose of the inbound RSP functionality is to return the OR result from the RSP stage at which the outbound RSP flow is halted back to the stage 1 RSP data lines, and from there store it in the bl-sects.
- the inbound RSP flow is only engaged when the objective of an RSP engagement includes storing the RSP result in the bl-sects. If the sole objective of an RSP engagement is to use the RSP result outside the processing array, then the outbound RSP flow is engaged to stage “s” and the inbound RSP flow is not engaged.
- additional RSP circuitry is implemented in each bl-sect such that the logical OR result produced on the stage 1 RSP data line can be stored in a computational memory cell in the bl-sect. Additional RSP circuitry is also implemented in each section such that the logical OR result produced on each stage “s-1” RSP data line (spanning an entire section) is driven to circuitry outside the processing array. Additional RSP circuitry is also implemented such that the logical OR result produced on the single stage “s” RSP data line (spanning multiple sections) is driven to circuitry outside the processing array.
- FIG. 10 illustrates a processing array 30 with 2048 bit lines (and cells 0 , 0 , . . . , cell m, 2047 ), 2K bit line sections (BL-Sect[ 0 , 0 ], . . . , BL-Sect[ 0 , 2047 ]) and four RSP data lines including rsp16 (stage 1), rsp256 (stage 2), rsp2K (stage 3), and rsp32K (stage 4).
- FIG. 10 shows that each bit line section has its read/write logic with RSP circuitry 341 , 342 connected to one of the RSP data lines and each section has RSP logic 343 connected to all of the RSP lines.
- All of the read/write logic and RSP circuits are also connected to the control signals (RW_Ctrl[p: 0 ] and RSP Ctrl) generated by the controller 32 of the processing array 30 .
- the RSP circuitry in each bl-sect ( 341 , 342 in the example in FIG. 10 ) is used to produce a combined computation result across groups of 16 bl-sects on each of the 128 rsp16 data lines in the section.
- the RSP circuitry 343 in each section is used to: 1) produce a combined computation result across groups of 256 bl-sects on each of the 8 rsp256 data lines in the section; 2) produce a combined computation result across all 2K bl-sects on the rsp2K data line in the section; and 3) produce a combined computation result across 32K bl-sects in 16 sections on the rsp32K data line associated with those 16 sections.
- FIG. 11 illustrates a single bit line section (Section[x], BL[y] in this example) from the processing array 30 in FIG. 10 that has the same read logic 50 , read register 52 and write multiplexer 54 as described above that operates in the same way as described above.
- FIG. 11 also shows RSP circuitry outside the bl-sect.
- the group of sixteen sections in the computation memory has RSP Control Logic circuitry 1100 .
- blocks 1100 and 1104 ⁇ 1114 shown in FIG. 1 are implemented outside of the bl-sect.
- each rspXXX data line there are multiple circuits per rspXX data line. For example, there is one 1104 block and one 1110 block per rsp16 data line, as noted immediately to the right of those blocks; since each rsp16 data line spans 16 bl-sects which means there are one of these particular blocks per 16 bl-sects.
- the RSP control logic 1100 has inputs that are rsp16sel, rsp256sel, rsp2Ksel, rsp32Ksel, rspStartRet, and rspEnd control signals that are generated by the control logic 32 of the processing array 30 described above, and whose outputs are rsp16_LATen, rsp16_PDen, rsp256_LATen, rsp256_PDen, rsp2K_LATen, rsp2K_PDen, rsp32K_LATen, rsp32K_PDen, and rspReturn control signals that control the RSP output circuits and logic and RSP input circuits and logic for the bit line section as shown in FIG. 11 .
- the logic illustrasted in FIG. 11 may further comprise:
- FIG. 12 illustrates the detailed RSP logic implemented in FIG. 11 .
- the circuits include:
- FIGS. 13-17 illustrate the RSP signal timing associated with the circuit in FIG. 12 .
- FIGS. 13-17 illustrate the RSP signal timing associated with the circuit in FIG. 12 .
- FIG. 13 illustrates an example of the signal timing for generating an rsp2K result and an resp32K result and transmitting those results out of the processing array.
- the RSP implementation in FIG. 9 allows for “RSP” read operations to be initiated every 3 cycles, when RSP is engaged to drive RSP results out of the processing array.
- FIG. 14 illustrates an example of the signal timing for generating an rsp16 result and transmitting that result to the write multiplexer in each bit line section when RSP is engaged to produce and store rsp16 results in the bl_sects.
- the RSP implementation in FIG. 11 allows for “RSP” read operations to be initiated every 3 cycles, when RSP is engaged to produce and store rsp16 results in the bl-sects.
- FIGS. 15A and 15B illustrate an example of the signal timing for generating an rsp256 result, transmitting the result back to rsp16 and transmitting that result to the write multiplexer in each bit line section when RSP is engaged to produce and store rsp256 results in the bl_sects.
- the RSP implementation in FIG. 9 allows for “RSP” read operations to be initiated every 6 cycles, when RSP is engaged to produce and store rsp256 results in the bl-sects.
- FIGS. 16A and 16B illustrate an example of the signal timing for generating an rsp2K result, transmitting the result back to rsp256 and then to rsp16 and transmitting that result to the write multiplexer in each bit line section when RSP is engaged to produce and store rsp2K results in the bl_sects.
- rsp256sel is asserted to “1” for a half-cycle to engage stage 2 outbound RSP functionality.
- the RSP implementation in FIG. 11 allows for “RSP” read operations to be initiated every 8 cycles, when RSP is engaged to produce and store rsp2K results in the bl-sects.
- FIGS. 17A and 17B illustrate an example of the signal timing for generating an rsp32K result, transmitting the result back to rsp2K, to rsp256 and then to rsp16 and transmitting that result to the write multiplexer in each bit line section.
- the RSP implementation in FIG. 11 allows for “RSP” read operations to be initiated every 10 cycles, when RSP is engaged to produce and store rsp32K results in the bl-sects.
- FIGS. 18A and 18B illustrate the high level RSP functionality for sixteen sections with 2K bit lines when 16 sections with 2K bit lines each are implemented in the processing array device according to FIGS. 10-12 . As shown, within a section there are:
- system and method disclosed herein may be implemented via one or more components, systems, servers, appliances, other subcomponents, or distributed between such elements.
- systems may include an/or involve, inter alia, components such as software modules, general-purpose CPU, RAM, etc. found in general-purpose computers.
- components such as software modules, general-purpose CPU, RAM, etc. found in general-purpose computers.
- a server may include or involve components such as CPU, RAM, etc., such as those found in general-purpose computers.
- system and method herein may be achieved via implementations with disparate or entirely different software, hardware and/or firmware components, beyond that set forth above.
- components e.g., software, processing components, etc.
- computer-readable media associated with or embodying the present inventions
- aspects of the innovations herein may be implemented consistent with numerous general purpose or special purpose computing systems or configurations.
- exemplary computing systems, environments, and/or configurations may include, but are not limited to: software or other components within or embodied on personal computers, servers or server computing devices such as routing/connectivity components, hand-held or laptop devices, multiprocessor systems, microprocessor-based systems, set top boxes, consumer electronic devices, network PCs, other existing computer platforms, distributed computing environments that include one or more of the above systems or devices, etc.
- aspects of the system and method may be achieved via or performed by logic and/or logic instructions including program modules, executed in association with such components or circuitry, for example.
- program modules may include routines, programs, objects, components, data structures, etc. that performs particular tasks or implement particular instructions herein.
- the inventions may also be practiced in the context of distributed software, computer, or circuit settings where circuitry is connected via communication buses, circuitry or links. In distributed settings, control/instructions may occur from both local and remote computer storage media including memory storage devices.
- Computer readable media can be any available media that is resident on, associable with, or can be accessed by such circuits and/or computing components.
- Computer readable media may comprise computer storage media and communication media.
- Computer storage media includes volatile and nonvolatile, removable and non-removable media implemented in any method or technology for storage of information such as computer readable instructions, data structures, program modules or other data.
- Computer storage media includes, but is not limited to, RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, digital versatile disks (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other medium which can be used to store the desired information and can accessed by computing component.
- Communication media may comprise computer readable instructions, data structures, program modules and/or other components. Further, communication media may include wired media such as a wired network or direct-wired connection, however no media of any such type herein includes transitory media. Combinations of the any of the above are also included within the scope of computer readable media.
- the terms component, module, device, etc. may refer to any type of logical or functional software elements, circuits, blocks and/or processes that may be implemented in a variety of ways.
- the functions of various circuits and/or blocks can be combined with one another into any other number of modules.
- Each module may even be implemented as a software program stored on a tangible memory (e.g., random access memory, read only memory, CD-ROM memory, hard disk drive, etc.) to be read by a central processing unit to implement the functions of the innovations herein.
- the modules can comprise programming instructions transmitted to a general purpose computer or to processing/graphics hardware via a transmission carrier wave.
- the modules can be implemented as hardware logic circuitry implementing the functions encompassed by the innovations herein.
- the modules can be implemented using special purpose instructions (SIMD instructions), field programmable logic arrays or any mix thereof which provides the desired level performance and cost.
- SIMD instructions special purpose instructions
- features consistent with the disclosure may be implemented via computer-hardware, software and/or firmware.
- the systems and methods disclosed herein may be embodied in various forms including, for example, a data processor, such as a computer that also includes a database, digital electronic circuitry, firmware, software, or in combinations of them.
- a data processor such as a computer that also includes a database
- digital electronic circuitry such as a computer
- firmware such as a firmware
- software such as a computer
- the systems and methods disclosed herein may be implemented with any combination of hardware, software and/or firmware.
- the above-noted features and other aspects and principles of the innovations herein may be implemented in various environments.
- Such environments and related applications may be specially constructed for performing the various routines, processes and/or operations according to the invention or they may include a general-purpose computer or computing platform selectively activated or reconfigured by code to provide the necessary functionality.
- the processes disclosed herein are not inherently related to any particular computer, network, architecture, environment, or other apparatus, and may be implemented by a suitable combination of hardware, software, and/or firmware.
- various general-purpose machines may be used with programs written in accordance with teachings of the invention, or it may be more convenient to construct a specialized apparatus or system to perform the required methods and techniques.
- aspects of the method and system described herein, such as the logic may also be implemented as functionality programmed into any of a variety of circuitry, including programmable logic devices (“PLDs”), such as field programmable gate arrays (“FPGAs”), programmable array logic (“PAL”) devices, electrically programmable logic and memory devices and standard cell-based devices, as well as application specific integrated circuits.
- PLDs programmable logic devices
- FPGAs field programmable gate arrays
- PAL programmable array logic
- Some other possibilities for implementing aspects include: memory devices, microcontrollers with memory (such as EEPROM), embedded microprocessors, firmware, software, etc.
- aspects may be embodied in microprocessors having software-based circuit emulation, discrete logic (sequential and combinatorial), custom devices, fuzzy (neural) logic, quantum devices, and hybrids of any of the above device types.
- the underlying device technologies may be provided in a variety of component types, e.g., metal-oxide semiconductor field-effect transistor (“MOSFET”) technologies like complementary metal-oxide semiconductor (“CMOS”), bipolar technologies like emitter-coupled logic (“ECL”), polymer technologies (e.g., silicon-conjugated polymer and metal-conjugated polymer-metal structures), mixed analog and digital, and so on.
- MOSFET metal-oxide semiconductor field-effect transistor
- CMOS complementary metal-oxide semiconductor
- ECL emitter-coupled logic
- polymer technologies e.g., silicon-conjugated polymer and metal-conjugated polymer-metal structures
- mixed analog and digital and so on.
- the words “comprise,” “comprising,” and the like are to be construed in an inclusive sense as opposed to an exclusive or exhaustive sense; that is to say, in a sense of “including, but not limited to.” Words using the singular or plural number also include the plural or singular number respectively. Additionally, the words “herein,” “hereunder,” “above,” “below,” and words of similar import refer to this application as a whole and not to any particular portions of this application. When the word “or” is used in reference to a list of two or more items, that word covers all of the following interpretations of the word: any of the items in the list, all of the items in the list and any combination of the items in the list.
Abstract
A read and write data processing apparatus and method associated with computational memory cells formed as a memory/processing array (having a plurality of bit line sections) provides a mechanism to logically combine the computation results across multiple bit line sections in a section and across multiple sections, and transmit the combined result as an output of the processing array and/or store the combined result into one or more of those multiple bit line sections.
Description
- This application is a divisional of and claims priority under 35 USC 120 and 121 to U.S. patent application Ser. No. 16/152,374 filed on Oct. 4, 2018 and entitled “Results Processing Circuits and Methods Associated with Computational Memory Cells” that is in turn a continuation in part of and claims priority under 35 USC 120 to U.S. patent application Ser. No. 15/709,399, filed Sep. 19, 2017 and entitled “Computational Dual Port Sram Cell And Processing Array Device Using The Dual Port Sram Cells For Xor And Xnor Computations”, U.S. patent application Ser. No. 15/709,401, filed Sep. 19, 2017 and entitled “Computational Dual Port Sram Cell And Processing Array Device Using The Dual Port Sram Cells For Xor And Xnor Computations”, U.S. patent application Ser. No. 15/709,379, filed Sep. 19, 2017 and entitled “Computational Dual Port Sram Cell And Processing Array Device Using The Dual Port Sram Cells”, U.S. patent application Ser. No. 15/709,382, filed Sep. 19, 2017 and entitled “Computational Dual Port Sram Cell And Processing Array Device Using The Dual Port Sram Cells”, and U.S. patent application Ser. No. 15/709,385, filed Sep. 19, 2017 and entitled “Computational Dual Port Sram Cell And Processing Array Device Using The Dual Port Sram Cells” that in turn claim priority under 35 USC 119(e) and 120 and claim the benefit of U.S. Provisional Patent Application No. 62/430,767, filed Dec. 6, 2016 and entitled “Computational Dual Port Sram Cell And Processing Array Device Using The Dual Port Sram Cells For Xor And Xnor Computations” and U.S. Provisional Patent Application No. 62/430,762, filed Dec. 6, 2016 and entitled “Computational Dual Port Sram Cell And Processing Array Device Using The Dual Port Sram Cells”, the entirety of all of which are incorporated herein by reference.
- The disclosure relates generally to a computational memory element and in particular to a computational memory element array having results processing circuitry.
- Memory cells have traditionally been used to store bits of data. It is also possible to architect a memory cell so that the memory cell is able to perform some simple logical functions when multiple memory cells are connected to the same read bit line. For example, when memory cells A, B, and C are connected to a particular read bit line and are read simultaneously, and the memory cells and read bit line circuitry are designed to produce a logical AND result, then the result that appears on the read bit line is AND (a,b,c) (i.e. “a AND b AND c”), where a, b, and c represent the binary data values stored in memory cells A, B, and C respectively. More particularly, in these computational memory cells, the read bit line is pre-charged to a logic “1” before each read operation, and the activation of one or more read enable signals to one or more memory cells discharges the read bit line to a logic “0” if the data stored in any one or more of those memory cells=“0”; otherwise, the read bit line remains a logic “1” (i.e. in its pre-charge state). In this way, the read bit line result is the logical AND of the data stored in those memory cells. The memory cells may be subdivided into a plurality of sections and each section may have a plurality of bit line sections that perform a logical function and their logical functions, may be used to execute a wide variety of computational algorithms.
- Typically, all bit line sections in a section execute the same computational algorithm on their respective data because the bit line sections share the same read and write control signals. Oftentimes, it is desirable to be able to combine the results across multiple bit line sections to generate a single result for those multiple bit line sections. For example, if the algorithm is designed to search for a particular data pattern within the data stored in the bit line sections, it is desirable to know if any of the one or more of the bit line sections contain that data pattern. It is to this end that this disclosure is directed.
-
FIG. 1 illustrates an example of a semiconductor memory that may include a plurality of computation memory cells and RSP circuitry; -
FIG. 2 illustrates an example of a computer system that may include a plurality of computation memory cells and RSP circuitry; -
FIG. 3A illustrates an example of a processing array with computational memory cells that may be incorporated into a semiconductor memory or computer system; -
FIG. 3B illustrates the processing array with computational memory cells having one section and multiple bit line sections; -
FIG. 3C illustrates the processing array with computational memory cells having multiple sections and multiple bit line sections; -
FIGS. 4A and 4B illustrate examples of two different types of computational memory cells that may be used in the semiconductor memory ofFIG. 1 , the computer system ofFIG. 2 or the processing array ofFIGS. 3A-3C ; -
FIG. 5 illustrates read/write logic including read logic, read data storage, and write logic associated with each bit line section in the processing array device depicted inFIG. 3C ; -
FIG. 6 illustrates the read/write logic that includes a results processing (RSP) data line and RSP circuitry in each bit line section that is used to produce combined computational results across all bit line sections on the RSP data line; -
FIG. 7 illustrates a single bit line section with the RSP circuitry including RSP logic that produces the combined computational results on the RSP line and an additional line to the write multiplexer that provides the ability to store the logic state of the RSP data line; -
FIG. 8 illustrates an example of the signal timing associated with the RSP logic shown inFIG. 7 ; -
FIG. 9 illustrates high level RSP functionality when “K” sections with “n” bit lines in the processing array including the generation of the RSP result from each section; -
FIG. 10 illustrates a processing array with 2048 bit lines and four RSP data lines; -
FIG. 11 illustrates a single bit line section from the processing array inFIG. 10 ; -
FIG. 12 illustrates the detailed RSP logic implemented inFIG. 11 ; -
FIG. 13 illustrates an example of the signal timing for generating an rsp2K result and an resp32K result and transmitting those results out of the processing array; -
FIG. 14 illustrates an example of the signal timing for generating an rsp16 result and transmitting that result to the write multiplexer in each bit line section; -
FIGS. 15A and 15B illustrate an example of the signal timing for generating an rsp256 result, transmitting the result back to rsp16 and transmitting that result to the write multiplexer in each bit line section; -
FIGS. 16A and 16B illustrate an example of the signal timing for generating an rsp2K result, transmitting the result back to rsp256 and then to rsp16 and transmitting that result to the write multiplexer in each bit line section; -
FIGS. 17A and 17B illustrate an example of the signal timing for generating an rsp32K result, transmitting the result back to rsp2K, to rsp256 and then to rsp16 and transmitting that result to the write multiplexer in each bit line section; and -
FIGS. 18A and 18B illustrate the high level RSP functionality for sixteen sections with 2K bit lines. - The disclosure is particularly applicable to a processing array, semiconductor memory or computer that utilizes a plurality of computational memory cells (with each cell being formed with a static random access memory (SRAM) cell) and additional response (RSP) circuitry to provide a mechanism to logically combine the computation results across multiple bit line sections in a section and across multiple sections, and transmit the combined result as an output of the processing array and/or store the combined result into one or more of those multiple bit line sections. It will be appreciated, however, that each computational memory cell may be other types of volatile and non-volatile memory cell that are within the scope of the disclosure, that other additional circuitry (including more, less or different logic) may be used and are within the scope of the disclosure or that different computational memory cell architectures that those disclosed below are within the scope of the disclosure.
- The disclosure is directed to a memory/processing array that has a plurality of computing memory cells in an array with additional RSP circuitry. Each computing memory cell in a column in the array may have a read bit line and the read bit line for each of the computing memory cells in the column may be tied together as a single read bit line. The memory/processing array may be subdivided into one or more sections (an example of which is shown in
FIGS. 3B and 3C ) wherein each section has a unique set of “n” bit lines (each bit line being part of a bit line section) where each bit line section (bl-sect) comprises a single read bit line and a pair of positive and negative write bit lines, with each bit line connected to “m” computational memory cells. Each bit line section also may have a read data storage that is used to capture and store the read result from the read bit line during read operations (so a read data storage is implemented per read bit line) and read circuitry for routing the read data or the selected write data for performing logical operations. In the disclosure, BL-Sect[x,y] is a shorthand notation indicating a bit line section with bit line “y” in section “x” and “bl-sect” means bit line section. -
FIG. 1 illustrates an example of asemiconductor memory 10 that may include a plurality of computation memory cells and circuitry that provides an RSP capability that is described below in more detail. The below disclosed plurality of computation memory cells and RSP circuitry allow thesemiconductor memory 10 to logically combine computation results across bit line sections of the plurality of computation memory cells.FIG. 2 illustrates an example of acomputer system 20 that may include a plurality of computation memory cells and the RSP circuitry that are described below in more detail. The below disclosed plurality of computation memory cells and RSP circuitry allow thesemiconductor memory 10 orcomputer system 20 andmemory 24 to logically combine computation results across bit line sections of the plurality of computation memory cells. Thecomputer system 20 inFIG. 2 may have at least oneprocessor 22 and amemory 24 that may include the plurality of computation memory cells and read circuitry for selecting read or write data. -
FIG. 3A illustrates an example of aprocessing array 30 with computational memory cells in an array that may be incorporated into a semiconductor memory or computer system and may include RSP circuitry. Theprocessing array 30 may include an array of computational memory cells (cell 00, . . . , cell 0 n and cell m0, . . . , cell mn). In one embodiment, the array of computational memory cells may be rectangular as shown inFIG. 3A and may have a plurality of columns and a plurality of rows wherein the computational memory cells in a particular column may also be connected to the same read bit line (RBL0, . . . , RBLn). Theprocessing array 30 may further include a wordline (WL) generator and read/writelogic control circuit 32 that may be connected to and generate signals for the read word line (RE) and write word line (WE) for each memory cell (such as RE0, . . . , REn and WE0, . . . , WEn) to control the read and write operations is well known and one or more read/write circuitry 34 that are connected to the read and write bit lines of the computational memory cells. In the embodiment shown inFIG. 3A , the processing array may have read/write circuitry 34 for each set of bit line signals of the computational memory cells (e.g., for each column of the computational memory cells whose read bit lines are connected to each other). For example, BL0 read/write logic 340 may be coupled to the read and write bit lines (WBLb0, WBL0 and RBL0) for the computational memory cells incolumn 0 of the array and BLn read/write logic 34 n may be coupled to the read and write bit lines (WBLbn, WBLn and RBLn) for the computational memory cells in column n of the array as shown inFIG. 3A . - The wordline (WL) generator and read/write
logic control circuit 32 may also generate one or more control signals that control each read/write circuitry 34. For example, for the different embodiments of the read/write logic described in the co-pending U.S. patent application Ser. No. 16/111,178 filed Aug. 23, 2018 and entitled “Read Data Processing Circuits and Methods Associated with Computational Memory Cells” and incorporated herein by reference, the one or more control signals may include a Read_Done control signal, an XORacc_En control signal, an ANDacc_En control signal and an ORacc_En control signal whose operation and details are described in the above incorporated by reference application. Note that for each different embodiment, a different one or more of the control signals is used so that the wordline (WL) generator and read/writelogic control circuit 32 may generate different control signals for each embodiment or the wordline (WL) generator and read/writelogic control circuit 32 may generate each of the control signals, but then only certain of the control signals or all of the control signals may be utilized as described in the above incorporated by reference co-pending patent application. - During a read operation, the wordline (WL) generator and read/write
logic control circuit 32 may activate one or more word lines that activate one or more computational memory cells so that the read bit lines of those one or more computational memory cells may be read out. Further details of the read operation are not provided here since the read operation is well known. -
FIGS. 3B and 3C illustrate theprocessing array 30 with computational memory cells having sections having the same elements as shown inFIG. 3A . Thearray 30 inFIG. 3B has one section (Section 0) with “n” bit lines (bit line 0 (BL0), . . . , bit line n (BLn)) in different bit line sections (bl-sect), where each bit line connects to “m” computational memory cells (cell 00, . . . , cell m0 forbit line 0, for example). In the example inFIG. 3B , the m cells may be the plurality of computational memory cells that are part of each column of thearray 30.FIG. 3C illustrates theprocessing array 30 with computational memory cells having multiple sections. In the example inFIG. 3C , theprocessing array device 30 comprises “k” sections with “n” bit lines each, where each bit line within each section connects to “m” computational memory cells. Note that the other elements of theprocessing array 30 are present inFIG. 3C , but not shown for clarity. InFIG. 3C , the BL-Sect(0,0) block shown corresponds to the BL-Sect(0,0) shown inFIG. 3B with the plurality of computational memory cells and the read/write logic 340 and each other block shown inFIG. 3C corresponds to a separate portion of the processing array. - As shown in
FIG. 3C , the set of control signals, generated by the wordline generator and read/write logic controller 32, for each section may include one or more read enable control signals (for example S[0]_RE[m:0] for section 0), one or more write enable control signals (for example S[0]_WE[m:0] for section 0) and one or more read/write control signals (for example S[0]_RW_Ctrl[p:0] for section 0). As shown inFIG. 3C , thearray 30 may have a plurality of sections (0, . . . , kin the example inFIG. 3C ) and each section may have multiple bit line sections (0, n per section, in the example inFIG. 3C ). -
FIGS. 4A and 4B illustrate examples of two different types of computational memory cells that may be used in the semiconductor memory ofFIG. 1 , the computer system ofFIG. 2 or the processing array ofFIGS. 3A-C . In the examples, the computational memory cell are based on an SRAM memory cell. -
FIG. 4A illustrates an example of a dualport SRAM cell 20 that may be used for computation. The dual port SRAM cell may include two cross coupled inverters I21, I22 and two access transistors M23 and M24 that interconnected together to form a 6T SRAM cell. The SRAM may be operated as storage latch and may have a write port. The two inverters are cross coupled since the input of the first inverter is connected to the output of the second inverter and the output of the first inverter is coupled to the input of the second inverter as shown inFIG. 4A . A Write Word line carries a signal and is called WE and a write bit line and its complement are called WBL and WBLb, respectively. The Write word line WE is coupled to the gates of the two access transistors M23, M24 that are part of the SRAM cell. The write bit line and its complement (WBL and WBLb) are each coupled to one side of the respective access transistors M23, M24 as shown inFIG. 4A while the other side of each of those access transistors M23, M24 are coupled to each side of the cross coupled inverters (labeled D and Db inFIG. 4A .) - The circuit in
FIG. 4A may also have a read word line RE, a read bit line RBL and a read port formed by transistors M21, M22 coupled together to form as isolation circuit as shown. The read word line RE may be coupled to the gate of transistor M21 that forms part of the read port while the read bit line is coupled to the source terminal of transistor M21. The gate of transistor M22 may be coupled to the Db output from the cross coupled inverters I21, I22. - During reading, multiple cells (with only a single cell being shown in
FIG. 4A ) can turn on to perform an AND function. Specifically, at the beginning of the read cycle, RBL is pre-charged high and if the Db signal of all cells that are turned on by RE is “0”, then RBL stays high since, although the gate of transistor M21 is turned on by the RE signal, the gate of M22 is not turned on and the RBL line is not connected to the ground to which the drain of transistor M22 is connected. If the Db signal of any or all of the cells is “1” then RBL is discharged to 0 since the gate of M22 is turned on and the RBL line is connected to ground. As a result, RBL=NOR (Db0, Db1, etc.) where Db0, Db1, etc. are the complementary data of the SRAM cells that have been turned on by the RE signal. Alternatively, RBL=NOR (Db0, Db1, etc.)=AND (D0, D1, etc.), where D0, D1, etc. are the true data of the cells that have been turned on by the RE signal. - As shown in
FIG. 4A , the Db signal of thecell 20 may be coupled to a gate of transistor M22 to drive the RBL. However, unlike the typical 6T cell, the Db signal is isolated from the RBL line and its signal/voltage level by the transistors M21, M22. Because the Db signal/value is isolated from the RBL line and signal/voltage level, the Db signal is not susceptive to the lower bit line level caused by multiple “0” data stored in multiple cells in contrast to the typical SRAM cell. Therefore, for the cell inFIG. 4A , there is no limitation of how many cells can be turned on to drive RBL. As a result, the cell (and the device made up for multiple cells) offers more operands for the AND function since there is no limit of how many cells can be turned on to drive RBL. Furthermore, in the cell inFIG. 4A , the RBL line is pre-charged (not a static pull up transistor as with the typical 6T cell) so this cell can provide much faster sensing because the current generated by the cell is all be used to discharge the bit line capacitance with no current being consumed by a static pull up transistor so that the bit line discharging rate can be faster by more than 2 times. The sensing for the disclosed cell is also lower power without the extra current consumed by a static pull up transistor and the discharging current is reduced by more than half. - The write port of the cell in
FIG. 4A is operated in the same manner as the 6T typical SRAM cell. As a result, the write cycle and Selective Write cycle for the cell have the same limitation as the typical 6T cell. In addition to the AND function described above, theSRAM cell 20 inFIG. 4A also may perform a NOR function by storing inverted data. Specifically, if D is stored at the gate of M22, instead of Db, then RBL=NOR (D0, D1, etc.). One skilled in the art understand that the cell configuration shown inFIG. 4A would be slightly altered to achieve this, but that modification is within the scope of the disclosure. Further details of this exemplary computational memory cell is found in co-pending U.S. patent application Ser. Nos. 15/709,379, 15/709,382 and 15/709,385 all filed on Sep. 19, 2017 and entitled “Computational Dual Port Sram Cell And Processing Array Device Using The Dual Port Sram Cells” which are incorporated herein by reference. -
FIG. 4B illustrates an implementation of a dual port SRAM cell 100 with an XOR function. The dual port SRAM cell 100 may include two cross coupled inverters I31, I32 and two access transistors M33 and M34 that are interconnected together as shown inFIG. 4B to form the basic SRAM cell. The SRAM may be operated as storage latch and may have a write port. The two inverters I31, I32 are cross coupled since the input of the first inverter is connected to the output of the second inverter (labeled D) and the output of the first inverter (labeled Db) is coupled to the input of the second inverter as shown inFIG. 4B . The cross coupled inverters I31, I32 form the latch of the SRAM cell. The access transistor M33 and M34 may have their respective gates connected to write bit line and its complement (WBL, WBLb) respectively. A Write Word line carries a signal WE. The Write word line WE is coupled to the gate of a transistor M35 that is part of the access circuitry for the SRAM cell. - The circuit in
FIG. 4B may also have a read word line RE, a read bit line RBL and a read port formed by transistors M31, M32 coupled together to form as isolation circuit as shown. - The read word line RE may be coupled to the gate of transistor M31 that forms part of the read port while the read bit line RBL is coupled to the drain terminal of transistor M31. The gate of transistor M32 may be coupled to the Db output from the cross coupled inverters I31, I32. The isolation circuit isolates the latch output Db (in the example in
FIG. 4B ) from the read bit line and signal/voltage level so that the Db signal is not susceptive to the lower bit line level caused by multiple “0” data stored in multiple cells in contrast to the typical SRAM cell. - The cell 100 may further include two more read word line transistors M36, M37 and one extra complementary read word line, REb. When the read port is active, either RE or REb is high and the REb signal/voltage level is the complement of RE signal/voltage level. RBL is pre-charged high, and if one of (M31, M32) or (M36, M37) series transistors is on, RBL is discharged to 0. If none of (M31, M32) or (M36, M37) series transistors is on, then RBL stay high as 1 since it was precharged high. The following equation below, where D is the data stored in the cell and Db is the complement data stored in the cell, describes the functioning/operation of the cell:
-
RBL=AND(NAND(RE,Db),NAND(REb,D))=XNOR(RE,D) (EQ1) - If the word size is 8, then it needs to be stored in 8 cells (with one cell being shown in
FIG. 4B ) on the same bit line. On a search operation, an 8 bit search key can be entered using the RE, REb lines of eight cells to compare the search key with cell data. If the search key bit is 1, then the corresponding RE=1 and REb=0 for that cell. If the search key bit is 0, then the corresponding RE=0 and REb=1. If all 8 bits match the search key, then RBL will be equal to 1. IF any 1 of the 8 bits is not matched, then RBL will be discharged and be 0. Therefore, this cell 100 (when used with 7 other cells for an 8 bit search key) can perform the same XNOR function but uses half the number of cell as the typical SRAM cell. The following equation for the multiple bits on the bit line may describe the operation of the cells as: -
RBL=AND(XNOR(RE1,D1),XNOR(RE2,D2), . . . ,XNOR(REi,Di)), where i is the number of active cell. (EQ2) - By controlling either RE or REb to be a high signal/on, the circuit 100 may also be used to do logic operations mixing true and complement data as shown below:
-
RBL=AND(D1,D2, . . . Dn,Dbn+1,Dbn+2, . . . Dbm) (EQ3) - where D1, D2, Dn are “n” number of data with RE on and Dbn+1, Dbn+2, Dbm are m-n number of data with REb on.
- Furthermore, if the cell 100 stores inverse data, meaning WBL and WBLb shown in
FIG. 4B is swapped, then the logic equation EQ1 becomes XOR function and logic equation EQ3 becomes NOR a function and can be expressed asEQ 4 and EQ5 -
RBL=XOR(RE,D) (EQ4) -
RBL=NOR(D1,D2, . . . ,Dn,Dbn+1,Dbn+2, . . . Dbm) (EQ5) - where D1, D2, . . . Dn are n number of data with RE on and Dbn+1, Dbn+2, . . . Dbm are m-n number of data with REb on.
- In another embodiment, the read port of the circuit 100 is
FIG. 4B may be reconfigured differently to achieve different Boolean equation. Specifically, transistors M31, M32, M36 and M37 may be changed to PMOS and the source of M32 and M37 is VDD instead of VSS, the bit line is pre-charged to 0 instead of 1 and the word line RE active state is 0. In this embodiment, the logic equations EQ1 is inverted so that RBL is an XOR function of RE and D (EQ6). EQ3 is rewritten as an OR function (EQ7) as follows: -
RBL=XOR(RE,D) (EQ6) -
RBL=OR(D1,D2, . . . ,Dn,Dbn+1,Dbn+2, . . . Dbm) (EQ7) - where D1, D2, . . . Dn are n number of data with RE on and Dbn+1, Dbn+2, Dbm are m-n number of data with REb on.
- If the cell stores the inverse data of the above discussed PMOS read port, meaning WBL and WBLb is swapped, then
-
RBL=XNOR(RE,D) (EQ8) -
RBL=NAND(D1,D2, . . . ,Dn,Dbn+1,Dbn+2, . . . Dbm) (EQ9) - where D1, D2, . . . Dn are n number of data with RE on and Dbn+1, Dbn+2, . . . Dbm are m-n number of data with REb on.
- For example, consider a search operation where a digital word needs to be found in a memory array in which the memory array can be configured as each bit of the word stored on the same bit line. To compare 1 bit of the word, then the data is stored in a cell and its RE is the search key, then EQ1 can be written as below:
-
RBL=XNOR(Key,D) EQ10 - If Key=D, then RBL=1. If the word size is 8 bits as D[0:7], then the search key Key[0:7] is its RE, then EQ2 can be expressed as search result and be written as below:
-
RBL=AND(XNOR(Key[0],D[0]),XNOR(Key[1],D[1], . . . ,Key[7],D[7]) EQ11 - If all Key[i] is equal to D[i] where i=0-7, then the search result RBL is match. Any one of Key[i] is not equal to D[i], then the search result is not match. Parallel search can be performed in 1 operation by arranging multiple data words along the same word line and on parallel bit lines with each word on 1 bit line. Further details of this computation memory cell may be found in U.S. patent application Ser. Nos. 15/709,399 and 15/709,401 both filed on Sep. 19, 2017 and entitled “Computational Dual Port Sram Cell And Processing Array Device Using The Dual Port Sram Cells For Xor And Xnor Computations”, which are incorporated herein by reference.
-
FIG. 5 illustrates more details of the read/write circuitry 34 including read logic, read data storage, and write logic for each bl-sect in the processing array device depicted inFIG. 3C . The read/write circuitry 34 for each bit line section may include readcircuitry 50, aread storage 52, implemented as a register, and writecircuitry 54. The read/write circuitry 34 may also implement one embodiment of the RSP circuitry as described below. Theread circuitry 50 and readstorage 52 allows the data on the read bit lines connected to the particular read circuitry and read storage to accumulate so that more complex Boolean logic operations may be performed. Various implementations of the readcircuitry 50 and readstorage 52 may be found in Ser. No. 16/111,178 filed Aug. 23, 2018 and entitled “Read Data Processing Circuits and Methods Associated with Computational Memory Cells” that is co-pending and co-owned and is incorporated herein by reference. Thewrite circuitry 54 manages the writing of data from each section. Each of the readcircuitry 50, readstorage 52 and writecircuitry 54 may be connected to one or more control signals (S[x]_RW_Ctrl[p:0] in the example implementation shown inFIG. 5 ) that control the operation of each of the circuits. The control signals may include the read control signals that are described above in the incorporated by reference patent application. - The
read circuitry 50 may receive inputs from the read bit line of the computing memory cells of the section (S[x]_RBL[y]) and thewrite circuitry 54 may receive an input from the readdata storage 52 and output data to the write bit lines of the computing memory cells of the section (S[x]_WBL[y] and S[x]_WBLb[y] in the example inFIG. 5 ). - First Embodiment of RSP Circuitry
- One way to provide a mechanism to logically combine the computation results across multiple bl-sects in a section, where each bl-sect produces a computation result that is ultimately captured in its
Read Register 52, is to: -
- Implement an RSP data line (S[x]_RSP for example) that spans all bl-sects in the section.
- Utilize additional circuitry in the bl-sects to produce a pre-defined logical function—in particular, a logical OR—on the RSP data line of the computation results captured in all bl-sect Read Registers in the section. That is, when the section has “n” bl-sects, to produce on the RSP data line the function:
-
RBL[0]_Reg_Out OR RBL[1]_Reg_Out OR . . . RBL[n−1]_Reg_Out - A logical OR is chosen because a common use case of RSP functionality is to determine if the algorithm (e.g. a search algorithm) produced a result=“1” (e.g. indicating a positive match on the search data) in any one or more of the bl-sects in the section. In the embodiments below, each of the RSP logic circuits, control logic may be implemented using known circuits including Boolean logic circuits.
- One way to produce the logical OR on the RSP data line is to define its default state as “0”, and enable (by default) a pull-down transistor on the RSP data line to pull it to its “0” default state. Then when RSP functionality is engaged, temporarily and unconditionally disable the pull-down transistor on the RSP data line, and temporarily enable a pull-up transistor—one per bl-sect—on the RSP data line if the computation result captured in its associated bl-sect Read Register=“1”. In that way, the RSP data line remains in its “0” default state only if the computation result captured in all bl-sect Read Registers is “0”; otherwise, the RSP data line is pulled to “1”, thereby producing a logical OR of all bl-sect Read Register states on the RSP data line.
- Functionally, this scheme only requires one pull-down transistor on the RSP data line. However, several pull-down transistors (controlled by the same control signal) may be implemented on the RSP data line if needed (e.g. to decrease the amount of time required to discharge the RSP data line to “0”).
- Furthermore, additional RSP circuitry is implemented in each bl-sect such that the logical OR result produced on the RSP data line can be stored in a computational memory cell in the bl-sect.
- And furthermore, additional RSP circuitry is implemented in the section such that the logical OR result produced on the RSP data line is driven to circuitry outside the processing array.
-
FIG. 6 illustrates the read/write logic 34 that includes a results processing (RSP) data line and RSP circuitry in each bit line section that is used to produce combined computational results across all bit line sections on the RSP data line. The read/write logic 34 has thesame read logic 50 and readregister 52 as the circuitry inFIG. 5 that operates in the same manner. As shown inFIG. 6 , thewrite logic 54 may further have awrite multiplexer 62, such as a 6:1 multiplexer) that selects, based on the RW_Ctrl[p:0] control signal, the data to be written into the memory cells for this bit line section wherein the data is selected from the readregister 52 output (S[x]_RBL[y]_Reg_Out signal for example) as well as the read register output data from the neighboring bit line sections (S[x−1]_RBL[y]_Reg_Out, S[x+1]_RBL[y]_Reg_Out, S[x]_RBL[y−1]_Reg_Out and S[x]_RBL[y+1]_Reg_Out). - For the RSP circuitry, the read/
write circuitry 34 may further includeRSP Logic 64 that is used to produce the combined computation result on the RSP data line using the “n” bl-sect Read Register outputs as the data sources. In addition, the RSP data line data may be input to thewrite MUX 62 as shown that provides the ability to store the logic state of the RSP data line in a bl-sect memory cell. The RSP logic is controlled by the RSPsel and RSPend control signals (generated by the read/write logic control 32 inFIG. 3A ) and has an output that drives the RSP data line. -
FIG. 7 illustrates a single bit line section with theRSP circuitry 70 including RSP logic that produces the combined computational results on the RSP line and an additional line to the write multiplexer that provides the ability to store the logic state of the RSP data line. The circuitry may include a pull down enable signal (S[x]_RSP_PDen for example) control signal that defaults to “1”, and is only “0” from the rising edge of the RSPsel control signal to the falling edge of the RSPend control signal. The circuitry may further include a latch 72 (RSP latch) whose data input is the “RBL_Reg_Out” output (S[x]_RBL[y]_Reg_out for example) of theRead Register 52 of the bit line section, whose clock input is the RSPsel control signal, whose reset input is the RSP_PDen control signal, and whose data output is an RSP Pull-Up Enable (“RSP_PUen”) control signal. The functionality ofRSP LAT 72 is such that when RSP_PDen=1, then RSP_PUen=0; when RSP_PDen=0 and RSPsel=1, then RSP_PUen=RBL_Reg_Out; otherwise, RSP_PUen remains unchanged. In the depicted implementation, there is one RSP LAT per bit line section. - The
RSP circuitry 70 may further include an RSP Pull-Down Transistor 74 (“RSP PD”) implemented on the RSP data line, whose enable input is the RSP_PDen control signal. In the depicted implementation, there is one RSP PD per bl-sect. Thecircuitry 70 may further include an RSP Pull-Up Transistor 76 (“RSP PU”) implemented on the RSP data line, whose enable input is the RSP_PUen control signal. In the depicted implementation, there is one RSP PU per bl-sect. Thecircuitry 70 may further include a first Buffer 78 (“BUF1”) whose input is the RSP data line, and whose output “RSP in” is driven to the bl-sect Write Mux shown inFIG. 6 . In the depicted implementation, there is one such buffer per bl-sect. Thecircuitry 70 may further include a second Buffer 79 (“BUF2”) whose input is the RSP data line, and whose output “RSP_out” is driven out of the processing array. In the depicted implementation, there is one such buffer for the entire section. -
FIG. 8 illustrates an example of the signal timing associated with the RSP logic shown inFIG. 7 when RSP is engaged to store the RSP result in the bl-sect and to drive the RSP result out of the processing array. -
- 1. In
cycle 0, GRE is asserted to “1” for a half-cycle to initiate a read operation that causes a computation result=“1” to be captured in the bl-sect Read Register. In this cycle, RSP_PDen=“1” (its default state), thereby causing RSP_PUen=“0” (its default state), thereby causing the RSP data line=“0” (its default state). - 2. In
cycle 1, RSPsel is asserted to “1” for a half-cycle to engage RSP functionality. The rising edge of RSPsel causes RSP_PDen=“0”, and RSPsel=1 causes RSP LAT to capture the RBL_Reg_Out output of the Read Register (i.e. the logic “1” from cycle 0), thereby causing RSP_PUen=“1”, thereby causing the RSP data line=“1”. - 3. In
cycle 2, RSPend is asserted to “1” for a half-cycle to disengage RSP functionality. The falling edge of RSPend causes RSP_PDen=“1”, thereby causing RSP_PUen=“0”, thereby causing the RSP data line=“0” (back to its default state). - 4. Simultaneously with the assertion of RSPend, GWE is asserted to “1” for a half-cycle to initiate a write operation that stores the state of the RSP data line in a computational memory cell in the bl-sect. Note that the assertion of GWE unnecessary if the sole objective of the RSP engagement is to use the RSP result outside the processing array.
- 1. In
- The timing diagram illustrates the minimum delay required between the assertion of RSPsel and the assertions of RSPend and GWE. If needed, this delay can be increased without affecting the depicted RSP functionality.
- The RSP implementation in
FIG. 7 allows for an RSP read operation (i.e. a read operation that generates a computation result in the bl-sect Read Register that is subsequently used for RSP engagement) to be initiated as often as every 3 cycles. The RSP implementation inFIG. 7 allows for a non-RSP read operation (i.e. a read operation that generates a computation result in the bl-sect Read Register that is not subsequently used for RSP engagement) to be initiated in any cycle(s) after an RSP read operation is initiated, without affecting the RSP result associated with that RSP read operation. -
FIG. 9 illustrates high level RSP functionality for a processing array having “K” sections (Section 0,Section 1, . . . , Section K) with each section having “n” bit line sections (BL-Sect[0,0], . . . , BL-Sect[0,n], etc.) with “n” bit lines in the processing array. As shown inFIG. 9 , each section of the processing array has RSP circuitry and logic in each bit line section and a buffer that generate the RSP result (S[0]_RSP_out, etc.) from each section. - Second Embodiment of RSP Circuitry
- In a second embodiment, a mechanism to logically combine the computation results across multiple bit line sections (bl-sects) in a section and across multiple sections where each bl-sect produces a computation result that is ultimately captured in its Read Register is provided. The circuitry for the second embodiment may include:
-
- A set of “s” RSP data lines (an example of which is shown in
FIG. 10 ) that, in stages, span an increasing number of bl-sects, where thestage 1 through stage “s-2” RSP data lines span a subset of bl-sects in a section, the stage “s-1” RSP data line spans all bl-sects in a section, and the stage “s” RSP data line spans all bl-sects across multiple sections. - The generation of outbound RSP functionality from the
stage 1 RSP data line to the stage “s” RSP data line, as follows:- Utilize additional circuitry in the bl-sects to produce a logical OR on each
stage 1 RSP data line of the computation results captured in the bl-sect Read Registers spanned by eachstage 1 RSP data line. - Utilize additional circuitry in the section to produce a logical OR on each
stage 2 RSP data line of the states of themultiple stage 1 RSP data lines spanned by eachstage 2 RSP data line. - Utilize additional circuitry in the section to produce a logical OR on each additional stage RSP data line from the
stage 3 RSP data line to the stage “s-1” RSP data line (e.g., perform this for each successive stage RSP data line). - Utilize additional circuitry in each section to produce a logical OR on the stage “s” RSP data line of the states of the multiple stage “s-1” RSP data lines spanned by the stage “s” RSP data line.
- Utilize additional circuitry in the bl-sects to produce a logical OR on each
- A set of “s” RSP data lines (an example of which is shown in
- One way to produce the logical OR on each RSP data line during the outbound flow is (as was done in the first embodiment) to define their default states as “0”, and enable (by default) pull-down transistor(s) to pull each RSP data line to its “0” default state. Then, when RSP functionality is engaged:
-
- At the first stage of the outbound RSP flow, temporarily and unconditionally disable the pull-down transistor(s) on the
stage 1 RSP data line, and temporarily enable a pull-up transistor—one per bl-sect—if the computation result captured in the bl-sect Read Register=“1”. - At the second stage of the outbound RSP flow, temporarily and unconditionally disable the pull-down transistor(s) on the
stage 2 RSP data line, and temporarily enable a pull-up transistor—one perstage 1 RSP data line—if the state of thestage 1 RSP data line=“1”. - At the last stage of the outbound RSP flow, temporarily and unconditionally disable the pull-down transistor(s) on the stage “s” RSP data line, and temporarily enable a pull-up transistor—one per stage “s-1” RSP data line—if the state of the stage “s-1” RSP data line=“1”.
- At the first stage of the outbound RSP flow, temporarily and unconditionally disable the pull-down transistor(s) on the
- The implementation is such that the outbound RSP flow can be halted at any RSP stage before the inbound RSP flow (see below) is engaged to return the OR result from the RSP stage at which the outbound RSP flow was halted back to all
stage 1 RSP data lines, and from there store it in the bl-sects. In this way any RSP stage result, fromstage 1 to stage “s”, can be stored in the bl-sects. -
- The generation of inbound RSP functionality from the stage “s” RSP data line to the stage “1” RSP data line, as follows:
- Utilize additional circuitry in each section to enable the state of the stage “s” RSP data line, if =“1”, onto all stage “s-1” RSP data lines spanned by the stage “s” RSP data line. Note that the state of the stage “s” RSP data line can only be “1” if the outbound flow is engaged all the way to stage “s”.
- Utilize additional circuitry in the section to enable the state of each stage “s-1” RSP data line, if =“1”, onto all stage “s-2” RSP data lines spanned by each stage “s-1” RSP data line. Note that the state of the stage “s-1” RSP data line can only be “1” if the outbound flow is engaged to stage “s-1” or further.
- Utilize additional circuitry in the section to enable the state of each
stage 2 RSP data line, if =“1”, onto allstage 1 RSP data lines spanned by eachstage 2 RSP data line. Note that the state of thestage 2 RSP data line can only be “1” if the outbound flow is engaged tostage 2 or further.
- The generation of inbound RSP functionality from the stage “s” RSP data line to the stage “1” RSP data line, as follows:
- The purpose of the inbound RSP functionality is to return the OR result from the RSP stage at which the outbound RSP flow is halted back to the
stage 1 RSP data lines, and from there store it in the bl-sects. The inbound RSP flow is only engaged when the objective of an RSP engagement includes storing the RSP result in the bl-sects. If the sole objective of an RSP engagement is to use the RSP result outside the processing array, then the outbound RSP flow is engaged to stage “s” and the inbound RSP flow is not engaged. - When the inbound RSP flow is engaged after the outbound RSP flow is halted at stage “r”, where 1<r<=s, one way to return the OR result from the stage “r” RSP data line to the stage “1” RSP data line is to:
-
- Enable a pull-up transistor on each stage “s-1” RSP data line if the state of the stage “s” RSP data line that spans it=“1” (which is only possible if r=s).
- Enable a pull-up transistor on each stage “s-2” RSP data line if the state of the stage “s-1” RSP data line that spans it=“1” (which is only possible if r>=s-1).
- Enable a pull-up transistor on each
stage 1 RSP data line if the state of thestage 2 RSP data line that spans it=“1” (which is only possible if r>=2).
- These pull-up transistors are enabled long enough for the stage “r” RSP data line state, if “1”, to propagate back to the
multiple stage 1 RSP data lines spanned by the stage “r” RSP data line. - If the outbound RSP flow is halted at the
stage 1 RSP data line (i.e. r=1), then the inbound RSP flow is still engaged as part of the process to store the result in the bl-sects even though, in this case, no return propagation from a latter RSP stage is involved. - Furthermore, additional RSP circuitry is implemented in each bl-sect such that the logical OR result produced on the
stage 1 RSP data line can be stored in a computational memory cell in the bl-sect. Additional RSP circuitry is also implemented in each section such that the logical OR result produced on each stage “s-1” RSP data line (spanning an entire section) is driven to circuitry outside the processing array. Additional RSP circuitry is also implemented such that the logical OR result produced on the single stage “s” RSP data line (spanning multiple sections) is driven to circuitry outside the processing array. -
FIG. 10 illustrates aprocessing array 30 with 2048 bit lines (andcells FIG. 10 shows that each bit line section has its read/write logic withRSP circuitry RSP logic 343 connected to all of the RSP lines. All of the read/write logic and RSP circuits are also connected to the control signals (RW_Ctrl[p:0] and RSP Ctrl) generated by thecontroller 32 of theprocessing array 30. The RSP circuitry in each bl-sect (341, 342 in the example inFIG. 10 ) is used to produce a combined computation result across groups of 16 bl-sects on each of the 128 rsp16 data lines in the section. TheRSP circuitry 343 in each section is used to: 1) produce a combined computation result across groups of 256 bl-sects on each of the 8 rsp256 data lines in the section; 2) produce a combined computation result across all 2K bl-sects on the rsp2K data line in the section; and 3) produce a combined computation result across 32K bl-sects in 16 sections on the rsp32K data line associated with those 16 sections. -
FIG. 11 illustrates a single bit line section (Section[x], BL[y] in this example) from theprocessing array 30 inFIG. 10 that has thesame read logic 50, readregister 52 and writemultiplexer 54 as described above that operates in the same way as described above. In addition to the circuitry in the bit line section,FIG. 11 also shows RSP circuitry outside the bl-sect. As shown inFIG. 11 , the group of sixteen sections in the computation memory has RSPControl Logic circuitry 1100. In particular, blocks 1100 and 1104˜1114 shown inFIG. 1 are implemented outside of the bl-sect. For the RSP out logic and the RSP in logic (1104-1114), there are multiple circuits per rspXXX data line. For example, there is one 1104 block and one 1110 block per rsp16 data line, as noted immediately to the right of those blocks; since each rsp16 data line spans 16 bl-sects which means there are one of these particular blocks per 16 bl-sects. - The
RSP control logic 1100 has inputs that are rsp16sel, rsp256sel, rsp2Ksel, rsp32Ksel, rspStartRet, and rspEnd control signals that are generated by thecontrol logic 32 of theprocessing array 30 described above, and whose outputs are rsp16_LATen, rsp16_PDen, rsp256_LATen, rsp256_PDen, rsp2K_LATen, rsp2K_PDen, rsp32K_LATen, rsp32K_PDen, and rspReturn control signals that control the RSP output circuits and logic and RSP input circuits and logic for the bit line section as shown inFIG. 11 . - The logic illustrasted in
FIG. 11 may further comprise: -
-
RSP16 Out Logic 1102 whose inputs are the output of the bl-sect Read Register, rsp16sel, rsp16 LATen, and rsp16 PDen as control signals, and whose output drives the rsp16 data line (S[x]_rsp16). -
RSP256 Out Logic 1104 whose inputs are the rsp16 data line, rsp256sel, rsp256 LATen, and rsp256 PDen control signals, and whose output drives the rsp256 data line. -
RSP2K Out Logic 1106 whose inputs are the rsp256 data line, rsp2Ksel, rsp2K_LATen, and rsp2K_PDen control signals, and whose output drives the rsp2K data line. -
RSP32K Out Logic 1108 whose inputs are the rsp2K data line, rsp32Ksel, rsp32K_LATen, and rsp32K_PDen control signals, and whose output drives the rsp32K data line. - RSP2K In Logic 1114 whose inputs are the rsp32K data line and rspReturn control signal, and whose output drives the rsp2K data line.
-
RSP256 In Logic 1112 whose inputs are the rsp2K data line and rspReturn control signal, and whose output drives the rsp256 data line. -
RSP16 In Logic 1110 whose inputs are the rsp256 data line and rspReturn control signal, and whose output drives the rsp16 data line. - The 6:1
Write Mux 54 in the bl sect whose 6th input is the rsp16 data line.
-
-
FIG. 12 illustrates the detailed RSP logic implemented inFIG. 11 . The circuits include: -
- The RSP control logic 1100 (shown in
FIG. 11 ) that generates RSP control signals including:- An rspReturn control signal, that defaults to “0” and is only “1” from the rising edge of rspStartRet to the rising edge of rspEnd.
- An RSP16 Latch Enable (“rsp16 LATen”) control signal, that defaults to “0” and is only “1” from the rising edge of rsp16sel to the falling edge of rsp256sel or the falling edge of rspReturn.
- An RSP16 Pull-Down Enable (“rsp16 PDen”) control signal that defaults to “1”, and is only “0” when rsp16_LATen=1 or rspReturn=1.
- An RSP256 Latch Enable (“rsp256 LATen”) control signal, that defaults to “0” and is only “1” from the rising edge of rsp256sel to the falling edge of rsp2Ksel or the falling edge of rspReturn.
- An RSP256 Pull-Down Enable (“rsp256 PDen”) control signal, that defaults to “1” and is only “0” when rsp256 LATen=1 or rspReturn=1.
- An RSP2K Latch Enable (“rsp2K_LATen”) control signal, that defaults to “0” and is only “1” from the rising edge of rsp2Ksel to the falling edge of rsp32Ksel or the falling edge of rspReturn.
- An RSP2K Pull-Down Enable (“rsp2K_PDen”) control signal, that defaults to “1” and is only “0” when rsp2K_LATen=1 or rspReturn=1.
- An RSP32K Latch Enable (“rsp32K_LATen”) control signal, that defaults to “0” and is only “1” from the rising edge of rsp32Ksel to the falling edge of rspEnd.
- An RSP32K Pull-Down Enable (“rsp32K_PDen”) control signal, that defaults to “1” and is only “0” when rsp32K_LATen=1.
- RSP16 Out Logic (one set per bl-sect) consisting of:
- An RSP16 Latch (“rsp16_LAT”) whose data input is the “RBL_Reg_Out” output of the bl-sect's Read Register, whose clock input is the rsp16sel control signal, whose reset input is the rsp16 LATen control signal, and whose data output is an RSP16 Out Pull-Up Enable (“rsp16_PUen_out”) control signal. The functionality of rsp16 LAT is such that when rsp16 LATen=0, then rsp16_PUen_out=0; when rsp16_LATen=1 and rsp16sel=1, then rsp16_PUen_out=RBL_Reg_Out; otherwise, rsp16_PUen_out remains unchanged.
- An RSP16 Pull-Down Transistor (“rsp16 PD”) implemented on the rsp16 data line, whose enable input is the rsp16 PDen control signal.
- An RSP16 Out Pull-Up Transistor (“rsp16 out PU”) implemented on the rsp16 data line, whose enable input is the rsp16_PUen_out control signal.
- A first Buffer (“BUF1”) whose input is the rsp16 data line, and whose output “rsp16_in” is driven to the bl-sect's Write Mux.
- RSP16 In Logic (one set per rsp16 data line) consisting of:
- An RSP16 In AND gate (“rsp16 in AND”) whose output is an RSP16 In Pull-Up Enable (“rsp16_PUen_in”) control signal equal to the logical AND of the rsp256 data line and rspReturn.
- An RSP16 In Pull-Up Transistor (“rsp16 in PU”) implemented on the rsp16 data line, whose enable input is the rsp16_PUen_in control signal.
- RSP256 Out Logic (one set per rsp16 data line) consisting of:
- An RSP256 Latch (“rsp256 LAT”) whose data input is the rsp16 data line, whose clock input is the rsp256sel control signal, whose reset input is the rsp256_LATen control signal, and whose data output is an RSP256 Out Pull-Up Enable (“rsp256_PUen_out”) control signal. The functionality of rsp256 LAT is such that when rsp256_LATen=0, then rsp256_PUen_out=0; when rsp256_LATen=1 and rsp256sel=1, then rsp256_PUen_out=RBL_Reg_Out; otherwise, rsp256_PUen_out remains unchanged.
- An RSP256 Pull-Down Transistor (“rsp256 PD”) implemented on the rsp256 data line, whose enable input is the rsp256 PDen control signal.
- An RSP256 Out Pull-Up Transistor (“rsp256 out PU”) implemented on the rsp256 data line, whose enable input is the rsp256_PUen_out control signal.
- RSP256 In Logic (one set per rsp256 data line) consisting of:
- An RSP256 In AND gate (“rsp256 in AND”) whose output is an RSP256 In Pull-Up Enable (“rsp256_PUen_in”) control signal equal to the logical AND of the rsp2K data line and rspReturn.
- An RSP256 In Pull-Up Transistor (“rsp256 in PU”) implemented on the rsp256 data line, whose enable input is the rsp256_PUen_in control signal.
- RSP2K Out Logic (one set per rsp256 data line) consisting of:
- An RSP2K Latch (“rsp2K LAT”) whose data input is the rsp256 data line, whose clock input is the rsp2Ksel control signal, whose reset input is the rsp2K_LATen control signal, and whose data output is an RSP2K Out Pull-Up Enable (“rsp2K_PUen_out”) control signal. The functionality of rsp2K LAT is such that when rsp2K_LATen=0, then rsp2K_PUen_out=0; when rsp2K_LATen=1 and rsp2Ksel=1, then rsp2K_PUen_out=RBL_Reg_Out; otherwise, rsp2K_PUen remains unchanged.
- An RSP2K Pull-Down Transistor (“rsp2K PD”) implemented on the rsp2K data line, whose enable input is the rsp2K_PDen control signal.
- An RSP2K Out Pull-Up Transistor (“rsp2K out PU”) implemented on the rsp2K data line, whose enable input is the rsp2K_PUen_out control signal.
- RSP2K In Logic (one set per rsp2K data line) consisting of:
- An RSP2K In AND gate (“rsp2K in AND”) whose output is an RSP2K In Pull-Up Enable (“rsp2K_PUen_in”) control signal equal to the logical AND of the rsp32K data line and rspReturn.
- An RSP2K In Pull-Up Transistor (“rsp2K in PU”) implemented on the rsp2K data line, whose enable input is the rsp2K_PUen_in control signal.
- rsp32K Out Logic (one set per rsp2K data line) consisting of:
- An RSP32K Latch (“rsp32K LAT”) whose data input is the rsp2K data line, whose clock input is the rsp32Ksel control signal, whose reset input is the rsp32K_LATen control signal, and whose data output is an RSP32K Pull-Up Enable (“rsp32K_PUen_out”) control signal. The functionality of rsp32K LAT is such that when rsp32K_LATen=0, then rsp32K_PUen_out=0; when rsp32K_LATen=1 and rsp32Ksel=1, then rsp32K_PUen_out=RBL_Reg_Out; otherwise, rsp32K_PUen remains unchanged.
- An RSP32K Pull-Down Transistor (“rsp32K PD”) implemented on the rsp32K data line, whose enable input is the rsp32K_PDen control signal.
- An RSP32K Out Pull-Up Transistor (“rsp32K out PU”) implemented on the rsp32K data line, whose enable input is the rsp32K_PUen control signal.
- A second Buffer (“BUF2”) whose input is the rsp32K_PUen_out control signal, and whose output “rsp2K_out” is driven out of the processing array. In the depicted implementation, there is one BUF2 per section.
- A third Buffer (“BUF3”) whose input is the rsp32K data line, and whose output “rsp32K_out” is driven out of the processing array. In the depicted implementation, there is one BUF3 per 16 sections.
- The RSP control logic 1100 (shown in
-
FIGS. 13-17 illustrate the RSP signal timing associated with the circuit inFIG. 12 . In all five figures: -
- The timing diagrams illustrates the minimum delays required between the assertion of rsp16sel and the assertion of each successive RSP control signal (i.e. rsp256sel, rsp2Ksel, rsp32Ksel, rspStartRet, rspEnd), and the assertion of GWE, required for the depicted flow. If needed, these delays can be increased without affecting the depicted RSP functionality.
- The RSP implementation in
FIG. 12 allows for a non-RSP read operation to be initiated in any cycle(s) after an RSP read operation is initiated, without affecting the RSP result associated with that RSP read operation.
-
FIG. 13 illustrates an example of the signal timing for generating an rsp2K result and an resp32K result and transmitting those results out of the processing array. -
- 1. In
cycle 0, GRE is asserted to “1” for a half-cycle to initiate a read operation that causes a computation result=“1” to be captured in the bl-sect Read Register. In this cycle, rsp16_LATen=“0” (its default state) and rsp16_PDen=“1” (its default state), thereby causing rsp16_PUen_out=“0” (its default state), thereby causing the rsp16 data line=“0” (its default state). - 2. In
cycle 1, rsp16sel is asserted to “1” for a half-cycle to engagestage 1 outbound RSP functionality. The rising edge of rsp16sel causes rsp16_LATen=“1” and rsp16_PDen=“0”, and rsp16sel=1 causes rsp16 LAT to capture the RBL_Reg_Out output of the Read Register (i.e. the logic “1” from cycle 0), thereby causing rsp16 PUen_out=“1”, thereby causing the rsp16 data line=“1”. - 3. In
cycle 2, rsp256sel is asserted to “1” for a half-cycle to engagestage 2 outbound RSP functionality. The rising edge of rsp256sel causes rsp256_LATen=“1” and rsp256_PDen=“0”, and rsp256sel=1 causes rsp256 LAT to capture the state of the rsp16 data line (i.e. the logic “1” from cycle 1), thereby causing rsp256_PUen_out=“1”, thereby causing the rsp256 data line=“1”. The falling edge of rsp256sel causes rsp16 LATen=“0” and rsp16_PDen=“1”, thereby causing rsp16_PUen_out=“0”, thereby causing the rsp16 data line=“0” (back to its default state). - 4. In
cycle 3, rsp2Ksel is asserted to “1” for a half-cycle to engagestage 3 outbound RSP functionality. The rising edge of rsp2Ksel causes rsp2K_LATen=“1” and rsp2K_PDen=“0”, and rsp2Ksel=1 causes rsp2K LAT to capture the state of the rsp256 data line (i.e. the logic “1” from cycle 2), thereby causing rsp2K_PUen_out=“1”, thereby causing the rsp2K data line=“1”. The falling edge of rsp2Ksel causes rsp256_LATen=“0” and rsp256_PDen=“1”, thereby causing rsp256_PUen_out=“0”, thereby causing the rsp256 data line=“0” (back to its default state). - 5. In
cycle 4, rsp32Ksel is asserted to “1” for a half-cycle to engagestage 4 outbound RSP functionality. The rising edge of rsp32Ksel causes rsp32K_LATen=“1” and rsp32K_PDen=“0”, and rsp32Ksel=1 causes rsp32K LAT to capture the state of the rsp2K data line (i.e. the logic “1” from cycle 3), thereby causing rsp32K_PUen_out=“1”, thereby causing the rsp32K data line=“1”. The falling edge of rsp32Ksel causes rsp2K_LATen=“0” and rsp2K_PDen=“1”, thereby causing rsp2K_PUen_out=“0”, thereby causing the rsp2K data line=“0” (back to its default state). - 6. In
cycle 5, rspEnd is asserted to “1” for a half-cycle to disengage outbound RSP functionality. The falling edge of rspEnd causes rsp32K_LATen=“0” and rsp32K_PDen=“1”, thereby causing rsp32K_PUen_out=“0”, thereby causing the rsp32K data line=“0” (back to its default state).
- 1. In
- The RSP implementation in
FIG. 9 allows for “RSP” read operations to be initiated every 3 cycles, when RSP is engaged to drive RSP results out of the processing array. -
FIG. 14 illustrates an example of the signal timing for generating an rsp16 result and transmitting that result to the write multiplexer in each bit line section when RSP is engaged to produce and store rsp16 results in the bl_sects. -
- 1. In
cycle 0, GRE is asserted to “1” for a half-cycle to initiate a read operation that causes a computation result=“1” to be captured in the bl-sect Read Register. In this cycle, rsp16_LATen=“0” (its default state) and rsp16_PDen=“1” (its default state), thereby causing rsp16_PUen_out=“0” (its default state), thereby causing the rsp16 data line=“0” (its default state). - 2. In
cycle 1, rsp16sel is asserted to “1” for a half-cycle to engagestage 1 outbound RSP functionality. The rising edge of rsp16sel causes rsp16_LATen=“1” and rsp16_PDen=“0”, and rsp16sel=1 causes rsp16 LAT to capture the RBL_Reg_Out output of the Read Register (i.e. the logic “1” from cycle 0), thereby causing rsp16_PUen_out=“1”, thereby causing the rsp16 data line=“1”. - 3. In
cycle 2, rspStartRet is asserted to “1” for a half-cycle to disengage outbound RSP functionality and engage inbound RSP functionality. The rising edge of rspStartRet causes rspReturn=“1”. - 4. In cycle 2 (and one cycle before rspEnd is asserted), GWE is asserted to “1” for a half-cycle to initiate a write operation that stores the state of the rsp16 data line in a computational memory cell in the bl-sect.
- 5. In
cycle 3, rspEnd is asserted to “1” for a half-cycle to disengage inbound RSP functionality. The rising edge of rspEnd causes rspReturn=“0”, thereby causing rsp16_LATen=“0” and rsp16_PDen=“1”, thereby causing rsp16_PUen_out=“0”, thereby causing the rsp16 data line=“0” (back to its default state).
- 1. In
- The RSP implementation in
FIG. 11 allows for “RSP” read operations to be initiated every 3 cycles, when RSP is engaged to produce and store rsp16 results in the bl-sects. -
FIGS. 15A and 15B illustrate an example of the signal timing for generating an rsp256 result, transmitting the result back to rsp16 and transmitting that result to the write multiplexer in each bit line section when RSP is engaged to produce and store rsp256 results in the bl_sects. -
- 1. In
cycle 0, GRE is asserted to “1” for a half-cycle to initiate a read operation that causes a computation result=“1” to be captured in the bl-sect Read Register. In this cycle, rsp16_LATen=“0” (its default state) and rsp16_PDen=“1” (its default state), thereby causing rsp16_PUen_out=“0” (its default state), thereby causing the rsp16 data line=“0” (its default state). - 2. In
cycle 1, rsp16sel is asserted to “1” for a half-cycle to engagestage 1 outbound RSP functionality. The rising edge of rsp16sel causes rsp16_LATen=“1” and rsp16_PDen=“0”, and rsp16sel=1 causes rsp16 LAT to capture the RBL_Reg_Out output of the Read Register (i.e. the logic “1” from cycle 0), thereby causing rsp16_PUen_out=“1”, thereby causing the rsp16 data line=“1”. - 3. In
cycle 2, rsp256sel is asserted to “1” for a half-cycle to engagestage 2 outbound RSP functionality. The rising edge of rsp256sel causes rsp256_LATen=“1” and rsp256_PDen=“0”, and rsp256sel=1 causes rsp256 LAT to capture the state of the rsp16 data line (i.e. the logic “1” from cycle 1), thereby causing rsp256_PUen_out=“1”, thereby causing the rsp256 data line=“1”. The falling edge of rsp256sel causes rsp16_LATen=“0” and rsp16_PDen=“1”, thereby causing rsp16_PUen_out=“0”, thereby causing the rsp16 data line=“0” (back to its default state). - 4. In
cycle 3, rspStartRet is asserted to “1” for a half-cycle to disengage outbound RSP functionality and engage inbound RSP functionality. The rising edge of rspStartRet causes rspReturn=“1”, thereby causing rsp16 PDen=“0” and (because the rsp256 data line=“1”) rsp16_PUen_in=“1”, thereby causing the rsp16 data line=“1”. - 5. In cycle 5 (and one cycle before rspEnd is asserted), GWE is asserted to “1” for a half-cycle to initiate a write operation that stores the state of the rsp16 data line in a computational memory cell in the bl-sect.
- 6. In
cycle 6, rspEnd is asserted to “1” for a half-cycle to disengage inbound RSP functionality. The rising edge of rspEnd causes rspReturn=“0”, thereby causing rsp256_LATen=“0” and rsp256_PDen=“1”, thereby causing rsp256_PUen_out=“0”, thereby causing the rsp256 data line=“0” (back to its default state). The falling edge of rspReturn also causes rsp16_PDen=“1” and rsp16_PUen_in=“0”, thereby causing the rsp16 data line=“0” (back to its default state).
- 1. In
- The RSP implementation in
FIG. 9 allows for “RSP” read operations to be initiated every 6 cycles, when RSP is engaged to produce and store rsp256 results in the bl-sects. -
FIGS. 16A and 16B illustrate an example of the signal timing for generating an rsp2K result, transmitting the result back to rsp256 and then to rsp16 and transmitting that result to the write multiplexer in each bit line section when RSP is engaged to produce and store rsp2K results in the bl_sects. -
- 1. In
cycle 0, GRE is asserted to “1” for a half-cycle to initiate a read operation that causes a computation result=“1” to be captured in the bl-sect Read Register. In this cycle, rsp16_LATen=“0” (its default state) and rsp16_PDen=“1” (its default state), thereby causing rsp16_PUen_out=“0” (its default state), thereby causing the rsp16 data line=“0” (its default state). - 2. In
cycle 1, rsp16sel is asserted to “1” for a half-cycle to engagestage 1 outbound RSP functionality. The rising edge of rsp16sel causes rsp16_LATen=“1” and rsp16 PDen=“0”, and rsp16sel=1 causes rsp16 LAT to capture the RBL_Reg_Out output of the Read Register (i.e. the logic “1” from cycle 0), thereby causing rsp16_PUen_out=“1”, thereby causing the rsp16 data line=“1”.
- 1. In
- In
cycle 2, rsp256sel is asserted to “1” for a half-cycle to engagestage 2 outbound RSP functionality. The rising edge of rsp256sel causes rsp256_LATen=“1” and rsp256_PDen=“0”, and rsp256sel=1 causes rsp256 LAT to capture the state of the rsp16 data line (i.e. the logic “1” from cycle 1), thereby causing rsp256_PUen_out=“1”, thereby causing the rsp256 data line=“1”. The falling edge of rsp256sel causes rsp16_LATen=“0” -
- 3. and rsp16_PDen=“1”, thereby causing rsp16_PUen_out=“0”, thereby causing the rsp16 data line=“0” (back to its default state).
- 4. In
cycle 3, rsp2Ksel is asserted to “1” for a half-cycle to engagestage 3 outbound RSP functionality. The rising edge of rsp2Ksel causes rsp2K_LATen=“1” and rsp2K_PDen=“0”, and rsp2Ksel=1 causes rsp2K LAT to capture the state of the rsp256 data line (i.e. the logic “1” from cycle 2), thereby causing rsp2K_PUen_out=“1”, thereby causing the rsp2K data line=“1”. The falling edge of rsp2Ksel causes rsp256_LATen=“0” and rsp256_PDen=“1”, thereby causing rsp256_PUen_out=“0”, thereby causing the rsp256 data line=“0” (back to its default state). - 5. In
cycle 4, rspStartRet is asserted to “1” for a half-cycle to disengage outbound RSP functionality and engage inbound RSP functionality. The rising edge of rspStartRet causes rspReturn=“1”, thereby causing rsp256_PDen=“0” and (because the rsp2K data line=“1”) rsp256_PUen_in=“1”, thereby causing the rsp256 data line=“1”. The rising edge of rspReturn also causes rsp16_PDen=“0” and (because the rsp256 data line=“1”) rsp16_PUen_in=“1”, thereby causing the rsp16 data line=“1”. - 6. In cycle 7 (and one cycle before rspEnd is asserted), GWE is asserted to “1” for a half-cycle to initiate a write operation that stores the state of the rsp16 data line in a computational memory cell in the bl-sect.
- 7. In
cycle 8, rspEnd is asserted to “1” for a half-cycle to disengage inbound RSP functionality. The rising edge of rspEnd causes rspReturn=“0”, thereby causing rsp2K_LATen=“0” and rsp2K_PDen=“1”, thereby causing rsp2K_PUen_out=“0”, thereby causing the rsp2K data line=“0” (back to its default state). The falling edge of rspReturn also causes rsp256_PDen=“1” and rsp256_PUen_in=“0”, thereby causing the rsp256 data line=“0” (back to its default state). The falling edge of rspReturn also causes rsp16_PDen=“1” and rsp16_PUen_in=“0”, thereby causing the rsp16 data line=“0” (back to its default state).
- The RSP implementation in
FIG. 11 allows for “RSP” read operations to be initiated every 8 cycles, when RSP is engaged to produce and store rsp2K results in the bl-sects. -
FIGS. 17A and 17B illustrate an example of the signal timing for generating an rsp32K result, transmitting the result back to rsp2K, to rsp256 and then to rsp16 and transmitting that result to the write multiplexer in each bit line section. -
- 1. In
cycle 0, GRE is asserted to “1” for a half-cycle to initiate a read operation that causes a computation result=“1” to be captured in the bl-sect Read Register. In this cycle, rsp16_LATen=“0” (its default state) and rsp16_PDen=“1” (its default state), thereby causing rsp16_PUen_out=“0” (its default state), thereby causing the rsp16 data line=“0” (its default state). - 2. In
cycle 1, rsp16sel is asserted to “1” for a half-cycle to engagestage 1 outbound RSP functionality. The rising edge of rsp16sel causes rsp16_LATen=“1” and rsp16_PDen=“0”, and rsp16sel=1 causes rsp16 LAT to capture the RBL_Reg_Out output of the Read Register (i.e. the logic “1” from cycle 0), thereby causing rsp16_PUen_out=“1”, thereby causing the rsp16 data line=“1”. - 3. In
cycle 2, rsp256sel is asserted to “1” for a half-cycle to engagestage 2 outbound RSP functionality. The rising edge of rsp256sel causes rsp256_LATen=“1” and rsp256_PDen=“0”, and rsp256sel=1 causes rsp256 LAT to capture the state of the rsp16 data line (i.e. the logic “1” from cycle 1), thereby causing rsp256_PUen_out=“1”, thereby causing the rsp256 data line=“1”. The falling edge of rsp256sel causes rsp16_LATen=“0” and rsp16_PDen=“1”, thereby causing rsp16_PUen_out=“0”, thereby causing the rsp16 data line=“0” (back to its default state). - 4. In
cycle 3, rsp2Ksel is asserted to “1” for a half-cycle to engagestage 3 outbound RSP functionality. The rising edge of rsp2Ksel causes rsp2K_LATen=“1” and rsp2K_PDen=“0”, and rsp2Ksel=1 causes rsp2K LAT to capture the state of the rsp256 data line (i.e. the logic “1” from cycle 2), thereby causing rsp2K_PUen_out=“1”, thereby causing the rsp2K data line=“1”. The falling edge of rsp2Ksel causes rsp256_LATen=“0” and rsp256_PDen=“1”, thereby causing rsp256_PUen_out=“0”, thereby causing the rsp256 data line=“0” (back to its default state). - 5. In
cycle 4, rsp32Ksel is asserted to “1” for a half-cycle to engagestage 4 outbound RSP functionality. The rising edge of rsp32Ksel causes rsp32K_LATen=“1” and rsp32K_PDen=“0”, and rsp32Ksel=1 causes rsp32K LAT to capture the state of the rsp2K data line (i.e. the logic “1” from cycle 3), thereby causing rsp32K_PUen_out=“1”, thereby causing the rsp32K data line=“1”. The falling edge of rsp32Ksel causes rsp2K_LATen=“0” and rsp2K_PDen=“1”, thereby causing rsp2K_PUen_out=“0”, thereby causing the rsp2K data line=“0” (back to its default state). - 6. In
cycle 5, rspStartRet is asserted to “1” for a half-cycle to disengage outbound RSP functionality and engage inbound RSP functionality. The rising edge of rspStartRet causes rspReturn=“1”, thereby causing rsp2K_PDen=“0” and (because the rsp32K data line=“1”) rsp2K_PUen_in=“1”, thereby causing the rsp2K data line=“1”. The rising edge of rspReturn also causes rsp256_PDen=“0” and (because the rsp2K data line=“1”) rsp256_PUen_in=“1”, thereby causing the rsp256 data line=“1”. The rising edge of rspReturn also causes rsp16_PDen=“0” and (because the rsp256 data line=“1”) rsp16_PUen_in=“1”, thereby causing the rsp16 data line=“1”. - 7. In cycle 9 (and one cycle before rspEnd is asserted), GWE is asserted to “1” for a half-cycle to initiate a write operation that stores the state of the rsp16 data line in a computational memory cell in the bl-sect.
- 8. In
cycle 10, rspEnd is asserted to “1” for a half-cycle to disengage inbound RSP functionality. The rising edge of rspEnd causes rspReturn=“0”, thereby causing rsp2K_PDen=“1” and rsp2K_PUen_in=“0”, thereby causing the rsp2K data line=“0” (back to its default state). The falling edge of rspReturn also causes rsp256_PDen=“1” and rsp256_PUen_in=“0”, thereby causing the rsp256 data line=“0” (back to its default state). The falling edge of rspReturn also causes rsp16_PDen=“1” and rsp16_PUen_in=“0”, thereby causing the rsp16 data line=“0” (back to its default state). The falling edge of rspEnd causes rsp32K_LATen=“0” and rsp32K_PDen=“1”, thereby causing rsp32K_PUen_out=“0”, thereby causing the rsp32K data line=“0” (back to its default state).
- 1. In
- The RSP implementation in
FIG. 11 allows for “RSP” read operations to be initiated every 10 cycles, when RSP is engaged to produce and store rsp32K results in the bl-sects. -
FIGS. 18A and 18B illustrate the high level RSP functionality for sixteen sections with 2K bit lines when 16 sections with 2K bit lines each are implemented in the processing array device according toFIGS. 10-12 . As shown, within a section there are: -
- 128 rsp16 data lines—one per each group of 16 bls of 2K.
- 8 rsp256 data lines—one per each group of 16 rsp16 data lines, i.e. one per each group of 256 bls of 2K.
- 1 rsp2K data line—one per a single group of 8 rsp256 data lines, i.e. one per all 2K bls.
- And across all 16 sections there is:
-
- 1 rsp32K data line—one per a single group of 16 rsp2K data lines, i.e. one per 32K bls.
- The foregoing description, for purpose of explanation, has been described with reference to specific embodiments. However, the illustrative discussions above are not intended to be exhaustive or to limit the disclosure to the precise forms disclosed. Many modifications and variations are possible in view of the above teachings. The embodiments were chosen and described in order to best explain the principles of the disclosure and its practical applications, to thereby enable others skilled in the art to best utilize the disclosure and various embodiments with various modifications as are suited to the particular use contemplated.
- The system and method disclosed herein may be implemented via one or more components, systems, servers, appliances, other subcomponents, or distributed between such elements. When implemented as a system, such systems may include an/or involve, inter alia, components such as software modules, general-purpose CPU, RAM, etc. found in general-purpose computers. In implementations where the innovations reside on a server, such a server may include or involve components such as CPU, RAM, etc., such as those found in general-purpose computers.
- Additionally, the system and method herein may be achieved via implementations with disparate or entirely different software, hardware and/or firmware components, beyond that set forth above. With regard to such other components (e.g., software, processing components, etc.) and/or computer-readable media associated with or embodying the present inventions, for example, aspects of the innovations herein may be implemented consistent with numerous general purpose or special purpose computing systems or configurations. Various exemplary computing systems, environments, and/or configurations that may be suitable for use with the innovations herein may include, but are not limited to: software or other components within or embodied on personal computers, servers or server computing devices such as routing/connectivity components, hand-held or laptop devices, multiprocessor systems, microprocessor-based systems, set top boxes, consumer electronic devices, network PCs, other existing computer platforms, distributed computing environments that include one or more of the above systems or devices, etc.
- In some instances, aspects of the system and method may be achieved via or performed by logic and/or logic instructions including program modules, executed in association with such components or circuitry, for example. In general, program modules may include routines, programs, objects, components, data structures, etc. that performs particular tasks or implement particular instructions herein. The inventions may also be practiced in the context of distributed software, computer, or circuit settings where circuitry is connected via communication buses, circuitry or links. In distributed settings, control/instructions may occur from both local and remote computer storage media including memory storage devices.
- The software, circuitry and components herein may also include and/or utilize one or more type of computer readable media. Computer readable media can be any available media that is resident on, associable with, or can be accessed by such circuits and/or computing components. By way of example, and not limitation, computer readable media may comprise computer storage media and communication media. Computer storage media includes volatile and nonvolatile, removable and non-removable media implemented in any method or technology for storage of information such as computer readable instructions, data structures, program modules or other data. Computer storage media includes, but is not limited to, RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, digital versatile disks (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other medium which can be used to store the desired information and can accessed by computing component. Communication media may comprise computer readable instructions, data structures, program modules and/or other components. Further, communication media may include wired media such as a wired network or direct-wired connection, however no media of any such type herein includes transitory media. Combinations of the any of the above are also included within the scope of computer readable media.
- In the present description, the terms component, module, device, etc. may refer to any type of logical or functional software elements, circuits, blocks and/or processes that may be implemented in a variety of ways. For example, the functions of various circuits and/or blocks can be combined with one another into any other number of modules. Each module may even be implemented as a software program stored on a tangible memory (e.g., random access memory, read only memory, CD-ROM memory, hard disk drive, etc.) to be read by a central processing unit to implement the functions of the innovations herein. Or, the modules can comprise programming instructions transmitted to a general purpose computer or to processing/graphics hardware via a transmission carrier wave. Also, the modules can be implemented as hardware logic circuitry implementing the functions encompassed by the innovations herein. Finally, the modules can be implemented using special purpose instructions (SIMD instructions), field programmable logic arrays or any mix thereof which provides the desired level performance and cost.
- As disclosed herein, features consistent with the disclosure may be implemented via computer-hardware, software and/or firmware. For example, the systems and methods disclosed herein may be embodied in various forms including, for example, a data processor, such as a computer that also includes a database, digital electronic circuitry, firmware, software, or in combinations of them. Further, while some of the disclosed implementations describe specific hardware components, systems and methods consistent with the innovations herein may be implemented with any combination of hardware, software and/or firmware. Moreover, the above-noted features and other aspects and principles of the innovations herein may be implemented in various environments. Such environments and related applications may be specially constructed for performing the various routines, processes and/or operations according to the invention or they may include a general-purpose computer or computing platform selectively activated or reconfigured by code to provide the necessary functionality. The processes disclosed herein are not inherently related to any particular computer, network, architecture, environment, or other apparatus, and may be implemented by a suitable combination of hardware, software, and/or firmware. For example, various general-purpose machines may be used with programs written in accordance with teachings of the invention, or it may be more convenient to construct a specialized apparatus or system to perform the required methods and techniques.
- Aspects of the method and system described herein, such as the logic, may also be implemented as functionality programmed into any of a variety of circuitry, including programmable logic devices (“PLDs”), such as field programmable gate arrays (“FPGAs”), programmable array logic (“PAL”) devices, electrically programmable logic and memory devices and standard cell-based devices, as well as application specific integrated circuits. Some other possibilities for implementing aspects include: memory devices, microcontrollers with memory (such as EEPROM), embedded microprocessors, firmware, software, etc. Furthermore, aspects may be embodied in microprocessors having software-based circuit emulation, discrete logic (sequential and combinatorial), custom devices, fuzzy (neural) logic, quantum devices, and hybrids of any of the above device types. The underlying device technologies may be provided in a variety of component types, e.g., metal-oxide semiconductor field-effect transistor (“MOSFET”) technologies like complementary metal-oxide semiconductor (“CMOS”), bipolar technologies like emitter-coupled logic (“ECL”), polymer technologies (e.g., silicon-conjugated polymer and metal-conjugated polymer-metal structures), mixed analog and digital, and so on.
- It should also be noted that the various logic and/or functions disclosed herein may be enabled using any number of combinations of hardware, firmware, and/or as data and/or instructions embodied in various machine-readable or computer-readable media, in terms of their behavioral, register transfer, logic component, and/or other characteristics. Computer-readable media in which such formatted data and/or instructions may be embodied include, but are not limited to, non-volatile storage media in various forms (e.g., optical, magnetic or semiconductor storage media) though again does not include transitory media. Unless the context clearly requires otherwise, throughout the description, the words “comprise,” “comprising,” and the like are to be construed in an inclusive sense as opposed to an exclusive or exhaustive sense; that is to say, in a sense of “including, but not limited to.” Words using the singular or plural number also include the plural or singular number respectively. Additionally, the words “herein,” “hereunder,” “above,” “below,” and words of similar import refer to this application as a whole and not to any particular portions of this application. When the word “or” is used in reference to a list of two or more items, that word covers all of the following interpretations of the word: any of the items in the list, all of the items in the list and any combination of the items in the list.
- Although certain presently preferred implementations of the invention have been specifically described herein, it will be apparent to those skilled in the art to which the invention pertains that variations and modifications of the various implementations shown and described herein may be made without departing from the spirit and scope of the invention. Accordingly, it is intended that the invention be limited only to the extent required by the applicable rules of law.
- While the foregoing has been with reference to a particular embodiment of the disclosure, it will be appreciated by those skilled in the art that changes in this embodiment may be made without departing from the principles and spirit of the disclosure, the scope of which is defined by the appended claims.
Claims (37)
1. A method for response for a processing device, the method comprising:
generating a computation result using a processing array device having a plurality of memory cells arranged in an array having a plurality of columns and a plurality of rows, each memory cell having a storage element wherein the array has a plurality of sections and each section has a plurality of bit line sections and a plurality of bit lines with one bit line per bit line section, wherein the memory cells in each bit line section are all connected to a single read bit line that generates the computation result and the plurality of bit lines in each section are distinct from the plurality of bit lines included in the other sections of the array;
communicating, using a plurality of RSP data lines of the processing array device, a combined computation result; and
performing an response (RSP) read operation by engaging in stages the plurality of RSP data lines.
2. The method of claim 1 further comprising initiating a read operation without using the RSP data lines in any clock cycle after the RSP read operation is initiated without affecting the combined computation result.
3. The method of claim 1 further comprising driving an RSP result on the plurality of RSP data lines out of the processing array.
4. The method of claim 3 , wherein driving the RSP result further comprises initiating a read operation that causes a “1” computation result to be captured in a read storage of each bit line section and setting each of a first stage of the plurality of RSP data line stages to a default state.
5. The method of claim 4 , wherein, after initiating the read operation and setting each of the first stage of the plurality of RSP data line stages to a default state, driving the RSP result further comprises engaging RSP outbound circuits in each of the first stage of the plurality of RSP data line stages and causing each of the first stage of the plurality of RSP data line stages to store a value of “1”.
6. The method of claim 5 , wherein, after engaging the RSP outbound circuits in each of the first stage of the plurality of RSP data line stages, engaging the RSP outbound circuits in each of a second stage of the plurality of RSP data line stages, causing each of the second stages of the plurality of RSP data line stages to store a value of “1” and causing each of the first stage of the plurality of RSP data line stages to return back to a default state.
7. The method of claim 6 , wherein, after engaging the RSP outbound circuits in each of the second stage of the plurality of RSP data line stages, engaging the RSP outbound circuits in each of a third stage of the plurality of RSP data line stages, causing each of the third stage of the plurality of RSP data line stages to store a value of “1” and causing each of the second stage of the plurality of RSP data line stages to return back to a default state.
8. The method of claim 7 , wherein, after engaging the RSP outbound circuits in each of the third stage of the plurality of RSP data line stages, engaging the RSP outbound circuits in each of a fourth stage of the plurality of RSP data line stages, causing each of the fourth stage of the plurality of RSP data line stages to store a value of “1” and causing each of the third stage of the plurality of RSP data line stages to return back to a default state.
9. The method of claim 8 , wherein, after engaging the RSP outbound circuits in each of a fourth stage of the plurality of RSP data line stages, disengaging the outbound RSP circuits and causing each of the fourth stage of the plurality of RSP data line stages to return back to a default state.
10. The method of claim 1 further comprising engaging the RSP data lines to produce and store a result for a first stage of the plurality of RSP data line stages in each bit line section.
11. The method of claim 10 , wherein storing the result further comprises initiating a read operation that causes a “1” computation result to be captured in a read storage of each bit line section and setting each of a first stage of the plurality of RSP data line stages to a default state.
12. The method of claim 11 , wherein, after initiating the read operation and setting each of the first stage of the plurality of RSP data line stages to a default state, driving the RSP result further comprises engaging RSP outbound circuits in each of the first stage of the plurality of RSP data line stages and causing each of the first stage of the plurality of RSP data line stages to store a value of “1”.
13. The method of claim 12 , wherein, after engaging the RSP outbound circuits in each of the first stage of the plurality of RSP data line stages, disengaging the RSP outbound circuits, engaging RSP inbound circuits and initiating a write operation that stores a state of the first stage RSP data line in the computational memory cell in the bit line section.
14. The method of claim 13 , wherein, after initiating the write operation, disengaging the RSP inbound circuits causing the first stage RSP data line to return to a default value.
15. The method of claim 1 further comprising engaging the RSP data lines to produce and store a result for a second stage of the plurality of RSP data line stages in each bit line section.
16. The method of claim 15 , wherein storing the result further comprises initiating a read operation that causes a “1” computation result to be captured in a read storage of each bit line section and setting each of a first stage of the plurality of RSP data line stages to a default state.
17. The method of claim 16 , wherein, after initiating the read operation and setting each of the first stage of the plurality of RSP data line stages to a default state, driving the RSP result further comprises engaging RSP outbound circuits in each of the first stage of the plurality of RSP data line stages and causing each of the first stage of the plurality of RSP data line stages to store a value of “1”.
18. The method of claim 17 , wherein, after engaging the RSP outbound circuits in each of the first stage of the plurality of RSP data line stages, engaging the RSP outbound circuits in each of a second stage of the plurality of RSP data line stages, causing each of the second stages of the plurality of RSP data line stages to store a value of “1” and causing each of the first stage of the plurality of RSP data line stages to return back to a default state.
19. The method of claim 18 , wherein, after engaging the RSP outbound circuits in each of the second stage of the plurality of RSP data line stages, disengaging the RSP outbound circuits, engaging RSP inbound circuits and initiating a write operation that stores a state of the first stage RSP data line in the computational memory cell in the bit line section.
20. The method of claim 19 , wherein, after initiating the write operation, disengaging the RSP inbound circuits causing the first stage RSP data line to return to a default value.
21. The method of claim 1 further comprising engaging the RSP data lines to produce and store a result for a third stage of the plurality of RSP data line stages in each bit line section.
22. The method of claim 21 , wherein storing the result further comprises initiating a read operation that causes a “1” computation result to be captured in a read storage of each bit line section and setting each of a first stage of the plurality of RSP data line stages to a default state.
23. The method of claim 22 , wherein, after initiating the read operation and setting each of the first stage of the plurality of RSP data line stages to a default state, driving the RSP result further comprises engaging RSP outbound circuits in each of the first stage of the plurality of RSP data line stages and causing each of the first stage of the plurality of RSP data line stages to store a value of “1”.
24. The method of claim 23 , wherein, after engaging the RSP outbound circuits in each of the first stage of the plurality of RSP data line stages, engaging the RSP outbound circuits in each of a second stage of the plurality of RSP data line stages, causing each of the second stages of the plurality of RSP data line stages to store a value of “1” and causing each of the first stage of the plurality of RSP data line stages to return back to a default state.
25. The method of claim 24 , wherein, after engaging the RSP outbound circuits in each of the second stage of the plurality of RSP data line stages, engaging the RSP outbound circuits in each of a third stage of the plurality of RSP data line stages, causing each of the third stage of the plurality of RSP data line stages to store a value of “1” and causing each of the second stage of the plurality of RSP data line stages to return back to a default state.
26. The method of claim 25 , wherein, after engaging the RSP outbound circuits in each of the third stage of the plurality of RSP data line stages, disengaging the outbound RSP circuits and causing each of the first and second stage of the plurality of RSP data line stages to be set to “1” and engaging the RSP inbound circuits.
27. The method of claim 26 , wherein, after disengaging the RSP outbound circuits and engaging the inbound RSP circuits, initiating a write operation that stores a state of the first stage RSP data line in the computational memory cell in the bit line section.
28. The method of claim 27 , wherein, after initiating the write operation, disengaging the RSP inbound circuits causing the first stage RSP data line to return to a default value.
29. The method of claim 1 , wherein engaging the RSP data lines to produce and store a result for a fourth stage of the plurality of RSP data line stages in each bit line section.
30. The method of claim 29 , wherein engaging the RSP data lines further comprises initiating a read operation that causes a “1” computation result to be captured in a read storage of each bit line section and setting each of a first stage of the plurality of RSP data line stages to a default state.
31. The method of claim 30 , wherein, after initiating the read operation and setting each of the first stage of the plurality of RSP data line stages to a default state, driving the RSP result further comprises engaging RSP outbound circuits in each of the first stage of the plurality of RSP data line stages and causing each of the first stage of the plurality of RSP data line stages to store a value of “1”.
32. The method of claim 31 , wherein, after engaging the RSP outbound circuits in each of the first stage of the plurality of RSP data line stages, engaging the RSP outbound circuits in each of a second stage of the plurality of RSP data line stages, causing each of the second stages of the plurality of RSP data line stages to store a value of “1” and causing each of the first stage of the plurality of RSP data line stages to return back to a default state.
33. The method of claim 32 , wherein, after engaging the RSP outbound circuits in each of the second stage of the plurality of RSP data line stages, engaging the RSP outbound circuits in each of a third stage of the plurality of RSP data line stages, causing each of the third stage of the plurality of RSP data line stages to store a value of “1” and causing each of the second stage of the plurality of RSP data line stages to return back to a default state.
34. The method of claim 33 , wherein, after engaging the RSP outbound circuits in each of the third stage of the plurality of RSP data line stages, engaging the RSP outbound circuits in each of a fourth stage of the plurality of RSP data line stages, causing each of the fourth stage of the plurality of RSP data line stages to store a value of “1” and causing each of the third stage of the plurality of RSP data line stages to return back to a default state.
35. The method of claim 34 , wherein, after engaging the RSP outbound circuits in each of a fourth stage of the plurality of RSP data line stages, disengaging the outbound RSP circuits and causing each of the first, second and third stage RSP data lines to be set to “1” and engaging the inbound RSP circuits.
36. The method of claim 35 , wherein, after disengaging the RSP outbound circuits and engaging the inbound RSP circuits, initiating a write operation that stores a state of the first stage RSP data line in the computational memory cell in the bit line section.
37. The method of claim 36 , wherein, after initiating the write operation, disengaging the RSP inbound circuits causing the first, second and third stage RSP data line to return to a default value.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US17/081,869 US20210216246A1 (en) | 2016-12-06 | 2020-10-27 | Results processing circuits and methods associated with computational memory cells |
Applications Claiming Priority (9)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US201662430767P | 2016-12-06 | 2016-12-06 | |
US201662430762P | 2016-12-06 | 2016-12-06 | |
US15/709,399 US10998040B2 (en) | 2016-12-06 | 2017-09-19 | Computational memory cell and processing array device using the memory cells for XOR and XNOR computations |
US15/709,382 US10725777B2 (en) | 2016-12-06 | 2017-09-19 | Computational memory cell and processing array device using memory cells |
US15/709,401 US10249362B2 (en) | 2016-12-06 | 2017-09-19 | Computational memory cell and processing array device using the memory cells for XOR and XNOR computations |
US15/709,379 US10521229B2 (en) | 2016-12-06 | 2017-09-19 | Computational memory cell and processing array device using memory cells |
US15/709,385 US10860318B2 (en) | 2016-12-06 | 2017-09-19 | Computational memory cell and processing array device using memory cells |
US16/152,374 US10891076B1 (en) | 2016-12-06 | 2018-10-04 | Results processing circuits and methods associated with computational memory cells |
US17/081,869 US20210216246A1 (en) | 2016-12-06 | 2020-10-27 | Results processing circuits and methods associated with computational memory cells |
Related Parent Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
US16/152,374 Division US10891076B1 (en) | 2016-12-06 | 2018-10-04 | Results processing circuits and methods associated with computational memory cells |
Publications (1)
Publication Number | Publication Date |
---|---|
US20210216246A1 true US20210216246A1 (en) | 2021-07-15 |
Family
ID=70160198
Family Applications (3)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
US16/152,374 Active US10891076B1 (en) | 2016-12-06 | 2018-10-04 | Results processing circuits and methods associated with computational memory cells |
US16/713,383 Active US11194519B2 (en) | 2016-12-06 | 2019-12-13 | Results processing circuits and methods associated with computational memory cells |
US17/081,869 Abandoned US20210216246A1 (en) | 2016-12-06 | 2020-10-27 | Results processing circuits and methods associated with computational memory cells |
Family Applications Before (2)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
US16/152,374 Active US10891076B1 (en) | 2016-12-06 | 2018-10-04 | Results processing circuits and methods associated with computational memory cells |
US16/713,383 Active US11194519B2 (en) | 2016-12-06 | 2019-12-13 | Results processing circuits and methods associated with computational memory cells |
Country Status (1)
Country | Link |
---|---|
US (3) | US10891076B1 (en) |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US11194548B2 (en) | 2019-06-18 | 2021-12-07 | Gsi Technology, Inc. | Processing array device that performs one cycle full adder operation and bit line read/write logic features |
US11194519B2 (en) | 2016-12-06 | 2021-12-07 | Gsi Technology, Inc. | Results processing circuits and methods associated with computational memory cells |
US11205476B1 (en) | 2016-12-06 | 2021-12-21 | Gsi Technology, Inc. | Read data processing circuits and methods associated with computational memory cells |
US11409528B2 (en) | 2016-12-06 | 2022-08-09 | Gsi Technology, Inc. | Orthogonal data transposition system and method during data transfers to/from a processing array |
US11763881B2 (en) | 2016-12-06 | 2023-09-19 | Gsi Technology, Inc. | Computational memory cell and processing array device using the memory cells for XOR and XNOR computations |
Families Citing this family (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US10943648B1 (en) | 2016-12-06 | 2021-03-09 | Gsi Technology, Inc. | Ultra low VDD memory cell with ratioless write port |
US10847212B1 (en) | 2016-12-06 | 2020-11-24 | Gsi Technology, Inc. | Read and write data processing circuits and methods associated with computational memory cells using two read multiplexers |
US10847213B1 (en) | 2016-12-06 | 2020-11-24 | Gsi Technology, Inc. | Write data processing circuits and methods associated with computational memory cells |
US10770133B1 (en) | 2016-12-06 | 2020-09-08 | Gsi Technology, Inc. | Read and write data processing circuits and methods associated with computational memory cells that provides write inhibits and read bit line pre-charge inhibits |
US10725777B2 (en) | 2016-12-06 | 2020-07-28 | Gsi Technology, Inc. | Computational memory cell and processing array device using memory cells |
US11227653B1 (en) | 2016-12-06 | 2022-01-18 | Gsi Technology, Inc. | Storage array circuits and methods for computational memory cells |
US10930341B1 (en) | 2019-06-18 | 2021-02-23 | Gsi Technology, Inc. | Processing array device that performs one cycle full adder operation and bit line read/write logic features |
US10958272B2 (en) | 2019-06-18 | 2021-03-23 | Gsi Technology, Inc. | Computational memory cell and processing array device using complementary exclusive or memory cells |
Family Cites Families (372)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US3451694A (en) | 1967-09-21 | 1969-06-24 | Eaton Yale & Towne | Vehicle safety mechanism |
US3747952A (en) | 1971-09-17 | 1973-07-24 | R Graebe | Protective device |
US3795412A (en) | 1972-06-16 | 1974-03-05 | A John | Vehicle occupant restraint system |
US4227717A (en) | 1979-04-04 | 1980-10-14 | Bouvier Julien J | Motorcycle safety device |
US4308505A (en) | 1979-06-13 | 1981-12-29 | Trw, Inc. | Frequency detector device and method |
US4594564A (en) | 1984-06-11 | 1986-06-10 | Signetics Corporation | Frequency detector for use with phase locked loop |
US4741006A (en) | 1984-07-12 | 1988-04-26 | Kabushiki Kaisha Toshiba | Up/down counter device with reduced number of discrete circuit elements |
US4587496A (en) | 1984-09-12 | 1986-05-06 | General Signal Corporation | Fast acquisition phase-lock loop |
DE3526363A1 (en) | 1985-07-19 | 1987-01-22 | Siemens Ag | METHOD FOR PRODUCING AN ADJUSTABLE FREQUENCY GENERATOR |
US4716322A (en) | 1986-03-25 | 1987-12-29 | Texas Instruments Incorporated | Power-up control circuit including a comparator, Schmitt trigger, and latch |
DE3887924T3 (en) | 1987-06-02 | 1999-08-12 | Nat Semiconductor Corp | Non-volatile memory arrangement with a capacitive ferroelectric memory element. |
US4856035A (en) | 1988-05-26 | 1989-08-08 | Raytheon Company | CMOS binary up/down counter |
US5008636A (en) | 1988-10-28 | 1991-04-16 | Apollo Computer, Inc. | Apparatus for low skew system clock distribution and generation of 2X frequency clocks |
US5295258A (en) | 1989-12-22 | 1994-03-15 | Tandem Computers Incorporated | Fault-tolerant computer system with online recovery and reintegration of redundant components |
ATE179810T1 (en) | 1991-03-01 | 1999-05-15 | Advanced Micro Devices Inc | MICROPROCESSOR WITH EXTERNAL MEMORY |
JP2625277B2 (en) | 1991-05-20 | 1997-07-02 | 富士通株式会社 | Memory access device |
EP0554489B1 (en) | 1992-02-06 | 1999-12-01 | International Business Machines Corporation | Multi-port static random access memory with fast write-thru scheme |
US5646547A (en) | 1994-04-28 | 1997-07-08 | Xilinx, Inc. | Logic cell which can be configured as a latch without static one's problem |
US5440506A (en) | 1992-08-14 | 1995-08-08 | Harris Corporation | Semiconductor ROM device and method |
JP2667941B2 (en) | 1992-09-17 | 1997-10-27 | 三菱電機株式会社 | Memory cell circuit |
US5302916A (en) | 1992-12-21 | 1994-04-12 | At&T Bell Laboratories | Wide range digital frequency detector |
US5375089A (en) | 1993-10-05 | 1994-12-20 | Advanced Micro Devices, Inc. | Plural port memory system utilizing a memory having a read port and a write port |
US5382922A (en) | 1993-12-23 | 1995-01-17 | International Business Machines Corporation | Calibration systems and methods for setting PLL gain characteristics and center frequency |
EP0665650A1 (en) | 1994-01-31 | 1995-08-02 | STMicroelectronics S.A. | Low voltage high speed phase frequency detector |
US5400274A (en) | 1994-05-02 | 1995-03-21 | Motorola Inc. | Memory having looped global data lines for propagation delay matching |
JPH0897711A (en) | 1994-09-26 | 1996-04-12 | Nec Eng Ltd | Pll circuit |
US5530383A (en) | 1994-12-05 | 1996-06-25 | May; Michael R. | Method and apparatus for a frequency detection circuit for use in a phase locked loop |
US5587672A (en) | 1995-09-25 | 1996-12-24 | Neomagic Corp. | Dynamic logic having power-down mode with periodic clock refresh for a low-power graphics controller |
US5744991A (en) | 1995-10-16 | 1998-04-28 | Altera Corporation | System for distributing clocks using a delay lock loop in a programmable logic circuit |
US5696468A (en) | 1996-02-29 | 1997-12-09 | Qualcomm Incorporated | Method and apparatus for autocalibrating the center frequency of a voltage controlled oscillator of a phase locked loop |
US5838631A (en) | 1996-04-19 | 1998-11-17 | Integrated Device Technology, Inc. | Fully synchronous pipelined ram |
US5661419A (en) | 1996-05-23 | 1997-08-26 | Sun Microsystems, Inc. | Dynamic phase-frequency detector circuit |
AUPO072096A0 (en) | 1996-06-28 | 1996-07-25 | Curtin University Of Technology | Precise digital frequency detection |
US5748044A (en) | 1996-10-11 | 1998-05-05 | Silicon Motion, Inc. | Dual VCO phase-locked loop |
US5784332A (en) | 1996-12-12 | 1998-07-21 | Micron Technology Corporation | Clock frequency detector for a synchronous memory device |
US6167487A (en) | 1997-03-07 | 2000-12-26 | Mitsubishi Electronics America, Inc. | Multi-port RAM having functionally identical ports |
US6072741A (en) | 1997-04-01 | 2000-06-06 | Ramtron International Corporation | First-in, first-out integrated circuit memory device incorporating a retransmit function |
US6172935B1 (en) | 1997-04-25 | 2001-01-09 | Micron Technology, Inc. | Synchronous dynamic random access memory device |
US5937204A (en) | 1997-05-30 | 1999-08-10 | Helwett-Packard, Co. | Dual-pipeline architecture for enhancing the performance of graphics memory |
US6100736A (en) | 1997-06-05 | 2000-08-08 | Cirrus Logic, Inc | Frequency doubler using digital delay lock loop |
US6005794A (en) | 1997-06-27 | 1999-12-21 | Texas Instruments Incorporated | Static memory with low power write port |
US5977801A (en) | 1997-07-07 | 1999-11-02 | International Business Machines Corporation | Self-resetting phase/frequency detector with reduced dead zone |
US5942949A (en) | 1997-10-14 | 1999-08-24 | Lucent Technologies Inc. | Self-calibrating phase-lock loop with auto-trim operations for selecting an appropriate oscillator operating curve |
KR100269313B1 (en) | 1997-11-07 | 2000-12-01 | 윤종용 | Semiconductor memory device for consuming small current at stand-by state |
JP3094977B2 (en) | 1997-11-28 | 2000-10-03 | 日本電気株式会社 | PLL circuit |
JPH11176158A (en) | 1997-12-10 | 1999-07-02 | Fujitsu Ltd | Latch circuit, data output circuit, and semiconductor device with it |
GB2381622B (en) | 1997-12-17 | 2003-06-25 | Fujitsu Ltd | Synchronous dynamic random access memory apparatus |
US5963059A (en) | 1997-12-19 | 1999-10-05 | Advanced Micro Devices, Inc. | Phase frequency detector having reduced blind spot |
US5969576A (en) | 1997-12-22 | 1999-10-19 | Philips Electronics North America Corporation | Phase locked loop lock condition detector |
KR100364127B1 (en) | 1997-12-29 | 2003-04-11 | 주식회사 하이닉스반도체 | Chip-set |
US5969986A (en) | 1998-06-23 | 1999-10-19 | Invox Technology | High-bandwidth read and write architectures for non-volatile memories |
US6115320A (en) | 1998-02-23 | 2000-09-05 | Integrated Device Technology, Inc. | Separate byte control on fully synchronous pipelined SRAM |
US6262937B1 (en) | 1998-03-13 | 2001-07-17 | Cypress Semiconductor Corp. | Synchronous random access memory having a read/write address bus and process for writing to and reading from the same |
JPH11261017A (en) | 1998-03-16 | 1999-09-24 | Fujitsu Ltd | Semiconductor storage device |
JP3254427B2 (en) | 1998-10-09 | 2002-02-04 | インターナショナル・ビジネス・マシーンズ・コーポレーション | VCO characteristic calibration method |
US6700588B1 (en) | 1998-11-09 | 2004-03-02 | Broadcom Corporation | Apparatus and method for blending graphics and video surfaces |
US6100721A (en) | 1999-02-01 | 2000-08-08 | Motorola, Inc. | Circuit and method of extending the linear range of a phase frequency detector |
US6381684B1 (en) | 1999-04-26 | 2002-04-30 | Integrated Device Technology, Inc. | Quad data rate RAM |
US6757854B1 (en) | 1999-09-16 | 2004-06-29 | Lsi Logic Corporation | Detecting faults in dual port FIFO memories |
JP4707204B2 (en) | 1999-10-08 | 2011-06-22 | 富士通セミコンダクター株式会社 | Semiconductor memory device |
US6961546B1 (en) | 1999-10-21 | 2005-11-01 | Broadcom Corporation | Adaptive radio transceiver with offset PLL with subsampling mixers |
US6265902B1 (en) | 1999-11-02 | 2001-07-24 | Ericsson Inc. | Slip-detecting phase detector and method for improving phase-lock loop lock time |
JP3292188B2 (en) | 1999-11-10 | 2002-06-17 | 日本電気株式会社 | PLL circuit |
US6853696B1 (en) | 1999-12-20 | 2005-02-08 | Nortel Networks Limited | Method and apparatus for clock recovery and data qualification |
US6683930B1 (en) | 1999-12-23 | 2004-01-27 | Cypress Semiconductor Corp. | Digital phase/frequency detector, and clock generator and data recovery PLL containing the same |
JP3375584B2 (en) | 2000-01-07 | 2003-02-10 | 松下電器産業株式会社 | Frequency comparator and phase-locked loop with it |
US6226217B1 (en) | 2000-02-18 | 2001-05-01 | Hewlett-Packard Company | Register structure with a dual-ended write mechanism |
US6538475B1 (en) | 2000-03-15 | 2003-03-25 | Intel Corporation | Phase detector |
US6310880B1 (en) | 2000-03-17 | 2001-10-30 | Silicon Aquarius, Inc. | Content addressable memory cells and systems and devices using the same |
JP4810051B2 (en) | 2000-06-07 | 2011-11-09 | コネクサント システムズ,アイエヌシー. | Method and apparatus for media access control in power line communication network system |
KR100358118B1 (en) | 2000-06-08 | 2002-10-25 | 한국전자통신연구원 | Phase locked loop having high-speed locking |
JP4748877B2 (en) | 2000-07-10 | 2011-08-17 | ルネサスエレクトロニクス株式会社 | Storage device |
US6538954B2 (en) | 2000-07-10 | 2003-03-25 | Mitsubishi Denki Kabushiki Kaisha | Multi-port static random access memory equipped with a write control line |
US6366524B1 (en) | 2000-07-28 | 2002-04-02 | Micron Technology Inc. | Address decoding in multiple-bank memory architectures |
KR100380347B1 (en) | 2000-11-21 | 2003-04-11 | 삼성전자주식회사 | Semiconductor memory device and data read method thereof |
US6732247B2 (en) | 2001-01-17 | 2004-05-04 | University Of Washington | Multi-ported memory having pipelined data banks |
US6385122B1 (en) | 2001-01-31 | 2002-05-07 | Virage Logic Corp. | Row and column accessible memory with a built-in multiplex |
JP4164241B2 (en) | 2001-02-15 | 2008-10-15 | 株式会社ルネサステクノロジ | Semiconductor device |
US6441691B1 (en) | 2001-03-09 | 2002-08-27 | Ericsson Inc. | PLL cycle slip compensation |
US7003065B2 (en) | 2001-03-09 | 2006-02-21 | Ericsson Inc. | PLL cycle slip detection |
US7088604B2 (en) | 2001-03-15 | 2006-08-08 | Micron Technology, Inc. | Multi-bank memory |
KR100508083B1 (en) | 2001-03-21 | 2005-08-17 | 삼성전자주식회사 | Symbol combining device for multi-path diversity |
KR100415192B1 (en) | 2001-04-18 | 2004-01-16 | 삼성전자주식회사 | Method and apparatus for read operation and write operation in semiconductor memory device |
US6744277B1 (en) | 2001-05-06 | 2004-06-01 | Altera Corporation | Programmable current reference circuit |
US6483361B1 (en) | 2001-05-18 | 2002-11-19 | National Semiconductor Corporation | Lock detector for determining phase lock condition in PLL on a period-by-period basis according to desired phase error |
ITMI20011291A1 (en) | 2001-06-19 | 2002-12-19 | St Microelectronics Srl | AUTOMATIC CALIBRATION METHOD OF A PHASE LOCKING SYSTEM |
US6594194B2 (en) | 2001-07-11 | 2003-07-15 | Sun Microsystems, Inc. | Memory array with common word line |
WO2003009142A1 (en) | 2001-07-17 | 2003-01-30 | Alcatel Internetworking, Inc. | Switch fabric with dual port memory emulation scheme |
JP2003045200A (en) | 2001-08-02 | 2003-02-14 | Mitsubishi Electric Corp | Semiconductor module and semiconductor memory used for the same |
US6504417B1 (en) | 2001-08-15 | 2003-01-07 | International Business Machines Corporation | Active trim circuit for CMOS on-chip resistors |
US6859107B1 (en) | 2001-09-05 | 2005-02-22 | Silicon Image, Inc. | Frequency comparator with hysteresis between locked and unlocked conditions |
JP3883126B2 (en) | 2001-10-26 | 2007-02-21 | 富士通株式会社 | Semiconductor integrated circuit device, electronic device incorporating the same, and power consumption reduction method |
US6473334B1 (en) | 2001-10-31 | 2002-10-29 | Compaq Information Technologies Group, L.P. | Multi-ported SRAM cell with shared bit and word lines and separate read and write ports |
JP2003173681A (en) | 2001-12-07 | 2003-06-20 | Mitsubishi Electric Corp | Semiconductor memory circuit and latch circuit |
US7471756B2 (en) | 2002-02-21 | 2008-12-30 | Intersil Americas Inc. | Phase detector and method for a shortening phase-error-correction pulse |
US6683502B1 (en) | 2002-03-12 | 2004-01-27 | Xilinx, Inc. | Process compensated phase locked loop |
US6642747B1 (en) | 2002-03-15 | 2003-11-04 | National Semiconductor Corporation | Frequency detector for a phase locked loop system |
US6661267B2 (en) | 2002-05-06 | 2003-12-09 | International Business Machines Corporation | Coarse calibration circuit using variable step sizes to reduce jitter and a dynamic course calibration (DCC) circuit for a 2 GHz VCO |
US7340577B1 (en) | 2002-05-29 | 2008-03-04 | Nvidia Corporation | Method and system for efficiently executing reads after writes in a memory employing delayed write data |
GB2389254B (en) | 2002-05-31 | 2005-09-07 | Hitachi Ltd | Semiconductor integrated circuit device for communication |
US6838951B1 (en) | 2002-06-12 | 2005-01-04 | Rf Micro Devices, Inc. | Frequency synthesizer having VCO bias current compensation |
US7043599B1 (en) | 2002-06-20 | 2006-05-09 | Rambus Inc. | Dynamic memory supporting simultaneous refresh and data-access transactions |
KR100447030B1 (en) | 2002-08-22 | 2004-09-07 | 삼성전자주식회사 | Semiconductor device applying well bias and method of fabricating the same |
US6836419B2 (en) | 2002-08-23 | 2004-12-28 | Micron Technology, Inc. | Split word line ternary CAM architecture |
US6940328B2 (en) | 2002-08-28 | 2005-09-06 | Micron Technology, Inc. | Methods and apparatus for duty cycle control |
US6938142B2 (en) | 2002-08-28 | 2005-08-30 | Micron Technology, Inc. | Multi-bank memory accesses using posted writes |
WO2004021574A1 (en) | 2002-08-30 | 2004-03-11 | Koninklijke Philips Electronics N.V. | Phase locked loop |
US20040053510A1 (en) | 2002-09-16 | 2004-03-18 | Little Casey J. | System for and method of unlimited voltage multi ported sram cells |
JP2004110906A (en) | 2002-09-17 | 2004-04-08 | Renesas Technology Corp | Semiconductor memory |
US6985400B2 (en) | 2002-09-30 | 2006-01-10 | Infineon Technologies Ag | On-die detection of the system operation frequency in a DRAM to adjust DRAM operations |
US7327597B1 (en) | 2002-10-02 | 2008-02-05 | Cisco Technology, Inc. | Static random access memory architecture |
TWI283515B (en) | 2002-10-02 | 2007-07-01 | Via Tech Inc | Method and device for adjusting reference level |
KR100481504B1 (en) | 2002-11-12 | 2005-04-07 | 삼성전자주식회사 | Controlling apparatus of sampling phase for digital display apparatus and controlling method thereof |
JP2004173177A (en) | 2002-11-22 | 2004-06-17 | Nec Corp | Pll circuit |
CN100524148C (en) | 2002-11-29 | 2009-08-05 | 松下电器产业株式会社 | Parameter correction circuit and parameter correction method |
KR100531469B1 (en) | 2003-01-09 | 2005-11-28 | 주식회사 하이닉스반도체 | Analog Delay Lock Loop with store circuit about delay lock |
DE10308527B4 (en) | 2003-02-27 | 2012-05-31 | Lantiq Deutschland Gmbh | Tuning circuit and method for tuning a filter stage |
GB2400760B (en) | 2003-04-14 | 2005-12-21 | Wolfson Ltd | Improved phase/frequency detector and phase lock loop circuit |
US6975554B1 (en) | 2003-04-30 | 2005-12-13 | Advanced Micro Devices, Inc. | Method and system for providing a shared write driver |
US6882237B2 (en) | 2003-04-30 | 2005-04-19 | Zarlink Semiconductor Inc. | Capture range control mechanism for voltage controlled oscillators |
CA2526467C (en) | 2003-05-20 | 2015-03-03 | Kagutech Ltd. | Digital backplane recursive feedback control |
US7099643B2 (en) | 2003-05-27 | 2006-08-29 | Broadcom Corporation | Analog open-loop VCO calibration method |
US7139213B2 (en) | 2003-06-02 | 2006-11-21 | Silicon Aquarius, Inc. | Multiple data path memories and systems |
US7254690B2 (en) | 2003-06-02 | 2007-08-07 | S. Aqua Semiconductor Llc | Pipelined semiconductor memories and systems |
US7454555B2 (en) | 2003-06-12 | 2008-11-18 | Rambus Inc. | Apparatus and method including a memory device having multiple sets of memory banks with duplicated data emulating a fast access time, fixed latency memory device |
US7477716B2 (en) | 2003-06-25 | 2009-01-13 | Mosaid Technologies, Inc. | Start up circuit for delay locked loop |
US6845059B1 (en) * | 2003-06-26 | 2005-01-18 | International Business Machines Corporation | High performance gain cell architecture |
JP4019021B2 (en) | 2003-07-14 | 2007-12-05 | 日本テキサス・インスツルメンツ株式会社 | Semiconductor memory cell |
US7158425B2 (en) | 2003-07-28 | 2007-01-02 | Mosaic Systems, Inc. | System and method for providing a redundant memory array in a semiconductor memory integrated circuit |
US7046093B1 (en) | 2003-08-27 | 2006-05-16 | Intergrated Device Technology, Inc. | Dynamic phase-locked loop circuits and methods of operation thereof |
US7464282B1 (en) | 2003-09-03 | 2008-12-09 | T-Ram Semiconductor, Inc. | Apparatus and method for producing dummy data and output clock generator using same |
US6998922B2 (en) | 2003-09-08 | 2006-02-14 | Broadcom Corp. | Phase locked loop modulator calibration techniques |
US7349515B1 (en) | 2003-09-22 | 2008-03-25 | Cypress Semiconductor Corporation | Method and an apparatus to improve production yield of phase locked loops |
US7352249B2 (en) | 2003-10-03 | 2008-04-01 | Analog Devices, Inc. | Phase-locked loop bandwidth calibration circuit and method thereof |
US6933789B2 (en) | 2003-11-13 | 2005-08-23 | Skyworks Solutions, Inc. | On-chip VCO calibration |
US7263152B2 (en) | 2003-11-18 | 2007-08-28 | Analog Devices, Inc. | Phase-locked loop structures with enhanced signal stability |
US6954091B2 (en) | 2003-11-25 | 2005-10-11 | Lsi Logic Corporation | Programmable phase-locked loop |
US7246215B2 (en) | 2003-11-26 | 2007-07-17 | Intel Corporation | Systolic memory arrays |
US7047146B2 (en) | 2003-12-19 | 2006-05-16 | Airoha Technology Corp | Method for automatically calibrating the frequency range of a PLL and associated PLL capable of automatic calibration |
US7042792B2 (en) | 2004-01-14 | 2006-05-09 | Integrated Device Technology, Inc. | Multi-port memory cells for use in FIFO applications that support data transfers between cache and supplemental memory arrays |
JP4260034B2 (en) | 2004-01-30 | 2009-04-30 | 三洋電機株式会社 | Clock generation method and clock generation apparatus |
US7200713B2 (en) | 2004-03-29 | 2007-04-03 | Intel Corporation | Method of implementing off-chip cache memory in dual-use SRAM memory for network processors |
US20050226079A1 (en) | 2004-04-08 | 2005-10-13 | Yiming Zhu | Methods and apparatus for dual port memory devices having hidden refresh and double bandwidth |
JP4651298B2 (en) | 2004-04-08 | 2011-03-16 | 三菱電機株式会社 | Automatic frequency correction PLL circuit |
TWI237119B (en) | 2004-04-15 | 2005-08-01 | Mediatek Inc | Phase locked loop, adaptive frequency detector and related method thereof |
KR100574980B1 (en) | 2004-04-26 | 2006-05-02 | 삼성전자주식회사 | Phase-Locked Loop for fast frequency locking |
US7042271B2 (en) | 2004-05-06 | 2006-05-09 | Broadcom Corporation | Resistor compensation apparatus |
US7489164B2 (en) | 2004-05-17 | 2009-02-10 | Raminda Udaya Madurawe | Multi-port memory devices |
TWI232023B (en) | 2004-05-21 | 2005-05-01 | Sunplus Technology Co Ltd | Voltage control oscillator |
JP4662532B2 (en) | 2004-06-03 | 2011-03-30 | パナソニック株式会社 | Semiconductor memory device |
JP2006127460A (en) | 2004-06-09 | 2006-05-18 | Renesas Technology Corp | Semiconductor device, semiconductor signal processing apparatus and crossbar switch |
US7142477B1 (en) | 2004-06-18 | 2006-11-28 | Cypress Semiconductor Corp. | Memory interface system and method for reducing cycle time of sequential read and write accesses using separate address and data buses |
US7002416B2 (en) | 2004-06-29 | 2006-02-21 | Micrel, Inc. | Circuit for driving a voltage controlled oscillator for frequency modulation |
US7019569B2 (en) | 2004-07-09 | 2006-03-28 | Faraday Technology Corp. | Method of implementing multi-transfer curve phase lock loop |
DE102004037162A1 (en) | 2004-07-30 | 2006-03-23 | Infineon Technologies Ag | Method and device for generating a clock signal |
KR100684067B1 (en) | 2004-08-13 | 2007-02-16 | 삼성전자주식회사 | Voltage- controlled oscillator capable of limiting an operating frequency range and having temperature compensation |
US8190808B2 (en) | 2004-08-17 | 2012-05-29 | Rambus Inc. | Memory device having staggered memory operations |
KR100618860B1 (en) | 2004-09-09 | 2006-08-31 | 삼성전자주식회사 | Address coding method and address decoder for reducing sensing noise during refresh operation of memory device |
US7280428B2 (en) | 2004-09-30 | 2007-10-09 | Rambus Inc. | Multi-column addressing mode memory system including an integrated circuit memory device |
US7330080B1 (en) | 2004-11-04 | 2008-02-12 | Transmeta Corporation | Ring based impedance control of an output driver |
KR100668360B1 (en) | 2004-11-09 | 2007-01-16 | 한국전자통신연구원 | Phase frequency detector |
US20060119443A1 (en) | 2004-12-08 | 2006-06-08 | Via Technologies Inc. | Damping coefficient variation mechanism in a phase locked loop |
JP2006164183A (en) | 2004-12-10 | 2006-06-22 | Renesas Technology Corp | Semiconductor signal processing device |
US7095287B2 (en) | 2004-12-28 | 2006-08-22 | Silicon Laboratories Inc. | Method and apparatus to achieve a process, temperature and divider modulus independent PLL loop bandwidth and damping factor using open-loop calibration techniques |
US7375593B2 (en) | 2005-01-19 | 2008-05-20 | Paul William Ronald Self | Circuits and methods of generating and controlling signals on an integrated circuit |
US7746181B1 (en) | 2005-01-28 | 2010-06-29 | Cypress Semiconductor Corporation | Circuit and method for extending the usable frequency range of a phase locked loop (PLL) |
JP4873182B2 (en) | 2005-02-03 | 2012-02-08 | 日本電気株式会社 | Semiconductor memory device and driving method thereof |
KR100562655B1 (en) | 2005-02-28 | 2006-03-20 | 주식회사 하이닉스반도체 | Driving limitation filter in semiconductor memory device and its method |
US20070189101A1 (en) | 2005-05-17 | 2007-08-16 | Atmel Corporation | Fast read port for register file |
US20060248305A1 (en) | 2005-04-13 | 2006-11-02 | Wayne Fang | Memory device having width-dependent output latency |
US7636019B1 (en) | 2005-06-07 | 2009-12-22 | Cypress Semiconductor Corporation | Phase lock loop pre-charging system and method |
US20070001721A1 (en) | 2005-07-01 | 2007-01-04 | Chi-Yang Chen | Power-on reset circuit |
US7577225B2 (en) | 2005-07-28 | 2009-08-18 | Agere Systems Inc. | Digital phase-looked loop |
JP4749089B2 (en) | 2005-08-26 | 2011-08-17 | ルネサスエレクトロニクス株式会社 | Semiconductor device |
JP2005346922A (en) | 2005-09-01 | 2005-12-15 | Renesas Technology Corp | Synchronous semiconductor memory apparatus |
JP4738112B2 (en) | 2005-09-12 | 2011-08-03 | ルネサスエレクトロニクス株式会社 | Semiconductor memory device |
US7487315B2 (en) | 2005-09-14 | 2009-02-03 | Via Technologies, Inc. | Accessing apparatus capable of reducing power consumption and accessing method thereof |
US7439816B1 (en) | 2005-09-28 | 2008-10-21 | Cypress Semiconductor Corporation | Phase-locked loop fast lock circuit and method |
US7389457B2 (en) | 2005-10-28 | 2008-06-17 | Sony Corporation | Shift registers free of timing race boundary scan registers with two-phase clock control |
US7313040B2 (en) | 2005-10-28 | 2007-12-25 | Sony Corporation | Dynamic sense amplifier for SRAM |
US8542050B2 (en) | 2005-10-28 | 2013-09-24 | Sony Corporation | Minimized line skew generator |
US7355907B2 (en) | 2005-10-28 | 2008-04-08 | Sony Corporation | Performing read and write operations in the same cycle for an SRAM device |
TWI295806B (en) | 2005-11-24 | 2008-04-11 | Via Tech Inc | Output circuit of sram |
JP2007193928A (en) | 2005-12-19 | 2007-08-02 | Matsushita Electric Ind Co Ltd | Semiconductor memory |
US7663415B2 (en) | 2005-12-30 | 2010-02-16 | Stmicroelectronics Pvt. Ltd. | Phase locked loop (PLL) method and architecture |
WO2007085003A2 (en) | 2006-01-20 | 2007-07-26 | Sirf Technology, Inc. | Method of eliminating temperature induced band switching in ultra wideband voltage controlled oscillator |
US7336533B2 (en) | 2006-01-23 | 2008-02-26 | Freescale Semiconductor, Inc. | Electronic device and method for operating a memory circuit |
US7355482B2 (en) | 2006-02-17 | 2008-04-08 | Seiko Epson Corporation | Methods and apparatus for compensating a variable oscillator for process, voltage, and temperature variations using a replica oscillator |
US7362184B2 (en) | 2006-02-28 | 2008-04-22 | International Business Machines Corporation | Frequency divider monitor of phase lock loop |
JP2007266935A (en) | 2006-03-28 | 2007-10-11 | Nec Corp | Pll circuit |
US7728675B1 (en) | 2006-03-31 | 2010-06-01 | Cypress Semiconductor Corporation | Fast lock circuit for a phase lock loop |
US7516385B2 (en) | 2006-04-28 | 2009-04-07 | Sony Corporation | Test semiconductor device in full frequency with half frequency tester |
US7312629B2 (en) | 2006-05-17 | 2007-12-25 | Sony Corporation | Programmable impedance control circuit calibrated at Voh, Vol level |
US7689941B1 (en) | 2006-06-01 | 2010-03-30 | Altera Corporation | Write margin calculation tool for dual-port random-access-memory circuitry |
JP2008034037A (en) | 2006-07-28 | 2008-02-14 | Toshiba Corp | Semiconductor storage device |
US7813161B2 (en) | 2006-08-30 | 2010-10-12 | Stmicroelectronics Pvt. Ltd | Dual port SRAM with dedicated read and write ports for high speed read operation and low leakage |
TWI332318B (en) | 2006-09-07 | 2010-10-21 | Realtek Semiconductor Corp | Multiloop phase locked loop circuit |
JP4842064B2 (en) | 2006-09-14 | 2011-12-21 | ルネサスエレクトロニクス株式会社 | PLL circuit |
US7471158B2 (en) | 2006-09-19 | 2008-12-30 | Faraday Technology Corp. | Automatic switching phase-locked loop |
US7502273B2 (en) | 2006-09-27 | 2009-03-10 | Taiwan Semiconductor Manufacturing Co., Ltd. | Two-port SRAM with a high speed sensing scheme |
US7965112B2 (en) | 2006-09-29 | 2011-06-21 | Mediatek Inc. | Power-on reset circuits |
US8261138B2 (en) | 2006-10-24 | 2012-09-04 | International Business Machines Corporation | Test structure for characterizing multi-port static random access memory and register file arrays |
US7623404B2 (en) | 2006-11-20 | 2009-11-24 | Freescale Semiconductor, Inc. | Memory device having concurrent write and read cycles and method thereof |
US7660149B2 (en) | 2006-12-07 | 2010-02-09 | Taiwan Semiconductor Manufacturing Co., Ltd. | SRAM cell with separate read and write ports |
FR2910999B1 (en) | 2006-12-28 | 2009-04-03 | Commissariat Energie Atomique | MEMORY CELL WITH DOUBLE-GRID TRANSISTORS, INDEPENDENT AND ASYMMETRIC GRIDS |
US7495969B2 (en) | 2007-01-30 | 2009-02-24 | International Business Machines Corporation | Techniques for improving write stability of memory with decoupled read and write bit lines |
JP4357538B2 (en) | 2007-03-07 | 2009-11-04 | 株式会社日立製作所 | Semiconductor integrated circuit device |
US7592847B2 (en) | 2007-03-22 | 2009-09-22 | Mediatek Inc. | Phase frequency detector and phase-locked loop |
KR100854419B1 (en) | 2007-03-31 | 2008-08-26 | 주식회사 하이닉스반도체 | Power-up signal generator |
US7633322B1 (en) | 2007-04-06 | 2009-12-15 | Altera Corporation | Digital loop circuit for programmable logic device |
US7760032B2 (en) | 2007-04-20 | 2010-07-20 | Tialinx, Inc. | Self-compensating voltage-controlled oscillator |
US20080265957A1 (en) | 2007-04-25 | 2008-10-30 | Luong Trong V | Self-Resetting Phase Frequency Detector with Multiple Ranges of Clock Difference |
US7605667B2 (en) | 2007-04-26 | 2009-10-20 | Mediatek Inc. | Frequency synthesizer with a harmonic locked phase/frequency detector |
US7595657B2 (en) | 2007-05-01 | 2009-09-29 | Sony Corporation | Dynamic dual control on-die termination |
US7558095B2 (en) | 2007-05-02 | 2009-07-07 | Agere Systems Inc. | Memory cell for content-addressable memory |
US8004920B2 (en) | 2007-05-29 | 2011-08-23 | Micron Technology, Inc. | Power saving memory apparatus, systems, and methods |
US7920409B1 (en) | 2007-06-05 | 2011-04-05 | Arizona Board Of Regents For And On Behalf Of Arizona State University | SRAM cell with intrinsically high stability and low leakage |
US7956695B1 (en) | 2007-06-12 | 2011-06-07 | Altera Corporation | High-frequency low-gain ring VCO for clock-data recovery in high-speed serial interface of a programmable logic device |
US7719329B1 (en) | 2007-06-15 | 2010-05-18 | Cypress Semiconductor Corporation | Phase-locked loop fast lock circuit and method |
KR100944497B1 (en) | 2007-06-25 | 2010-03-03 | 삼성전자주식회사 | Digital frequency detector and digital Phase Locked Loop using the digital frequency detector |
US7848725B2 (en) | 2007-07-02 | 2010-12-07 | Broadcom Corporation | RF transmitter with stable on-chip PLL |
US7659783B2 (en) | 2007-07-16 | 2010-02-09 | Micrel, Inc. | System and method for phase-locked loop (PLL) for high-speed memory interface (HSMI) |
US7830212B2 (en) | 2007-07-30 | 2010-11-09 | Mediatek Inc. | Phase locked loop, voltage controlled oscillator, and phase-frequency detector |
JP2009038778A (en) | 2007-08-06 | 2009-02-19 | Nec Electronics Corp | Vco circuit and pll circuit using the same |
US7965564B2 (en) | 2007-09-18 | 2011-06-21 | Zikbit Ltd. | Processor arrays made of standard memory cells |
JP2009104757A (en) | 2007-10-02 | 2009-05-14 | Panasonic Corp | Semiconductor storage device |
US7885138B2 (en) | 2007-10-19 | 2011-02-08 | International Business Machines Corporation | Three dimensional twisted bitline architecture for multi-port memory |
US7746182B2 (en) | 2007-11-02 | 2010-06-29 | Texas Instruments Incorporated | Systems and methods for voltage controlled oscillator calibration |
US7635988B2 (en) | 2007-11-19 | 2009-12-22 | Tier Logic, Inc. | Multi-port thin-film memory devices |
US7646648B2 (en) | 2007-12-03 | 2010-01-12 | International Business Machines Corporation | Apparatus and method for implementing memory array device with built in computational capability |
US8117567B2 (en) | 2007-12-03 | 2012-02-14 | International Business Machines Corporation | Structure for implementing memory array device with built in computation capability |
JP5228468B2 (en) | 2007-12-17 | 2013-07-03 | 富士通セミコンダクター株式会社 | System device and method of operating system device |
US7719330B2 (en) | 2007-12-26 | 2010-05-18 | Ali Corporation | Phase locked loop device and control method thereof |
KR20090074412A (en) | 2008-01-02 | 2009-07-07 | 삼성전자주식회사 | Circuit of dividing the frequency and phase locked loop using the same |
US7737743B1 (en) | 2008-03-07 | 2010-06-15 | National Semiconductor Corporation | Phase-locked loop including sampling phase detector and charge pump with pulse width control |
US8120985B2 (en) | 2008-03-12 | 2012-02-21 | Qimonda Ag | Multi-bank memory device method and apparatus |
US7646215B2 (en) | 2008-03-24 | 2010-01-12 | Sony Corporation | Efficient method for implementing programmable impedance output drivers and programmable input on die termination on a bi-directional data bus |
GB2459108A (en) | 2008-04-09 | 2009-10-14 | Wolfson Microelectronics Plc | Dithered clock signal generator |
US8451971B2 (en) | 2008-05-30 | 2013-05-28 | Mediatek Inc. | Communication systems, clock generation circuits thereof, and method for generating clock signal |
KR20090126087A (en) | 2008-06-03 | 2009-12-08 | 현대자동차주식회사 | Air bag apparatus for a vehicle |
JP2010009141A (en) | 2008-06-24 | 2010-01-14 | Toshiba Corp | Data transfer method |
US7782655B2 (en) | 2008-07-01 | 2010-08-24 | Jeng-Jye Shau | Ultra-low power hybrid sub-threshold circuits |
US7782656B2 (en) | 2008-07-23 | 2010-08-24 | Taiwan Semiconductor Manufacturing Co., Ltd. | SRAM with improved read/write stability |
TWI353730B (en) | 2008-07-25 | 2011-12-01 | Etron Technology Inc | Phase/frequency detector |
US20100085086A1 (en) | 2008-07-29 | 2010-04-08 | Fujitsu Limited | Digital Frequency Detector |
US7859919B2 (en) | 2008-08-27 | 2010-12-28 | Freescale Semiconductor, Inc. | Memory device and method thereof |
US7843721B1 (en) | 2008-09-18 | 2010-11-30 | Suvolta, Inc. | Memory cell including an emitter follower and emitter follower sensing scheme and method of reading data therefrom |
US8044724B2 (en) | 2008-09-22 | 2011-10-25 | Mosys, Inc. | Low jitter large frequency tuning LC PLL for multi-speed clocking applications |
US7839177B1 (en) | 2008-11-07 | 2010-11-23 | Altera Corporation | Techniques for phase detection with fast reset |
US8194481B2 (en) | 2008-12-18 | 2012-06-05 | Mosaid Technologies Incorporated | Semiconductor device with main memory unit and auxiliary memory unit requiring preset operation |
JP5193846B2 (en) | 2008-12-25 | 2013-05-08 | 株式会社東芝 | Synchronization circuit |
US7929329B2 (en) | 2009-01-14 | 2011-04-19 | Micron Technology, Inc. | Memory bank signal coupling buffer and method |
JP2010170595A (en) | 2009-01-20 | 2010-08-05 | Panasonic Corp | Semiconductor memory device |
US7990780B2 (en) | 2009-02-20 | 2011-08-02 | Apple Inc. | Multiple threshold voltage register file cell |
US7940599B2 (en) | 2009-03-16 | 2011-05-10 | Freescale Semiconductor, Inc. | Dual port memory device |
US7940088B1 (en) | 2009-03-31 | 2011-05-10 | Pmc-Sierra, Inc. | High speed phase frequency detector |
US7961536B2 (en) | 2009-04-13 | 2011-06-14 | Advanced Micro Devices, Inc. | Memory device and methods thereof |
US7907022B2 (en) | 2009-04-23 | 2011-03-15 | Freescale Semiconductor, Inc. | Phase-locked loop and method for operating the same |
US8063707B2 (en) | 2009-05-08 | 2011-11-22 | Mediatek Inc. | Phase locked loop |
JP5441208B2 (en) | 2009-06-19 | 2014-03-12 | ルネサスエレクトロニクス株式会社 | Semiconductor memory device |
US8385106B2 (en) | 2009-09-11 | 2013-02-26 | Grandis, Inc. | Method and system for providing a hierarchical data path for spin transfer torque random access memory |
US8258831B1 (en) | 2009-11-09 | 2012-09-04 | Marvell Israel (M.I.S.L) Ltd. | Method and apparatus for clock generator lock detector |
KR101045072B1 (en) | 2009-12-28 | 2011-06-29 | 주식회사 하이닉스반도체 | Phase locked loop and method for driving the same |
US8638144B1 (en) | 2009-12-30 | 2014-01-28 | Gsi Technology, Inc. | Systems and methods involving phase detection with adaptive locking/detection features |
US9356611B1 (en) | 2009-12-30 | 2016-05-31 | Gsi Technology, Inc. | Systems and methods involving phase detection with adaptive locking/detection features |
US8325549B2 (en) | 2010-02-26 | 2012-12-04 | International Business Machines Corporation | Global bit select circuit interface with simplified write bit line precharging |
US8284593B2 (en) | 2010-04-14 | 2012-10-09 | Freescale Semiconductor, Inc. | Multi-port memory having a variable number of used write ports |
US8008956B1 (en) | 2010-05-18 | 2011-08-30 | Kwangwoon University Industry-Academic Collaboration Foundation | Frequency synthesizer and high-speed automatic calibration device therefor |
US8208314B2 (en) | 2010-06-01 | 2012-06-26 | Aptina Imaging Corporation | Sequential access memory elements |
JP5598161B2 (en) | 2010-08-26 | 2014-10-01 | ヤマハ株式会社 | Clock generation circuit |
US8213242B2 (en) | 2010-09-23 | 2012-07-03 | Taiwan Semiconductor Manufacturing Company, Ltd. | Memory cells having a row-based read and/or write support circuitry |
KR101241157B1 (en) | 2010-11-29 | 2013-03-11 | 현대자동차주식회사 | Intternal air bag device |
KR101183626B1 (en) | 2010-12-17 | 2012-09-17 | 에스케이하이닉스 주식회사 | Clock signal generation circuit |
US8625333B2 (en) | 2011-02-22 | 2014-01-07 | Lsi Corporation | Memory device having memory cells with write assist functionality |
US8294502B2 (en) | 2011-03-04 | 2012-10-23 | Altera Corporation | Delay circuitry |
JP2012205135A (en) | 2011-03-25 | 2012-10-22 | Toshiba Corp | Phase adjustment circuit and semiconductor device |
US8593896B2 (en) | 2011-03-30 | 2013-11-26 | Taiwan Semiconductor Manufacturing Company, Ltd. | Differential read write back sense amplifier circuits and methods |
US20120281459A1 (en) | 2011-05-08 | 2012-11-08 | Ben-Gurion University Of The Negev Research And Development Authority | Ultra low power memory cell with a supply feedback loop configured for minimal leakage operation |
US8643418B2 (en) | 2011-06-02 | 2014-02-04 | Micron Technology, Inc. | Apparatus and methods for altering the timing of a clock signal |
US8493774B2 (en) | 2011-06-17 | 2013-07-23 | International Business Machines Corporation | Performing logic functions on more than one memory cell within an array of memory cells |
US8576655B2 (en) | 2011-06-21 | 2013-11-05 | Taiwan Semiconductor Manufacturing Co., Ltd. | Semiconductor memories |
US8400200B1 (en) | 2011-07-09 | 2013-03-19 | Gsi Technology, Inc. | Systems and methods including features of power supply noise reduction and/or power-saving for high speed delay lines |
US8488408B1 (en) | 2011-07-20 | 2013-07-16 | Gsi Technology, Inc. | Systems and methods including clock features such as minimization of simultaneous switching outputs (SSO) effects involving echo clocks |
US8982649B2 (en) | 2011-08-12 | 2015-03-17 | Gsi Technology, Inc. | Systems and methods involving multi-bank, dual- or multi-pipe SRAMs |
US8526256B2 (en) | 2011-09-16 | 2013-09-03 | International Business Machines Corporation | Single-ended sense amplifier with read-assist |
US20130083591A1 (en) | 2011-09-29 | 2013-04-04 | John J. Wuu | Alternating Wordline Connection in 8T Cells for Improving Resiliency to Multi-Bit SER Upsets |
US8817550B1 (en) | 2011-12-02 | 2014-08-26 | Gsi Technology, Inc. | Systems and methods of semiconductor memory devices including features of output buffer initialization circuit(s) and/or multiple power-up detection/handling |
US8693236B2 (en) | 2011-12-09 | 2014-04-08 | Gsi Technology, Inc. | Systems and methods of sectioned bit line memory arrays, including hierarchical and/or other features |
US8593860B2 (en) | 2011-12-09 | 2013-11-26 | Gsi Technology, Inc. | Systems and methods of sectioned bit line memory arrays |
US8625334B2 (en) | 2011-12-16 | 2014-01-07 | Taiwan Semiconductor Manufacturing Company, Ltd. | Memory cell |
US8692621B2 (en) | 2011-12-21 | 2014-04-08 | Fairchild Semiconductor Corporation | Methods and apparatus for oscillator frequency calibration |
US8654570B2 (en) | 2011-12-29 | 2014-02-18 | Stmicroelectronics International N.V. | Low voltage write time enhanced SRAM cell and circuit extensions |
US9240229B1 (en) | 2012-03-15 | 2016-01-19 | Gsi Technology, Inc. | Systems and methods involving control-I/O buffer enable circuits and/or features of saving power in standby mode |
US8760958B2 (en) | 2012-03-15 | 2014-06-24 | Memoir Systems, Inc. | Methods and apparatus for designing and constructing multi-port memory circuits with voltage assist |
US8705310B2 (en) | 2012-08-24 | 2014-04-22 | Cypress Semiconductor Corporation | Access methods and circuits for memory devices having multiple banks |
US9954517B2 (en) | 2012-11-06 | 2018-04-24 | Micron Technology, Inc. | Apparatuses and methods for duty cycle adjustment |
US9093135B2 (en) | 2012-11-15 | 2015-07-28 | Nvidia Corporation | System, method, and computer program product for implementing a storage array |
US9692429B1 (en) | 2012-11-15 | 2017-06-27 | Gsi Technology, Inc. | Systems and methods involving fast-acquisition lock features associated with phase locked loop circuitry |
US9070477B1 (en) | 2012-12-12 | 2015-06-30 | Mie Fujitsu Semiconductor Limited | Bit interleaved low voltage static random access memory (SRAM) and related methods |
US9159391B1 (en) | 2012-12-13 | 2015-10-13 | Gsi Technology, Inc. | Systems and methods of double/quad data rate memory involving input latching, self-timing and/or other features |
US9431079B1 (en) | 2012-12-13 | 2016-08-30 | Gsi Technology, Inc. | Systems and methods of memory and memory operation involving input latching, self-timing and/or other features |
US9059691B2 (en) | 2012-12-31 | 2015-06-16 | Taiwan Semiconductor Manufacturing Company, Ltd. | Duty cycle detection and correction circuit in an integrated circuit |
US8879303B2 (en) | 2013-01-03 | 2014-11-04 | Lsi Corporation | Pre-charge tracking of global read lines in high speed SRAM |
US9030893B2 (en) | 2013-02-06 | 2015-05-12 | Qualcomm Incorporated | Write driver for write assistance in memory device |
US9171634B2 (en) | 2013-03-14 | 2015-10-27 | Arm Limited | Memory device and method of controlling leakage current within such a memory device |
US9053768B2 (en) | 2013-03-14 | 2015-06-09 | Gsi Technology, Inc. | Systems and methods of pipelined output latching involving synchronous memory arrays |
US9083356B1 (en) | 2013-03-14 | 2015-07-14 | Gsi Technology, Inc. | Systems and methods of phase-locked loop involving closed-loop, continuous frequency range, auto calibration and/or other features |
WO2014146027A2 (en) | 2013-03-15 | 2014-09-18 | Gsi Technology, Inc. | Systems and methods involving data bus inversion memory circuitry, configuration and/or operation. |
US9094025B1 (en) | 2013-03-15 | 2015-07-28 | Gsi Technology, Inc. | Systems and methods of phase frequency detection involving features such as improved clock edge handling circuitry/aspects |
US9208853B2 (en) | 2013-03-15 | 2015-12-08 | Intel Corporation | Dual-port static random access memory (SRAM) |
US9318174B1 (en) | 2013-03-15 | 2016-04-19 | Gsi Technology, Inc. | Memory systems and methods involving high speed local address circuitry |
US9413295B1 (en) | 2013-03-15 | 2016-08-09 | Gsi Technology, Inc. | Systems and methods of phase frequency detection with clock edge overriding reset, extending detection range, improvement of cycle slipping and/or other features |
TWI609371B (en) | 2013-03-15 | 2017-12-21 | 積佳半導體股份有限公司 | Dynamic random access memory (dram) device involving data signals grouped into 10 bits and method of operating the same |
TWI509606B (en) | 2013-04-23 | 2015-11-21 | Univ Nat Chiao Tung | Static memory and memory cell thereof |
US9177646B2 (en) | 2013-05-06 | 2015-11-03 | International Business Machines Corporation | Implementing computational memory from content-addressable memory |
US8971096B2 (en) | 2013-07-29 | 2015-03-03 | Qualcomm Incorporated | Wide range multiport bitcell |
US9153305B2 (en) | 2013-08-30 | 2015-10-06 | Micron Technology, Inc. | Independently addressable memory array address spaces |
CN103489477B (en) | 2013-09-04 | 2016-01-13 | 华中科技大学 | Resistant to total dose effect storage unit circuit |
JP6224401B2 (en) | 2013-10-04 | 2017-11-01 | ルネサスエレクトロニクス株式会社 | Semiconductor memory |
US10153042B2 (en) | 2013-11-28 | 2018-12-11 | Gsi Technology Inc. | In-memory computational device with bit line processors |
US9142285B2 (en) | 2013-12-13 | 2015-09-22 | National Chiao Tung University | Multi-port SRAM with shared write bit-line architecture and selective read path for low power operation |
KR102155511B1 (en) | 2013-12-27 | 2020-09-15 | 삼성전자 주식회사 | Semiconductor package and method for fabricating the same |
WO2015102569A2 (en) | 2013-12-30 | 2015-07-09 | The Regents Of The University Of Michigan | Static random access memory cell having improved write margin for use in ultra-low power application |
US9494647B1 (en) | 2013-12-31 | 2016-11-15 | Gsi Technology, Inc. | Systems and methods involving data inversion devices, circuitry, schemes and/or related aspects |
US9576622B2 (en) | 2014-01-24 | 2017-02-21 | Taiwan Semiconductor Manufacturing Company, Ltd. | Reading data from a memory cell |
US9257172B2 (en) | 2014-02-28 | 2016-02-09 | Taiwan Semiconductor Manufacturing Company, Ltd. | Multi-port memory cell |
US9449667B2 (en) | 2014-03-31 | 2016-09-20 | Taiwan Semiconductor Manufacturing Company, Ltd. | Memory circuit having shared word line |
US9305633B2 (en) | 2014-04-17 | 2016-04-05 | Stmicroelectronics International N.V. | SRAM cell and cell layout method |
US20150310901A1 (en) | 2014-04-24 | 2015-10-29 | Qualcomm Incorporated | Memory with a sleep mode |
US9613684B2 (en) | 2014-06-05 | 2017-04-04 | Gsi Technology, Inc. | Systems and methods involving propagating read and write address and data through multi-bank memory circuitry |
TWI675372B (en) | 2014-06-05 | 2019-10-21 | 美商積佳半導體股份有限公司 | Systems and methods involving multi-bank, dual-pipe memory circuitry |
TWI550607B (en) | 2014-07-24 | 2016-09-21 | 智原科技股份有限公司 | Write disturbance mitigation circuit for dual port sram |
US9336864B2 (en) | 2014-08-29 | 2016-05-10 | Qualcomm Incorporated | Silicon germanium read port for a static random access memory register file |
KR101548343B1 (en) | 2014-11-17 | 2015-09-01 | 연세대학교 산학협력단 | Memory device |
US9401200B1 (en) | 2014-12-22 | 2016-07-26 | Altera Corporation | Memory cells with p-type diffusion read-only port |
US9396795B1 (en) | 2014-12-30 | 2016-07-19 | The Regents Of The University Of Michigan | Storage device supporting logical operations, methods and storage medium |
US9786356B2 (en) | 2015-01-30 | 2017-10-10 | Qualcomm Incorporated | Memory device with adaptive voltage scaling based on error information |
US10431269B2 (en) | 2015-02-04 | 2019-10-01 | Altera Corporation | Methods and apparatus for reducing power consumption in memory circuitry by controlling precharge duration |
US9741452B2 (en) | 2015-02-23 | 2017-08-22 | Qualcomm Incorporated | Read-assist circuits for memory bit cells employing a P-type field-effect transistor (PFET) read port(s), and related memory systems and methods |
US20160284392A1 (en) | 2015-03-24 | 2016-09-29 | Globalfoundries Inc. | Memory cell, memory device including a plurality of memory cells and method including read and write operations at a memory cell |
US9558812B2 (en) | 2015-05-05 | 2017-01-31 | Gsi Technology Inc. | SRAM multi-cell operations |
US9853633B1 (en) | 2015-06-22 | 2017-12-26 | Gsi Technology, Inc. | Systems and methods involving lock-loop circuits, clock signal alignment, phase-averaging feedback clock circuitry |
US10659058B1 (en) | 2015-06-26 | 2020-05-19 | Gsi Technology, Inc. | Systems and methods involving lock loop circuits, distributed duty cycle correction loop circuitry |
US9396790B1 (en) | 2015-06-30 | 2016-07-19 | Stmicroelectronics International N.V. | Multi-supply dual port register file |
US9935635B2 (en) | 2015-08-28 | 2018-04-03 | Gsi Technology, Inc. | Systems and methods involving pseudo complementary output buffer circuitry/schemes, power noise reduction and/or other features |
US10249312B2 (en) | 2015-10-08 | 2019-04-02 | Qualcomm Incorporated | Quantization of spatial vectors |
JP6390596B2 (en) | 2015-11-18 | 2018-09-19 | トヨタ自動車株式会社 | Airbag device |
US10049709B2 (en) | 2015-12-31 | 2018-08-14 | Arm Limited | Port modes for use with memory |
US9789840B2 (en) | 2016-03-18 | 2017-10-17 | Ford Global Technologies, Llc | Downwardly inflatable vehicle airbag |
FR3051960B1 (en) | 2016-05-24 | 2019-10-18 | Commissariat A L'energie Atomique Et Aux Energies Alternatives | MEMORY CIRCUIT ADAPTED TO IMPLEMENT CALCULATION OPERATIONS |
US9697890B1 (en) | 2016-06-01 | 2017-07-04 | Taiwan Semiconductor Manufacturing Co., Ltd. | Memory and interface circuit for bit line of memory |
US9685210B1 (en) | 2016-07-08 | 2017-06-20 | Qualcomm Incorporated | Overlapping precharge and data write |
US9640540B1 (en) | 2016-07-19 | 2017-05-02 | Taiwan Semiconductor Manufacturing Co., Ltd. | Structure and method for an SRAM circuit |
KR102421299B1 (en) | 2016-09-12 | 2022-07-18 | 가부시키가이샤 한도오따이 에네루기 켄큐쇼 | Memory device, driving method thereof, semiconductor device, electronic component, and electronic device |
US10242728B2 (en) | 2016-10-27 | 2019-03-26 | Samsung Electronics Co., Ltd. | DPU architecture |
US9916889B1 (en) | 2016-12-01 | 2018-03-13 | Intel Corporation | Memory circuitry with row-wise gating capabilities |
US10891076B1 (en) | 2016-12-06 | 2021-01-12 | Gsi Technology, Inc. | Results processing circuits and methods associated with computational memory cells |
US10998040B2 (en) | 2016-12-06 | 2021-05-04 | Gsi Technology, Inc. | Computational memory cell and processing array device using the memory cells for XOR and XNOR computations |
US10770133B1 (en) | 2016-12-06 | 2020-09-08 | Gsi Technology, Inc. | Read and write data processing circuits and methods associated with computational memory cells that provides write inhibits and read bit line pre-charge inhibits |
US10725777B2 (en) | 2016-12-06 | 2020-07-28 | Gsi Technology, Inc. | Computational memory cell and processing array device using memory cells |
US10847212B1 (en) | 2016-12-06 | 2020-11-24 | Gsi Technology, Inc. | Read and write data processing circuits and methods associated with computational memory cells using two read multiplexers |
US10860320B1 (en) | 2016-12-06 | 2020-12-08 | Gsi Technology, Inc. | Orthogonal data transposition system and method during data transfers to/from a processing array |
US10854284B1 (en) | 2016-12-06 | 2020-12-01 | Gsi Technology, Inc. | Computational memory cell and processing array device with ratioless write port |
US10943648B1 (en) | 2016-12-06 | 2021-03-09 | Gsi Technology, Inc. | Ultra low VDD memory cell with ratioless write port |
US10534836B2 (en) | 2016-12-06 | 2020-01-14 | Gsi Technology Inc. | Four steps associative full adder |
US10777262B1 (en) | 2016-12-06 | 2020-09-15 | Gsi Technology, Inc. | Read data processing circuits and methods associated memory cells |
US10847213B1 (en) | 2016-12-06 | 2020-11-24 | Gsi Technology, Inc. | Write data processing circuits and methods associated with computational memory cells |
FR3088767B1 (en) | 2018-11-16 | 2022-03-04 | Commissariat Energie Atomique | MEMORY CIRCUIT SUITABLE FOR IMPLEMENTING CALCULATION OPERATIONS |
US10930341B1 (en) | 2019-06-18 | 2021-02-23 | Gsi Technology, Inc. | Processing array device that performs one cycle full adder operation and bit line read/write logic features |
US10958272B2 (en) | 2019-06-18 | 2021-03-23 | Gsi Technology, Inc. | Computational memory cell and processing array device using complementary exclusive or memory cells |
US10877731B1 (en) | 2019-06-18 | 2020-12-29 | Gsi Technology, Inc. | Processing array device that performs one cycle full adder operation and bit line read/write logic features |
US10673440B1 (en) | 2019-08-13 | 2020-06-02 | Xilinx, Inc. | Unified programmable computational memory and configuration network |
-
2018
- 2018-10-04 US US16/152,374 patent/US10891076B1/en active Active
-
2019
- 2019-12-13 US US16/713,383 patent/US11194519B2/en active Active
-
2020
- 2020-10-27 US US17/081,869 patent/US20210216246A1/en not_active Abandoned
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US11194519B2 (en) | 2016-12-06 | 2021-12-07 | Gsi Technology, Inc. | Results processing circuits and methods associated with computational memory cells |
US11205476B1 (en) | 2016-12-06 | 2021-12-21 | Gsi Technology, Inc. | Read data processing circuits and methods associated with computational memory cells |
US11409528B2 (en) | 2016-12-06 | 2022-08-09 | Gsi Technology, Inc. | Orthogonal data transposition system and method during data transfers to/from a processing array |
US11763881B2 (en) | 2016-12-06 | 2023-09-19 | Gsi Technology, Inc. | Computational memory cell and processing array device using the memory cells for XOR and XNOR computations |
US11194548B2 (en) | 2019-06-18 | 2021-12-07 | Gsi Technology, Inc. | Processing array device that performs one cycle full adder operation and bit line read/write logic features |
Also Published As
Publication number | Publication date |
---|---|
US10891076B1 (en) | 2021-01-12 |
US20200117398A1 (en) | 2020-04-16 |
US11194519B2 (en) | 2021-12-07 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US11194519B2 (en) | Results processing circuits and methods associated with computational memory cells | |
US11205476B1 (en) | Read data processing circuits and methods associated with computational memory cells | |
US10770133B1 (en) | Read and write data processing circuits and methods associated with computational memory cells that provides write inhibits and read bit line pre-charge inhibits | |
US10847212B1 (en) | Read and write data processing circuits and methods associated with computational memory cells using two read multiplexers | |
US11257540B2 (en) | Write data processing methods associated with computational memory cells | |
US11763881B2 (en) | Computational memory cell and processing array device using the memory cells for XOR and XNOR computations | |
US11150903B2 (en) | Computational memory cell and processing array device using memory cells | |
Agrawal et al. | X-SRAM: Enabling in-memory Boolean computations in CMOS static random access memories | |
US11194548B2 (en) | Processing array device that performs one cycle full adder operation and bit line read/write logic features | |
US10958272B2 (en) | Computational memory cell and processing array device using complementary exclusive or memory cells | |
US10930341B1 (en) | Processing array device that performs one cycle full adder operation and bit line read/write logic features | |
US10854284B1 (en) | Computational memory cell and processing array device with ratioless write port | |
JP4932856B2 (en) | Apparatus and method for adjusting operating parameters of an integrated circuit | |
US9558838B2 (en) | Semiconductor device for masking data stored in twin cell and outputting masked data | |
US9311971B1 (en) | Systems and methods of semiconductor memory devices including features of output buffer initialization circuit(s) and/or multiple power-up detection/handling | |
US7170805B2 (en) | Memory devices having bit line precharge circuits with off current precharge control and associated bit line precharge methods | |
US11227653B1 (en) | Storage array circuits and methods for computational memory cells | |
CN114077417A (en) | In-memory operation method and device, memory and storage medium | |
US7755949B2 (en) | Reset circuit for termination of tracking circuits in self timed compiler memories | |
US20230170012A1 (en) | In-memory compute sram with integrated toggle/copy operation and reconfigurable logic operations |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
AS | Assignment |
Owner name: GSI TECHNOLOGY, INC., CALIFORNIA Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNORS:HAIG, BOB;EHRMAN, ELI;ILAN, DAN;AND OTHERS;SIGNING DATES FROM 20181017 TO 20181207;REEL/FRAME:055130/0662 |
|
STPP | Information on status: patent application and granting procedure in general |
Free format text: DOCKETED NEW CASE - READY FOR EXAMINATION |
|
STPP | Information on status: patent application and granting procedure in general |
Free format text: NON FINAL ACTION MAILED |
|
STCB | Information on status: application discontinuation |
Free format text: ABANDONED -- FAILURE TO RESPOND TO AN OFFICE ACTION |