WO2015045135A1 - プログラマブルロジックデバイス、及び、論理集積ツール - Google Patents
プログラマブルロジックデバイス、及び、論理集積ツール Download PDFInfo
- Publication number
- WO2015045135A1 WO2015045135A1 PCT/JP2013/076467 JP2013076467W WO2015045135A1 WO 2015045135 A1 WO2015045135 A1 WO 2015045135A1 JP 2013076467 W JP2013076467 W JP 2013076467W WO 2015045135 A1 WO2015045135 A1 WO 2015045135A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- programmable logic
- cram
- logic device
- reliability
- logic
- Prior art date
Links
Images
Classifications
-
- H—ELECTRICITY
- H03—ELECTRONIC CIRCUITRY
- H03K—PULSE TECHNIQUE
- H03K19/00—Logic circuits, i.e. having at least two inputs acting on one output; Inverting circuits
- H03K19/02—Logic circuits, i.e. having at least two inputs acting on one output; Inverting circuits using specified components
- H03K19/173—Logic circuits, i.e. having at least two inputs acting on one output; Inverting circuits using specified components using elementary logic circuits as components
- H03K19/177—Logic circuits, i.e. having at least two inputs acting on one output; Inverting circuits using specified components using elementary logic circuits as components arranged in matrix form
- H03K19/17748—Structural details of configuration resources
- H03K19/17764—Structural details of configuration resources for reliability
-
- H—ELECTRICITY
- H03—ELECTRONIC CIRCUITRY
- H03K—PULSE TECHNIQUE
- H03K19/00—Logic circuits, i.e. having at least two inputs acting on one output; Inverting circuits
- H03K19/02—Logic circuits, i.e. having at least two inputs acting on one output; Inverting circuits using specified components
- H03K19/173—Logic circuits, i.e. having at least two inputs acting on one output; Inverting circuits using specified components using elementary logic circuits as components
- H03K19/177—Logic circuits, i.e. having at least two inputs acting on one output; Inverting circuits using specified components using elementary logic circuits as components arranged in matrix form
- H03K19/17724—Structural details of logic blocks
- H03K19/17728—Reconfigurable logic blocks, e.g. lookup tables
-
- H—ELECTRICITY
- H03—ELECTRONIC CIRCUITRY
- H03K—PULSE TECHNIQUE
- H03K19/00—Logic circuits, i.e. having at least two inputs acting on one output; Inverting circuits
- H03K19/02—Logic circuits, i.e. having at least two inputs acting on one output; Inverting circuits using specified components
- H03K19/173—Logic circuits, i.e. having at least two inputs acting on one output; Inverting circuits using specified components using elementary logic circuits as components
- H03K19/177—Logic circuits, i.e. having at least two inputs acting on one output; Inverting circuits using specified components using elementary logic circuits as components arranged in matrix form
- H03K19/17748—Structural details of configuration resources
- H03K19/1776—Structural details of configuration resources for memories
-
- H—ELECTRICITY
- H03—ELECTRONIC CIRCUITRY
- H03K—PULSE TECHNIQUE
- H03K19/00—Logic circuits, i.e. having at least two inputs acting on one output; Inverting circuits
- H03K19/20—Logic circuits, i.e. having at least two inputs acting on one output; Inverting circuits characterised by logic function, e.g. AND, OR, NOR, NOT circuits
- H03K19/21—EXCLUSIVE-OR circuits, i.e. giving output if input signal exists at only one input; COINCIDENCE circuits, i.e. giving output only if all input signals are identical
Definitions
- the present invention relates to a programmable logic device and a logic integration tool.
- Patent Document 1 discloses a technique for realizing a highly reliable and safe system by self-checking a comparator.
- PLD Programmable Logic Device
- FPGA Field Programmable Gate Array
- This programmable logic device is characterized in that logic information can be held in a rewritable memory (configuration memory: CRAM), and the logic can be set afterwards on the user side.
- CRAM configuration memory
- highly-integrated FPGAs mainly have a structure in which this memory is held by SRAM (Static Random Access Memory).
- SRAM Static Random Access Memory
- an object of the present invention is to provide a highly reliable and safe programmable logic device having high error resistance.
- the present invention provides a programmable logic device having a plurality of configuration memories, wherein the configuration memories are divided into a plurality of areas, and a part of the plurality of areas has a reliability against a failure of the configuration memory. It is characterized by being set to a highly reliable area higher than other areas.
- a highly reliable and safe programmable logic device with high error tolerance can be provided.
- FIG. 1 is a structural diagram of an FPGA according to Example 1.
- FIG. FIG. 2 shows a connection diagram of the CRAM and ROM of the FPGA according to the first embodiment.
- FIG. 2 is a layout diagram of the FPGA CRAM according to the first embodiment.
- FIG. 6 shows an operation waveform diagram of CRAM control of the FPGA according to the first embodiment.
- FIG. 3 shows a flowchart of logic separation integrated in the FPGA according to the first embodiment.
- FIG. 3 shows a structural diagram of an FPGA according to Example 2.
- FIG. 6 shows a layout diagram of logic elements of an FPGA according to a second embodiment.
- 7 shows a chip cross-sectional structure of a boundary portion between a highly reliable mounting portion and a large-capacity logic mounting portion of an FPGA according to a second embodiment.
- FIG. 6 shows a structural diagram of an FPGA according to Example 3.
- FIG. 6 shows a structural diagram of an FPGA comparator according to a third embodiment.
- FIG. 10 shows an operation waveform diagram of CRAM control of the FPGA according to the third embodiment.
- FIG. 6 shows a structural diagram of an FPGA according to Example 4;
- FIG. 10 is an operation waveform diagram of CRAM control of the FPGA according to the fourth embodiment.
- FIG. 9 shows a structural diagram of an FPGA according to Example 5.
- FIG. 9 shows a structural diagram of an FPGA according to Example 6.
- Logic LSIs include circuits such as flip-flops and latches as memory holding elements, but these circuits are known to be larger in size than SRAM memory cells and more resistant to soft errors than SRAMs.
- SRAM memory integrated in the logic LSI is added with parity and ECC (Error Correction Code), and can be systematically detected and corrected for SRAM bit inversion and the like.
- ECC Error Correction Code
- the CRAM has a form in which the data is retained after the data is set, is difficult to monitor by ECC, and is weak against bit inversion of the memory cell.
- the inventors have a part in which the reliability of the hardware alone is important as a requirement for the LSI, and a part in which the reliability is not so important in terms of hardware if the reliability is ensured in the system. I realized that.
- the portion requiring high reliability is, for example, a register for storing key information, a comparator for comparing multiplexed calculation results, and the like.
- reliability can be improved by redundancy by utilizing multiplexing and encoding, high reliability as a device is not necessarily required.
- the programmable logic device has a plurality of configuration memories, and the configuration memories are divided into a plurality of regions, and a part of the plurality of regions is the configuration memory.
- the reliability is set in a high reliability region in which the reliability against a memory failure is higher than in other regions.
- a calculation unit that performs calculation processing is provided in the other region, and a calculation result processing unit that processes a calculation result calculated by the calculation unit is arranged in the high reliability region.
- an SRAM-based FPGA will be described as an example of a programmable logic device.
- the contents of each embodiment are not limited to this, but to a programmable LSI or semiconductor that allows a user to program logic later. Is also applicable.
- the FPGA of this example is shown in FIG.
- This FPGA is configured by accumulating a plurality of logic elements in a tile shape.
- This FPGA is divided into at least two areas.
- One of the two areas corresponds to the above-described high-reliability area, and becomes the high-reliability mounting unit 2 that is a part that is more reliable against a soft error.
- the other corresponds to the above-described other area, and becomes the large-capacity logic implementation unit 1 which is a part for accumulating arithmetic logic for realizing the function of the apparatus.
- the high-capacity logic mounting unit 1 is a part of which reliability is lower than that of the high-reliability mounting unit 2, but in this embodiment, emphasis is placed on mounting logic at a higher density. Yes.
- the high-reliability area is set such that the CRAM check frequency is higher than other areas.
- the CRAM integrated in each area is checked independently, and for each CRAM, it is checked whether the data stored in the CRAM is destroyed.
- CRAM control units CRAMCTL1, 2 are provided.
- the CRAM check a CRC check is generally known that has a small hardware scale and can be detected at high speed, and this technique can also be used in this embodiment. The CRAM check method will be described in detail below.
- the reliability of a general logic circuit can be improved by a conventionally known operation such as duplication or duplication or an operation with an error correction code added.
- a part that holds settings as an LSI, a comparison circuit of calculation results, and the like can greatly contribute to improvement of availability and safety of the entire device by increasing the reliability of the circuit itself. Therefore, the reliability and safety of the entire apparatus can be improved by increasing the reliability of some limited circuits among the circuits integrated in the FPGA in terms of devices and circuits.
- an SRAM-based FPGA it is important to improve the soft error resistance of the SRAM.
- soft error resistance of SRAM it is known to increase the size of memory cells and increase the power supply voltage.
- general-purpose usage has been prioritized, so it has been important to have a homogeneous configuration in the FPGA.
- an architecture that can implement high-density (or large-capacity) logic has been regarded as important in FPGAs that are inferior in terms of mounting rate when compared with the same process and the same chip size.
- the reliability in the LSI it has been found that the reliability can be greatly improved while ensuring the versatility even in the heterogeneous configuration instead of the homogeneous configuration.
- FIG. 2 shows an example of connection between the CRAM in the FPGA of this embodiment and the nonvolatile ROM provided outside the FPGA.
- the FPGA of this embodiment can be configured to be connected to a nonvolatile ROM such as Flash via a memory bus MEMBUS that performs data transfer for CRAM access.
- a nonvolatile ROM such as Flash
- MEMBUS memory bus
- FIG. 3 shows the physical layout of the CRAM in the FPGA.
- an FPGA generally, memory cell columns constituting a CRAM and logic circuit portions such as switches are alternately arranged, and a plurality of CRAM memory cell columns are arranged in the vertical direction of the chip.
- this memory arrangement is physically divided and divided into a large-capacity logic mounting unit 1 and a high-reliability mounting unit 2 and integrated on an LSI chip. That is, the large-capacity logic mounting unit 1 and the high-reliability mounting unit 2 are provided with a plurality of cell rows in which CRAMs are arranged in series.
- FIG. 1 In the example of FIG.
- CRAM # 0 to 15 16 CRAM memory cell columns (CRAM # 0 to 15) are provided in the large-capacity logic mounting unit 1, and three highly reliable CRAM cell columns are included in the high-reliability mounting unit 2.
- RCRAM # 0 to 2 An example in which (RCRAM # 0 to 2) are integrated will be described.
- a highly reliable CRAM is indicated as RCRAM (Reliable CRAM).
- this figure shows an example in which the large-capacity logic mounting unit 1 is integrated on the left side of the page and the high-reliability mounting unit 2 is integrated on the right side. It doesn't matter. It is also possible to arrange them separately on the top and bottom of the chip.
- the CRAM control unit CRAMCTL1 corresponding to the large-capacity logic implementation unit 1 sequentially checks the stored data for CRAM # 0 to CRAM # 15. Further, the CRAM control unit CRAMCTL2 corresponding to the high-reliability mounting unit 2 sequentially checks RCRAM # 0 to RCRAM # 2.
- FIG. 4 shows a timing chart of CRAM access. An example in which 16 CRAM cell columns are checked for the large-capacity logic mounting unit 1 and 3 CRAM memory cell columns are checked for the high-reliability mounting unit 2 will be described.
- the CRAM control unit CRAMCTL1 performs a check while sequentially synchronizing CRAM # 0 to CRAM # 15 with the operation clock of the CRAM control unit CRAMCTL1. After checking up to CRAM # 15, return to CRAM # 0 again and check.
- the inspection cycle of the CRAM control unit CRAMCTL1 is T1.
- the high-reliability mounting unit 2 will be described in the case where three CRAM memory cell columns are checked in this figure.
- RCRAM # 0, RCRAM # 1, and RCRAM # 2 are sequentially checked, and when the RCRAM # 2 check is completed, the process returns to RCRAM # 0 again to perform the check.
- the check cycle at this time is T2.
- the check cycle of the high-reliability mounting unit 2 can be fixed without depending on the integrated capacity of the large-capacity logic mounting unit 1, and the CRAM of the high-reliability mounting unit 2 can be checked and repaired. By being able to carry out promptly, it becomes possible to lead to high reliability and high safety of the LSI as a whole.
- the high-capacity logical mounting unit 1 and the high-reliability mounting are provided.
- the CRAM of the part 2 can be constituted by transistors manufactured based on the same design process and design rule, and the memory cell size can be equally configured.
- a method of making the CRAM of the high-reliability mounting unit 2 larger than that of the large-capacity logic mounting unit 1 as in Example 2 described later may be further employed.
- FIG. 5 is a diagram for explaining a design flow of logic design utilizing the FPGA of this embodiment.
- the design of the functional logic installed in the FPGA is performed using a description language such as RTL (Register Transfer Level) as in the prior art.
- RTL Registered Transfer Level
- the degree of reliability is defined for the designed logic, and for example, an identification tag indicating the degree of reliability is added to the designed RTL.
- the logic data accumulated in the FPGA includes an identification tag indicating whether the target area where the logic is accumulated is the high-reliability mounting unit 2 or the large-capacity logic mounting unit 1.
- logic integration is performed using a logic integration tool.
- the logic integration tool performs logic integration on the large-capacity logic implementation unit 1 and the high-reliability implementation unit 2 based on the identification tag.
- the high-reliability mounting unit 2 requires less logic than the logic mounted in the large-capacity logic mounting unit 1 among the logic mounted in the LSI.
- a design with versatility is preferable and high It may be desirable to secure and define a certain amount of the capacity of the reliable implementation unit 2. In that case, even if the logic does not need to be mounted in the high-reliability mounting unit 2, if the logic integration is performed in the free area of the large-capacity logic mounting unit 1, LSI hardware resources can be used effectively.
- FIG. 6 shows the FPGA of this example.
- a large-capacity logic mounting unit 1 and a high-reliability mounting unit 2 are provided.
- the high-reliability mounting unit 2 uses a CRAM having a relatively large transistor size, and further includes a gate insulating film of a MISFET to be configured. By making the thickness large, it can withstand the application of a high voltage. As a result, the power supply voltage of the CRAM of the high-reliability mounting unit 2 is set higher than the CRAM in other areas.
- a signal level conversion circuit LS is added.
- the CRAM of the high-reliability mounting unit 2 can be designed to have a larger amount of retained charge than the CRAM of the large-capacity logic mounting unit 1, radiation such as neutrons collides with other nuclei. It is possible to have a high resistance to charge accumulation caused by the intrusion of charged particles.
- a capacitor can be added. What is necessary is just to comprise a capacity
- the high-reliability mounting unit 2 is composed of a relatively large transistor.
- the CRAM single unit soft error resistance of the high-reliability mounting unit 2 is improved by, for example, about one digit compared to the single-layer soft error resistance of the large-capacity logic mounting unit 1 CRAM, the CRAM check Even if the mechanism uses a method in which each CRAM memory cell column is sequentially checked cyclically by one CRAM control unit CRAMCTL3 in the chip, the MTTF (Mean Time To Failure) of the CRAM memory cell of the high-reliability mounting unit 2 is a large-capacity logic mounting unit.
- the high-reliability mounting unit 2 and the large-capacity logic mounting unit 1 may be provided with a CRAM control unit independently.
- FIG. 7 shows an example of integration of logic elements in the LSI for realizing the present embodiment.
- a transistor designed with a process node of 22 nm is arranged in the large-capacity logic mounting section 1, and a transistor having a size about twice as large as that of the high-reliability mounting section 2 (2 as a design rule).
- the lookup table LUT converts an input value from the lookup table LUT into an output value according to the bit information stored in the CRAM.
- the output of the lookup table LUT is taken into the flip-flop FF and connected to the next lookup table LUT or the like.
- the selector SEL is a circuit for switching the connection between the wiring and the lookup table LUT, and this is also switched by a bit stored in the CRAM. The connection switching between the wirings is switched by the switch SW.
- the signal level conversion circuit LS is provided between the subsequent stage of the selector that switches the connection between the lookup table LUT and the wiring, and the switch of the large-capacity logic mounting unit 1 and the switch of the high-reliability mounting unit 2. Once the signal level is converted, the subsequent high-reliability mounting unit 2 does not require a signal level conversion circuit.
- the transistor In order to further improve the resistance to soft errors against neutrons, it is desirable to design with a transistor used for I / O. In that case, the transistor should be designed with a transistor equivalent to 130 nm or 180 nm, which is 4 to 8 times larger.
- the cell will be configured.
- the high-reliability mounting unit 2 may be arranged in a tile shape as shown in FIG. Alternatively, it is possible to form a long cell in the horizontal direction in this figure. In any case, even if the transistor sizes are different, they can be arranged in a tile shape by an appropriate design, and a highly versatile circuit as an FPGA can be designed.
- FIG. 8 shows an example of a transistor structure constituting this embodiment.
- a P-type MISFET and an N-type MISFET are shown for the large-capacity logic mounting unit 1 and the high-reliability mounting unit 2, respectively.
- This figure shows an example of a so-called triple well structure in which an N type well is formed on a P type substrate.
- the power supply voltage VCC of the high reliability mounting unit 2 is higher than the power supply voltage VDD of the large-capacity logic mounting unit 1, it is necessary to separate the well on the NMOS side.
- the transistor structure is not limited to the triple well structure, and can be applied to a so-called dual well LSI in which no deep N well exists. In that case, although separation on the ground side may be difficult, a plurality of power supply voltages can be set.
- a CRAM manufactured by a so-called thick wire diameter process with an old process generation can be used.
- MOSs constituting an I / O need to be connected to peripheral devices of legacy voltage, so even in miniaturized LSIs, transistors with thick wire diameters are integrated and used on the same chip. Therefore, this may be used as a transistor of this thick wire diameter process.
- FIG. 9 shows an example of integration of the large-capacity logic mounting unit 1 and the high-reliability mounting unit 2 of the FPGA of this embodiment.
- a plurality of comparators are provided in a multiplexed manner.
- the large-capacity logic mounting unit 1 and the high-reliability mounting unit 2 are provided in the FPGA.
- the functions are realized by the duplicated function block FBA and function block FBB.
- the in each of the function blocks FBA and FBB CRAMs for storing logic information and wiring switching information are respectively defined. In this figure, not a physical arrangement but a logical relationship is shown so that the corresponding CRAM is clarified according to the function.
- the CRAM control unit 1CRAMCTL1 writes data from the ROM and checks the stored information in the CRAM.
- the high-reliability mounting unit 2 includes comparators CMPA and CMPB that compare signals from the functional blocks FBA and FBB, an output inspection unit ORA (Output Response Analyzer) mainly used for fail-safe applications, and functional blocks
- a test pattern generation circuit TPG for self-test for diagnosing a failure of the FBA, FBB, comparators CMPA, CMPB, a selector SEL for outputting one of the peripheral logic Plogic for switching wiring, the functional blocks FBA, FBB,
- the output control circuit OUTCTL is provided with a function of stopping the output OUT.
- comparators CMPA and CMPB in order to improve availability, an example of a dual system is shown, and an ORA control unit ORACTL for controlling the output inspection unit ORA is provided.
- EOR in the peripheral logic logic indicates an exclusive OR circuit.
- TPA and TPB indicate a test pattern A and a test pattern B, respectively.
- mapping When configuring the above-described dual system in this embodiment, it is preferable to perform mapping so that a logical operation related to the other RCRAM is not impossible during reconfiguration of one RCRAM. Therefore, as shown in FIG. 3, when RCRAM memory cells are arranged in stripes in the vertical direction, the RCRAM storing the A-system logic and the RCRAM storing the B-system logic are moved to different cell columns. Deploy.
- the CRAM is physically separated by disposing the comparators CMPA and CMPB in different cell columns of the plurality of cell columns, so that the read and write operations to the CRAM are not performed at the same timing.
- the comparators CMPA and CMPB have a circuit configuration in which an exclusive OR circuit is used in the first stage, and ORs are arranged in a tree form in the subsequent stage, thereby sending out from the functional blocks FBA and FBB. Inconsistency of detected signals can be detected and notified to the outside with one output signal.
- inspection circuits ORAA and ORAB are provided as circuit configurations that can inspect the outputs of the duplicated comparators CMPA and CMPB, and the outputs TOUT1 and TOUT2 are also supplied to the CTL. Entered.
- the output inspection unit ORA in order to output either the output TOUT1 of the inspection circuit ORAA or the output TOUT2 of the inspection circuit ORAB to the outside of the LSI, the output is switched by the selector SEL.
- the selector SEL is switched by a control signal SELC from the CTL.
- the ORA control unit ORACTL observes the alternating signal which is the output of the output TOUT1 of the inspection circuit ORAA and the output TOUT2 of the inspection circuit ORAB, and checks the stop of the alternating signal.
- the alternating signal means a signal in which a high level and a low level of the output appear alternately.
- a test pattern for detecting a failure in the comparators CMPA and CMPB is applied to one of the input signals at a level opposite to that of the signal (that is, the malfunction signal is arbitrarily transmitted to the comparators CMPA and CMPB).
- the failure of the comparators CMPA and CMPB is detected.
- the comparators CMPA and CMPB are checked simultaneously with the rising edge of the clock, and the data is compared at the falling edge of the clock.
- the former is called a comparator check period
- the latter is called a compared signal comparison period.
- the SEL selector switches so that one of the output TOUT1 of the inspection circuit ORAA and the output TOUT2 of the inspection circuit ORAB is output as the output TOUT.
- the control is performed by the error signal ERR from the CRAM control unit CRAMCTL2.
- the error signal ERR is not shown in detail here, if it is designed with a 2-bit signal, it is possible to indicate which of the CMPA and CMP2 has failed, and accordingly, the selector SEL can be controlled by the SELC. Is possible.
- Corresponding CRAM is associated with various logic circuits integrated in the high-reliability mounting unit 2. These RCRAMs are controlled by the CRAM control unit CRAMCTL2.
- the CRAM control units CRAMCTL1 and CRAMCTL2 are connected to the same memory bus MEMBUS, and configuration data for each CRAM is stored in the ROM outside the FPGA or the storage circuit RAM1 in the FPGA. ROM and RAM are also connected to the same memory bus.
- the RAM 1 is configured as a memory provided with an error detection and correction mechanism such as ECC.
- ROM 1 and RAM 1 are provided. It is assumed that the ROM outside the FPGA is constituted by a nonvolatile ROM such as Flash, and although the neutron resistance is high, the access speed of data is slow, such as so-called fast access takes several tens of microseconds.
- the RAM 1 in the FPGA is basically an SRAM-based memory, and has a feature that it can be accessed at high speed although it has low neutron resistance. However, since the circuit can be protected by ECC or the like, the neutron resistance can be increased. Further, this built-in RAM is integrated in the high-reliability mounting unit 2 and may be composed of a transistor having a high soft error tolerance like a highly reliable CRAM, or as a ROM capable of high-speed operation. It may be configured.
- the basic operation of the high-reliability comparator may be performed as described in Patent Document 1, but here, it is duplexed in order to improve the availability of the comparator itself.
- FIG. 11 shows an operation waveform diagram of this example.
- RCRAM # i is inspected at time t1 and there is a soft error in CRAM.
- an error occurs in the comparator CMPA, and the normal output from the comparator CMPA is not output in the output inspection unit ORA, so that the alternating signal that is the output signal of the inspection circuit ORAA is interrupted.
- error detection can be performed in one cycle from the CRAM check, it may take several cycles.
- the comparator CMPB since the comparator CMPB is normal, the outputs of the function blocks FBA and FBB are the same, so it is not necessary to stop the output.
- the reconfiguration signal RECONFIG which rewrites the CRAM is asserted, and the normal configuration data from the RAM 1 is written to the RCRAM.
- rewriting from the RAM 1 will be described as enhancing the resistance to multi-bit errors.
- 1-bit error correction can be performed, so that 1-bit correction can be restored more quickly.
- the address is calculated in one cycle from the error signal ERR, the address is transmitted to the RAM 1 at time t2, and the data is transmitted in the next cycle (time t3). Is assumed to be received.
- the data arrival latency may be several cycles depending on the specifications of the LSI to be designed, but here it is assumed to be one cycle.
- the data is written into RCRAM # i, and upon completion of writing at time t4, the reconfiguration signal RECONFIG is negated, and then the error signal ERR is negated.
- the test circuit ORAB may remain selected. In this case, the duplex system of the comparator is restored at time t4, but the complete duplex system can be restored at time t5 when the alternating signal of the test circuit ORAA returns to normal.
- the ROM, the output inspection unit ORA, and the test pattern generation circuit TPG that are in the FPGA in the configuration of the third embodiment are provided outside the FPGA.
- an error signal ERR2 is transmitted to the control unit CTL2.
- the control circuit CTL2 inputs the alternating signals TOUT11 and TOUT12 from the ORA together with the ERR2 signal, sends the control signal SCTL2 to the control circuit OUTCTL, controls the output stop of the output OUT, and controls the ORA circuit with the control signal SELC. It has a function to perform switching control of the alternating signal output selector.
- the alternating signals TOUT11 and TOUT12 output from the comparison circuits CMPA and CMPB are inspected, and when the waveforms are abnormal, that is, a comparator mismatch is detected during the comparison signal comparison period. In such a case, control is performed to stop the output.
- the OUTCTL continues to output.
- control for stopping the output OUT is performed for safety. In this case, since CRAM CTL2 can determine which comparator is broken, the comparator on the broken side is repaired.
- This embodiment has an advantage that the number of circuits integrated in the FPGA can be reduced.
- the output inspection unit ORA and the like are separately configured by a dedicated LSI or a general-purpose LSI.
- ORA, TPG, and the like can be configured by hard logic, there is an effect that resistance to soft errors is further increased.
- the ROM may be provided in the PGA.
- FIG. 13 shows an operation example of this embodiment.
- the basic operation is the same as described with reference to FIG.
- an external ROM access is used, if a general flash ROM is used, several tens of microseconds are required for the first access, and data reception starts from the address transmission (time t2 ′) (time t3 ′).
- a waiting time on the order of several tens of microseconds occurs. This waiting time is different from FIG.
- FIG. 13 shows an example in which the data transfer amount is large.
- the data transfer waiting time time t3 ′ to time t3 ′′
- the configuration time time t4 ′ to time t5 ′
- the highly reliable mounting unit 2 can reduce data transfer and configuration overhead by defining circuit mounting in a relatively small area.
- the comparator integrated in the high-reliability mounting unit 2 is tripled. Also in the present embodiment, a comparator or the like is integrated in the high reliability mounting unit 2.
- an error signal ERR3 is transmitted to the control unit CTL3.
- the control circuit CTL3 has a function of executing the output stop control of the output OUT by inputting the ERR3 signal and the alternating signal TOUT3 [0: 2] from the ORA and sending the signal SCTL3 to the control circuit OUTCTL.
- the SELC 3 has a function of performing switching control of the alternating signal output selector in the ORA circuit.
- the three alternating signals TOUT3 [0: 2] output from the comparison circuits CMPA, CMPB, and CMPC together with the error signal ERR3 are inspected, and when the waveform is abnormal, that is, the signal When a mismatch of comparators is detected during the comparison period, control is performed to stop the output.
- the OUTCTL performs control to continue outputting.
- the defective comparator immediately repairs.
- control for stopping the output OUT is performed for safety. Since the majority can be implemented by being triple, even if one of the comparators breaks down, the broken part can be repaired and the arithmetic processing can be continued.
- the logic element unit is not distinguished between the large-capacity logic mounting unit 1 and the high-reliability mounting unit 2.
- the example in which the large-capacity logic mounting unit 1 and the high-reliability mounting unit 2 are physically separated has been described.
- the embodiment shown in FIG. 15 corresponds to the CRAM check circuit corresponding to the high-reliability mounting unit 2 and the large-capacity logic mounting unit 1 in order to increase the inspection frequency of the CRAM soft error in the high-reliability mounting unit 2.
- the CRAM check circuit is made independent, the large-capacity logic mounting unit 1 and the high-reliability mounting unit 2 are made programmable.
- the same type of memory cell circuit and element are used for the CRAM of the large-capacity logic mounting unit 1 and the CRAM of the high-reliability mounting unit 2.
- the division between the CRAM of the large-capacity logic mounting unit 1 and the CRAM of the high-reliability mounting unit 2 may be determined by an address.
- the CRAM addresses A0 to AN correspond to the large-capacity logic mounting unit 1
- the CRAM control unit CRMTCL1 controls the CRAM
- the CRAM addresses AN + 1 to AM are set to high.
- the CRAM control unit CRMTCL2 controls the CRAM corresponding to the trust implementation unit 2.
- the logic of the large-capacity logic mounting unit 1 and the high-reliability mounting unit 2 and the transistors constituting the CRAM are manufactured according to the same process rule.
- the bit capacity size of the CRAM of the high-reliability mounting unit 2 The CRAM check frequency may be increased by setting the CRAM bit capacity size of the mounting unit 1 to 1/10 to 1/100.
- the operating frequency of the CRAM control unit CRAMCTL2 is made higher than the operating frequency of the CRAM control unit CRAMCTL1
- the CRAM check frequency of the highly reliable mounting unit 2 is relatively higher than the CRAM check frequency of the large-capacity logic mounting unit 1 Can do.
- the mounting logic scale that is, the capacity of the CRAM
- the high-reliability mounting unit 2 can be made variable, so that it can be used more universally.
- SYMBOLS 1 Large-capacity logic mounting part, 2 ... High reliability mounting part, CRAMCTL ... CRAM control part, LUT ... Look-up table SEL ... Selector, SW ... Switch, FF ... Flip-flop, LS ... Signal level conversion circuit, FBA, FBB ... Functional block, CMPA, CMPB ... comparator, TOUT ... output, ORA ... output inspection unit, TPG ... test pattern generation circuit, CRAMCTL ... CRAM control unit, MEMBUS ... memory bus CRAM access data transfer bus
Landscapes
- Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Engineering & Computer Science (AREA)
- Computer Hardware Design (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Design And Manufacture Of Integrated Circuits (AREA)
- Logic Circuits (AREA)
- For Increasing The Reliability Of Semiconductor Memories (AREA)
Abstract
Description
Claims (8)
- 複数のコンフィギュレーションメモリを有するプログラマブルロジックデバイスにおいて、
前記コンフィギュレーションメモリが複数の領域に区分して配置され、
前記複数の領域の一部は、前記コンフィギュレーションメモリの故障に対する信頼性を他の領域よりも高めた高信頼領域に設定されることを特徴とするプログラマブルロジックデバイス。 - 請求項1に記載のプログラマブルロジックデバイスにおいて、
前記高信頼領域は、前記コンフィギュレーションメモリのチェック頻度が他の領域よりも高く設定されることを特徴とするプログラマブルロジックデバイス。 - 請求項1に記載のプログラマブルロジックデバイスにおいて、
前記高信頼領域のコンフィギュレーションメモリは、他の領域のコンフィギュレーションメモリに比べて、トランジスタのゲート酸化膜の厚さ及びセルサイズが大きく設定されることを特徴とするプログラマブルロジックデバイス。 - 請求項1に記載のプログラマブルロジックデバイスにおいて、
前記他の領域には、演算処理を行う演算部が設けられ、
前記高信頼領域には、前記演算部で演算された演算結果を処理する演算結果処理部が配置されることを特徴とするプログラマブルロジックデバイス。 - 請求項1に記載のプログラマブルロジックデバイスにおいて、
前記他の領域には、同様の演算を多重化して行う多重化演算部が設けられ、
前記高信頼領域には、前記多重化演算部で演算された演算結果を比較する比較器が配置されることを特徴とするプログラマブルロジックデバイス。 - 請求項5に記載のプログラマブルロジックデバイスにおいて、
前記高信頼領域には、前記比較器からの出力を検査する出力検査部が配置されることを特徴とするプログラマブルロジックデバイス。 - 請求項5に記載のプログラマブルロジックデバイスにおいて、
前記高信頼領域には、前記コンフィギュレーションメモリが直列に配置されたセル列が複数並べて設けられ、
前記比較器が多重化して複数設けられ、
前記各比較器は、前記複数のセル列のうちの異なるセル列に配置されることを特徴とするプログラマブルロジックデバイス。 - 前記請求項1に記載のプログラマブルロジックデバイスに論理を集積する論理集積ツールであって、
前記プログラマブルロジックデバイスに集積される論理のデータには、前記論理が集積される対象領域が前記高信頼領域であるか前記他の領域であるかを示す識別タグが含まれ、
前記識別タグに基づいて前記対象領域に前記論理の集積を実行することを特徴とする論理集積ツール。
Priority Applications (3)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
PCT/JP2013/076467 WO2015045135A1 (ja) | 2013-09-30 | 2013-09-30 | プログラマブルロジックデバイス、及び、論理集積ツール |
JP2015538768A JP6326422B2 (ja) | 2013-09-30 | 2013-09-30 | プログラマブルロジックデバイス、及び、論理集積ツール |
US15/025,821 US9735784B2 (en) | 2013-09-30 | 2013-09-30 | Programmable logic device and logic integration tool |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
PCT/JP2013/076467 WO2015045135A1 (ja) | 2013-09-30 | 2013-09-30 | プログラマブルロジックデバイス、及び、論理集積ツール |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2015045135A1 true WO2015045135A1 (ja) | 2015-04-02 |
Family
ID=52742335
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/JP2013/076467 WO2015045135A1 (ja) | 2013-09-30 | 2013-09-30 | プログラマブルロジックデバイス、及び、論理集積ツール |
Country Status (3)
Country | Link |
---|---|
US (1) | US9735784B2 (ja) |
JP (1) | JP6326422B2 (ja) |
WO (1) | WO2015045135A1 (ja) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
EP3316135A1 (en) | 2016-10-26 | 2018-05-02 | Hitachi, Ltd. | Control system |
Families Citing this family (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP6464263B2 (ja) * | 2015-06-22 | 2019-02-06 | 株式会社日立製作所 | フィールドプログラマブルゲートアレイ |
JP7157773B2 (ja) * | 2020-01-27 | 2022-10-20 | 株式会社日立製作所 | プログラマブルデバイス及びこれを用いた制御コントローラ |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2005503668A (ja) * | 2001-09-18 | 2005-02-03 | ザイリンクス インコーポレイテッド | プログラマブルデバイスの一部にウェルバイアスを選択的にかけるための構造および方法 |
JP2007053761A (ja) * | 2005-08-16 | 2007-03-01 | Altera Corp | プログラマブルロジックデバイスの性能最適化装置および方法 |
Family Cites Families (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP3206275B2 (ja) | 1994-02-25 | 2001-09-10 | 株式会社日立製作所 | 誤り検出機能付き論理回路及びそれを用いたフォールトトレラントシステム |
US6924663B2 (en) * | 2001-12-28 | 2005-08-02 | Fujitsu Limited | Programmable logic device with ferroelectric configuration memories |
US7236000B1 (en) * | 2005-10-18 | 2007-06-26 | Xilinx, Inc. | Method and apparatus for error mitigation of programmable logic device configuration memory |
US7702978B2 (en) * | 2006-04-21 | 2010-04-20 | Altera Corporation | Soft error location and sensitivity detection for programmable devices |
US8630976B2 (en) * | 2008-08-20 | 2014-01-14 | Sap Ag | Fast search replication synchronization processes |
US8476927B2 (en) * | 2011-04-29 | 2013-07-02 | Semiconductor Energy Laboratory Co., Ltd. | Programmable logic device |
US8952723B2 (en) * | 2013-02-13 | 2015-02-10 | Semiconductor Energy Laboratory Co., Ltd. | Programmable logic device and semiconductor device |
US9275180B1 (en) * | 2014-07-14 | 2016-03-01 | Xilinx, Inc. | Programmable integrated circuit having different types of configuration memory |
-
2013
- 2013-09-30 JP JP2015538768A patent/JP6326422B2/ja not_active Expired - Fee Related
- 2013-09-30 US US15/025,821 patent/US9735784B2/en active Active
- 2013-09-30 WO PCT/JP2013/076467 patent/WO2015045135A1/ja active Application Filing
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2005503668A (ja) * | 2001-09-18 | 2005-02-03 | ザイリンクス インコーポレイテッド | プログラマブルデバイスの一部にウェルバイアスを選択的にかけるための構造および方法 |
JP2007053761A (ja) * | 2005-08-16 | 2007-03-01 | Altera Corp | プログラマブルロジックデバイスの性能最適化装置および方法 |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
EP3316135A1 (en) | 2016-10-26 | 2018-05-02 | Hitachi, Ltd. | Control system |
US10313095B2 (en) | 2016-10-26 | 2019-06-04 | Hitachi, Ltd. | Control system |
Also Published As
Publication number | Publication date |
---|---|
JPWO2015045135A1 (ja) | 2017-03-02 |
US20160241247A1 (en) | 2016-08-18 |
US9735784B2 (en) | 2017-08-15 |
JP6326422B2 (ja) | 2018-05-16 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
EP2492917B1 (en) | Error detection and correction circuitry | |
US8560927B1 (en) | Memory error detection circuitry | |
EP1848001B1 (en) | Soft error location and sensitivity detection for programmable devices | |
US9342402B1 (en) | Memory interface with hybrid error detection circuitry for modular designs | |
KR101519491B1 (ko) | 누적 테스트를 위한 반도체 메모리 장치 및 이를 포함하는 시스템 | |
US8639992B2 (en) | Soft error rate detector | |
US9575862B1 (en) | Integrated circuits with error handling capabilities | |
WO2017126451A1 (ja) | 論理集積回路および半導体装置 | |
JP5558269B2 (ja) | プログラマブル論理回路のエラー訂正回路 | |
Kyriakoulakos et al. | A novel SRAM-based FPGA architecture for efficient TMR fault tolerance support | |
JP6290934B2 (ja) | プログラマブルデバイス、エラー保持システム、及び電子システム装置 | |
JP6326422B2 (ja) | プログラマブルロジックデバイス、及び、論理集積ツール | |
EP2641174B1 (en) | Classifying a criticality of a soft error and mitigating the soft error based on the criticality | |
Ebrahim et al. | On enhancing the reliability of internal configuration controllers in FPGAs | |
US9684559B1 (en) | Methods and apparatus for storing error correction information on a memory controller circuit | |
CN110827878B (zh) | 存储器装置 | |
JP7339587B2 (ja) | 核反応検出装置及び核反応検出方法並びに核反応検出プログラム | |
KR102087509B1 (ko) | 메모리 디바이스 | |
Ferron et al. | Towards low-cost soft error mitigation in SRAM-based FPGAs: a case study on AT40K | |
JP6749965B2 (ja) | メモリデバイス | |
Argyrides et al. | Embedding current monitoring in H-tree RAM architecture for multiple SEU tolerance and reliability improvement | |
Gericota et al. | Robust configurable system design with built-in self-healing | |
McWilliam et al. | Stuck-at Fault Resilience using Redundant Transistor Logic Gates | |
Pánek et al. | The Fault-tolerant Single-FPGA Systems with a Self-repair Reconfiguration Controller | |
Leveugle et al. | On improving at no cost the quality of products built with SRAM-based FPGAs |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 13894602 Country of ref document: EP Kind code of ref document: A1 |
|
ENP | Entry into the national phase |
Ref document number: 2015538768 Country of ref document: JP Kind code of ref document: A |
|
WWE | Wipo information: entry into national phase |
Ref document number: 15025821 Country of ref document: US |
|
NENP | Non-entry into the national phase |
Ref country code: DE |
|
122 | Ep: pct application non-entry in european phase |
Ref document number: 13894602 Country of ref document: EP Kind code of ref document: A1 |