WO2017094091A1 - ダイ及びパッケージ、並びに、ダイの製造方法及びパッケージの生成方法 - Google Patents
ダイ及びパッケージ、並びに、ダイの製造方法及びパッケージの生成方法 Download PDFInfo
- Publication number
- WO2017094091A1 WO2017094091A1 PCT/JP2015/083669 JP2015083669W WO2017094091A1 WO 2017094091 A1 WO2017094091 A1 WO 2017094091A1 JP 2015083669 W JP2015083669 W JP 2015083669W WO 2017094091 A1 WO2017094091 A1 WO 2017094091A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- die
- core
- dies
- interface
- package
- Prior art date
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F15/00—Digital computers in general; Data processing equipment in general
- G06F15/76—Architectures of general purpose stored program computers
- G06F15/78—Architectures of general purpose stored program computers comprising a single central processing unit
- G06F15/7807—System on chip, i.e. computer system on a single chip; System in package, i.e. computer system on one or more chips in a single package
-
- H—ELECTRICITY
- H01—ELECTRIC ELEMENTS
- H01L—SEMICONDUCTOR DEVICES NOT COVERED BY CLASS H10
- H01L25/00—Assemblies consisting of a plurality of individual semiconductor or other solid state devices ; Multistep manufacturing processes thereof
- H01L25/03—Assemblies consisting of a plurality of individual semiconductor or other solid state devices ; Multistep manufacturing processes thereof all the devices being of a type provided for in the same subgroup of groups H01L27/00 - H01L33/00, or in a single subclass of H10K, H10N, e.g. assemblies of rectifier diodes
- H01L25/04—Assemblies consisting of a plurality of individual semiconductor or other solid state devices ; Multistep manufacturing processes thereof all the devices being of a type provided for in the same subgroup of groups H01L27/00 - H01L33/00, or in a single subclass of H10K, H10N, e.g. assemblies of rectifier diodes the devices not having separate containers
- H01L25/065—Assemblies consisting of a plurality of individual semiconductor or other solid state devices ; Multistep manufacturing processes thereof all the devices being of a type provided for in the same subgroup of groups H01L27/00 - H01L33/00, or in a single subclass of H10K, H10N, e.g. assemblies of rectifier diodes the devices not having separate containers the devices being of a type provided for in group H01L27/00
- H01L25/0655—Assemblies consisting of a plurality of individual semiconductor or other solid state devices ; Multistep manufacturing processes thereof all the devices being of a type provided for in the same subgroup of groups H01L27/00 - H01L33/00, or in a single subclass of H10K, H10N, e.g. assemblies of rectifier diodes the devices not having separate containers the devices being of a type provided for in group H01L27/00 the devices being arranged next to each other
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F15/00—Digital computers in general; Data processing equipment in general
- G06F15/16—Combinations of two or more digital computers each having at least an arithmetic unit, a program unit and a register, e.g. for a simultaneous processing of several programs
- G06F15/163—Interprocessor communication
- G06F15/17—Interprocessor communication using an input/output type connection, e.g. channel, I/O port
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F15/00—Digital computers in general; Data processing equipment in general
- G06F15/76—Architectures of general purpose stored program computers
- G06F15/78—Architectures of general purpose stored program computers comprising a single central processing unit
-
- H—ELECTRICITY
- H01—ELECTRIC ELEMENTS
- H01L—SEMICONDUCTOR DEVICES NOT COVERED BY CLASS H10
- H01L21/00—Processes or apparatus adapted for the manufacture or treatment of semiconductor or solid state devices or of parts thereof
- H01L21/70—Manufacture or treatment of devices consisting of a plurality of solid state components formed in or on a common substrate or of parts thereof; Manufacture of integrated circuit devices or of parts thereof
- H01L21/77—Manufacture or treatment of devices consisting of a plurality of solid state components or integrated circuits formed in, or on, a common substrate
- H01L21/78—Manufacture or treatment of devices consisting of a plurality of solid state components or integrated circuits formed in, or on, a common substrate with subsequent division of the substrate into plural individual devices
- H01L21/82—Manufacture or treatment of devices consisting of a plurality of solid state components or integrated circuits formed in, or on, a common substrate with subsequent division of the substrate into plural individual devices to produce devices, e.g. integrated circuits, each consisting of a plurality of components
-
- H—ELECTRICITY
- H01—ELECTRIC ELEMENTS
- H01L—SEMICONDUCTOR DEVICES NOT COVERED BY CLASS H10
- H01L21/00—Processes or apparatus adapted for the manufacture or treatment of semiconductor or solid state devices or of parts thereof
- H01L21/70—Manufacture or treatment of devices consisting of a plurality of solid state components formed in or on a common substrate or of parts thereof; Manufacture of integrated circuit devices or of parts thereof
- H01L21/77—Manufacture or treatment of devices consisting of a plurality of solid state components or integrated circuits formed in, or on, a common substrate
- H01L21/78—Manufacture or treatment of devices consisting of a plurality of solid state components or integrated circuits formed in, or on, a common substrate with subsequent division of the substrate into plural individual devices
- H01L21/82—Manufacture or treatment of devices consisting of a plurality of solid state components or integrated circuits formed in, or on, a common substrate with subsequent division of the substrate into plural individual devices to produce devices, e.g. integrated circuits, each consisting of a plurality of components
- H01L21/822—Manufacture or treatment of devices consisting of a plurality of solid state components or integrated circuits formed in, or on, a common substrate with subsequent division of the substrate into plural individual devices to produce devices, e.g. integrated circuits, each consisting of a plurality of components the substrate being a semiconductor, using silicon technology
-
- H—ELECTRICITY
- H01—ELECTRIC ELEMENTS
- H01L—SEMICONDUCTOR DEVICES NOT COVERED BY CLASS H10
- H01L25/00—Assemblies consisting of a plurality of individual semiconductor or other solid state devices ; Multistep manufacturing processes thereof
- H01L25/50—Multistep manufacturing processes of assemblies consisting of devices, each device being of a type provided for in group H01L27/00 or H01L29/00
-
- H—ELECTRICITY
- H01—ELECTRIC ELEMENTS
- H01L—SEMICONDUCTOR DEVICES NOT COVERED BY CLASS H10
- H01L27/00—Devices consisting of a plurality of semiconductor or other solid-state components formed in or on a common substrate
- H01L27/02—Devices consisting of a plurality of semiconductor or other solid-state components formed in or on a common substrate including semiconductor components specially adapted for rectifying, oscillating, amplifying or switching and having at least one potential-jump barrier or surface barrier; including integrated passive circuit elements with at least one potential-jump barrier or surface barrier
- H01L27/0203—Particular design considerations for integrated circuits
- H01L27/0207—Geometrical layout of the components, e.g. computer aided design; custom LSI, semi-custom LSI, standard cell technique
-
- H—ELECTRICITY
- H01—ELECTRIC ELEMENTS
- H01L—SEMICONDUCTOR DEVICES NOT COVERED BY CLASS H10
- H01L27/00—Devices consisting of a plurality of semiconductor or other solid-state components formed in or on a common substrate
- H01L27/02—Devices consisting of a plurality of semiconductor or other solid-state components formed in or on a common substrate including semiconductor components specially adapted for rectifying, oscillating, amplifying or switching and having at least one potential-jump barrier or surface barrier; including integrated passive circuit elements with at least one potential-jump barrier or surface barrier
- H01L27/04—Devices consisting of a plurality of semiconductor or other solid-state components formed in or on a common substrate including semiconductor components specially adapted for rectifying, oscillating, amplifying or switching and having at least one potential-jump barrier or surface barrier; including integrated passive circuit elements with at least one potential-jump barrier or surface barrier the substrate being a semiconductor body
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F15/00—Digital computers in general; Data processing equipment in general
- G06F15/76—Architectures of general purpose stored program computers
- G06F15/80—Architectures of general purpose stored program computers comprising an array of processing units with common control, e.g. single instruction multiple data processors
- G06F15/8007—Architectures of general purpose stored program computers comprising an array of processing units with common control, e.g. single instruction multiple data processors single instruction multiple data [SIMD] multiprocessors
- G06F15/803—Three-dimensional arrays or hypercubes
-
- G—PHYSICS
- G11—INFORMATION STORAGE
- G11C—STATIC STORES
- G11C5/00—Details of stores covered by group G11C11/00
- G11C5/02—Disposition of storage elements, e.g. in the form of a matrix array
- G11C5/04—Supports for storage elements, e.g. memory modules; Mounting or fixing of storage elements on such supports
-
- H—ELECTRICITY
- H10—SEMICONDUCTOR DEVICES; ELECTRIC SOLID-STATE DEVICES NOT OTHERWISE PROVIDED FOR
- H10B—ELECTRONIC MEMORY DEVICES
- H10B12/00—Dynamic random access memory [DRAM] devices
- H10B12/01—Manufacture or treatment
-
- H—ELECTRICITY
- H10—SEMICONDUCTOR DEVICES; ELECTRIC SOLID-STATE DEVICES NOT OTHERWISE PROVIDED FOR
- H10B—ELECTRONIC MEMORY DEVICES
- H10B12/00—Dynamic random access memory [DRAM] devices
- H10B12/50—Peripheral circuit region structures
Definitions
- the present invention relates to a die and a package, and a die manufacturing method and a package generation method.
- the ratio of the number of cores between the accelerator core and the CPU core varies depending on the request of the application, and thus varies depending on the type of computer that is often developed according to the main application.
- the number of cores is different for each type of computer. Since dies and packages with different ratios are manufactured, the cost is high and inefficiency.
- An object of the present invention is to provide a set of dies and a package that constitute a package having a ratio of the optimal number of cores for all types of computers at low cost and efficiency.
- a die set and package of one aspect of the present invention comprises: At least one of a first core composed of a CPU core or Latency core and a second core composed of an accelerator core or a throughput core, An external interface; A memory interface; Die interface to connect with other dies, A set of dies comprising: The die is Including a first type die and a second type die including both the first core and the second core; The ratio of the number of cores of the first core and the second core is different between the first type die and the second type die.
- the package is At least one set of dies is included.
- the memory interface is Includes an interface that conforms to the specifications for electrical contactless communication.
- the memory interface further includes: Includes a TCI compliant interface.
- the memory interface further includes: It includes an interface that conforms to the next-generation high-speed memory with a three-dimensional stack that communicates in electrical contact.
- the memory interface further includes: Including HBM compliant interface,
- the memory interface further includes: A general-purpose memory that communicates in electrical contact and an interface that conforms to DIMM (Dual Inline Memory Module) are included.
- DIMM Digital Inline Memory Module
- the memory interface further includes: Includes DDR4 compliant interface.
- the package further includes: Including two said dies interconnected by each of said die interfaces.
- the two dies connected to each other are of the same type.
- the package is further different in the two dies connected to each other.
- At least one of the two dies further includes: It is connected to another die through the die interface that is connected to each other.
- a package suitable for all types of computers and a die constituting the package can be provided at low cost and efficiently.
- non-contact communication means that one communication unit that performs communication and the other communication unit that performs communication are not in contact with each other, and a conductive member (solder, It means that communication is performed without going through any one or more of a conductive adhesive and a wire.
- Communication in contact with means that one communication unit that performs communication and a communication unit that performs communication contact each other to perform communication, or a conductive member (solder, conductive adhesive). It means that communication is performed via any one or more of an agent and a wire.
- the communication unit is a concept including a part that performs transmission and reception, a part that only transmits, and a part that includes only reception.
- FIG. 1 is a diagram showing a configuration example of a die set according to an embodiment of the present invention.
- the die set is configured to include a main die 11, a sub die 12, a sub die 13, and a memory interface die 14.
- FIG. 1A shows the configuration of the main die 11.
- the main die 11 includes an accelerator core 21, a CPU core 22, a GPIF (General Purpose Interface) 23, a TCI / MIF (Thru Chip Interface / Memory Interface) 24, and an HBM / MIF (High BandwidthM25). .
- GPIF General Purpose Interface
- TCI / MIF Thru Chip Interface / Memory Interface
- HBM / MIF High BandwidthM25
- the accelerator core 21 is a core having a small many-core configuration capable of obtaining a large amount of calculation results.
- the latency (from requesting data transfer to the device until the result is returned)
- the delay time) is large, but has a property of high throughput (a large amount of data that a computer or network can process within a certain time).
- the CPU core 22 is a large core that manages execution of an OS (Operating System), network control / load adjustment, accelerator control / load distribution adjustment, and performs low-latency and complicated calculation processing.
- OS Operating System
- network control / load adjustment network control / load adjustment
- accelerator control / load distribution adjustment accelerator control / load distribution adjustment
- the numbers “64” and “2,048” described in the accelerator core 21 and the CPU core 22 respectively indicate the number of cores of the accelerator core 21 and the CPU core 22. ing.
- the GPIF 23 is a general-purpose die interface that connects to other dies.
- the TCI / MIF 24 is a memory interface that performs non-contact communication with a memory by wireless communication between adjacent dies using magnetic field coupling.
- TCI is capable of high-speed communication with low power consumption when compared with existing wired communication methods, impedance matching is unnecessary because it is electrically non-contact, and necessary for magnetic field coupling
- the antenna can be formed in the wafer in the previous process, and there is an advantage that the yield is not affected because the post-process work does not increase.
- HBM / MIF25 is a TB / sec class broadband memory interface.
- various dies including the main die are provided with an external interface such as PCI Express.
- FIG. 1B shows the configuration of the sub-die 12. Similar to the main core 11, the sub die 12 includes an accelerator core 21, a CPU core 22, a GPIF 23, a TCI / MIF 24, and an HBM / MIF 25.
- the constituent elements of the sub die 12 in FIG. 1B are the same as those of the main die 11 in FIG.
- the ratio of the number of cores between the accelerator core 21 and the CPU core 22 is 2,048 to 64 in the main die 21, but 256 to 256 in the sub die 12.
- the number of TCI / MIF 24 is four in the main die 21 and two in the sub-die 12 and is different.
- FIG. 1C shows the configuration of the sub-die 13.
- the sub die 13 includes a CPU core 22, a GPIF 23, and an HBM / MIF 25.
- the number of cores of the CPU core 22 is 64, and the accelerator core 21 does not exist.
- the ratio of the number of cores of the accelerator core 21 and the CPU core 22 of the sub die 13 in FIG. 1C is 0 to 64, and the main core 11 in FIG. Different from the sub-core 12.
- the sub die 13 is different from the main core 11 and the sub core 12 in that the TCI / MIF 24 does not exist (0).
- FIG. 1D shows the configuration of the memory interface die 14.
- the memory interface die 14 includes a CPU core 22, a GPIF 23, and a DDR4 / MIF (Double Data Rate 4 / Memory Interface) 26.
- DDR4 / MIF26 is a memory interface that conforms to DDR4 SDRAM (Double-Data-Rate4 Synchronous Dynamic Access Memory) and supports both single memory and DIMM (Dual Inline Memory).
- DDR4 is a kind of DRAM (Dynamic Random Access Memory) standard composed of semiconductor integrated circuits.
- the above four types of dies constituting the set of dies are designed to have different sizes for the purpose of maximizing the use of an exposure mask 60 of 26 mm ⁇ 32 mm size (see FIG. 9) without any gaps. ing.
- a package is created by connecting an arbitrary number of dies of an arbitrary type and an arbitrary number of memories of an arbitrary type from the above-described four types of independent dies.
- the package means a die and a set of memories connected to the die, which are packaged with ceramics or a mold resin in order to suppress damage and impact to the die and the memory connected to the die.
- the optimal number of cores or the ratio of the number of cores of the accelerator core 21 and the CPU core 22 varies depending on the type of computer.
- packages according to the type of computer are individually manufactured. Without any problem, it is possible to provide a package with an optimal ratio or number of cores for all computers.
- FIG. 2 is a diagram showing an example of a large package for high-speed memory communication.
- TCI DRAM 30 connected to the TCI / MIF 24 of the main die 11.
- each of four TCI / MIFs 24 of the main die 11 is connected to each of four large-sized memories TCI DRAM 30.
- TCI DRAM 30 large-sized memories
- FIG. 3 is a diagram showing an example of a small package of the main die 11.
- HBM DRAM 40 connected to the HBM / MIF 25 of the main die 11.
- two HBM DRAMs 40 that are two small memories are connected to the two HBM / MIFs 25 of the main die 11.
- the package P2 in which the main die 11 and the HBM DRAM 40 are connected by the HBM / MIF 24 the package can be downsized and broadband memory communication can be realized.
- FIG. 4 is a diagram showing an example of a large package for high-speed memory communication in which the main die 11 and a plurality of types of memories are connected.
- a main die 4 includes a main die 11, a TCI DRAM 30 connected to the TCI / MIF 24 of the main die 11, and an HBM DRAM 40 connected to the HBM / MIF 25 of the main die 11.
- each of four TCI / MIFs 24 of the main die 11 is connected to each of four large-sized memories TCI DRAM 30.
- Each of the two HBM DRAMs 40 is connected to each of the two HBM / MIFs 25 of the main die 11.
- FIG. 5 is a diagram showing an example of a maximum configuration package in which the main die 11 and a plurality of types of memories are connected.
- the package P4 in FIG. 5 is for the memory interface connected to the main die 11, the TCI DRAM 30 connected to the TCI / MIF 24 of the main die 11, the HBM DRAM 40 connected to the HBM / MIF 25 of the main die 11, and the GPIF 23 of the main die 11. And a die 14.
- the memory interface die 14 includes a DDR4 / MIF 26.
- the DDR4 / MIF 26 is connected to the DDR4 DIMM 50, which is a memory module, outside the package P4.
- each of four TCI / MIFs 24 of the main die 11 is connected to each of four large memory TCI DRAMs 30.
- Each of the two HBM DRAMs 40 is connected to each of the two HBM / MIFs 25 of the main die 11.
- each of the plurality of DDR4 DIMMs 50 is connected to each of the memory interface dies 14 connected to each of the two GPIFs 23 of the main die 11.
- the main die 11 and the TCI DRAM 30 are connected by the TCI / MIF 24, the main die 11 and the HBM DRAM 40 are connected by the HBM / MIF 25, and the main die 11, the memory interface die 14, and the DDR4 DIMM 50 are connected.
- the package P4 connected by the GPIF 23, the HBM / MIF 25, and the DDR4 / MIF 26 the memory capacity of the main die 11 can be maximized.
- FIG. 6 is a diagram illustrating an example in which the same type of dies are connected to each other using the GPIF 23.
- the main die 11-A and the main die 11-B are connected to each other using the GPIF 23-A and the GPIF 23-B.
- the two GPIFs 23-B of the main die 11-B are connected to the two GPIFs 23-A of the main die 11-A, respectively.
- the number of cores of the CPU core 22 is 64, and the number of cores of the accelerator core 21 is 2,048.
- the ratio of the number of cores of the accelerator core 21 and the CPU core 22 after the connection between the main die 11-A and the main die 11-B in FIG. 6A is 4,096 to 128.
- the main die 11-A and the main die 11-B are connected while maintaining the ratio of the number of cores of the accelerator core 21 and the CPU core 22.
- the total number of cores can be increased.
- the sub die 12-C and the sub die 12-D are connected to each other by the GPIF 23-C and the GPIF 23-D.
- Each of the two GPIFs 23-D of the sub-die 12-D is connected to each of the two GPIFs 23-C of the sub-die 12-C.
- the number of cores of the CPU core 22 and the number of cores of the accelerator core 21 are both 256.
- the ratio of the number of cores of the accelerator core 21 and the CPU core 22 after the connection between the sub die 12-C and the sub die 12-D in FIG. 6B is 512: 512.
- the sub-die 12-C and the sub-die 12-D are connected while maintaining the ratio of the number of cores of the accelerator core 21 and the CPU core 22 as in the case of FIG.
- the total number of cores connecting the 12-C and the sub-die 12-D can be increased.
- the sub die 13-E and the sub die 13-F are connected to each other using the GPIF 23-E and the GPIF 23-F.
- the two GPIFs 23-F of the sub-die 13-F are connected to the two GPIFs 23-E of the sub-die 13-E, respectively.
- the number of cores of the CPU core 22 is 64, and there is no accelerator core.
- the ratio of the number of cores of the accelerator core 21 and the CPU core 22 after the connection between the sub die 13-E and the sub die 13-F in FIG. 6C is 0: 128.
- the total number of cores connecting the sub die 13-E and the sub die 13-F can be increased.
- the memory interface die 14-G and the memory interface die 14-H are connected to each other by the GPIF 23-G and the GPIF 23-H.
- the two GPIFs 23-H of the memory interface die 14-H are connected to the two GPIFs 23-G of the memory interface die 14-G, respectively.
- the number of cores of the CPU core 22 is 16 and there is no accelerator core. In other words, the ratio of the number of cores of the accelerator core 21 and the CPU core 22 after the connection between the memory interface die 14-G and the memory interface die 14-H in FIG. .
- FIG. 7 is a diagram illustrating an example in which different types of dies are connected to each other using the GPIF 23.
- the main die 11-J and the sub die 12-I are connected to each other using the GPIF 23-J and the GPIF 23-I.
- Each of the two GPIFs 23-I of the sub die 12-I is connected to each of the two GPIFs 23-J of the main die 11-J.
- the number of cores of the CPU core 22 of the main die 11-J is 64, and the number of cores of the accelerator core 21 is 2,048.
- the number of CPU cores 22 of the sub-die 12-I and the number of cores of the accelerator core 21 are both 256.
- the ratio of the number of cores of the accelerator core 21 and the CPU core 22 after the connection between the main die 11-J and the sub die 12-I in FIG. 7A is 2,304 to 320.
- the main die 11-L and the sub die 13-K are connected to each other by the GPIF 23-K and the GPIF 23-L.
- Each of the two GPIFs 23-K of the sub-die 13-K is connected to each of the two GPIFs 23-L of the main die 11-L.
- the number of cores of the CPU core 22 of the main die 11-L is 64, and the number of cores of the accelerator core 21 is 2,048.
- the number of cores of the CPU core 22 of the sub die 13-K is 64, and the accelerator core 21 does not exist.
- the ratio of the number of cores of the accelerator core 21 and the CPU core 22 after the connection between the main die 11-L and the sub die 13-K in FIG. 7B is 2,048 to 128.
- FIG. 8 is a diagram showing an example in which different types of dies having different numbers are connected to each other using the GPIF 23.
- the main die 11-N and the two memory interface dies 14-M are connected to each other by the GPIF 23-N and the GPIF 23-M.
- the two GPIFs 23-M of the memory interface die 14-M are connected to the two GPIFs 23-N of the main die 11-N, respectively.
- the number of cores of the CPU core 22 of the main die 11-N is 64, and the number of cores of the accelerator core 21 is 2,048.
- the number of CPU cores 22 of the memory interface die 14-M is 16, and the accelerator core 21 does not exist.
- the ratio of the number of cores of the accelerator core 21 and the CPU core 22 after the connection between the main die 11-N and the two memory interface dies 14-M in FIG. 8A is 2,048: 96. Become.
- the sub die 12-R and the sub die 13-S are connected to each other by the GPIF 23-R and the GPIF 23-S.
- Each of the two GPIFs 23-S of the sub die 13-S is connected to each of the two GPIFs 23-R of the sub die 12-R.
- the number of cores of the CPU core 22 of the sub die 12-R and the number of cores of the accelerator core 21 are both 256.
- the number of cores of the CPU core 22 of the sub die 13-S is 64, and the accelerator core 21 does not exist.
- the ratio of the number of cores of the accelerator core 21 and the CPU core 22 after the connection between the sub die 12-R and the sub die 13-S in FIG. 8B is 256: 320.
- the sub die 12-T and the two memory interface dies 14-U are connected to each other by the GPIF 23-T and the GPIF 23-U.
- the number of cores of the CPU core 22 of the sub die 12-T and the number of cores of the accelerator core 21 are both 256.
- the number of CPU cores 22 of the memory interface die 14-U is 16, and the accelerator core 21 does not exist.
- the ratio of the number of cores of the accelerator core 21 and the CPU core 22 after the connection between the sub die 12-T and the two memory interface dies 14-U in FIG. 8C is 256: 288.
- the sub die 13-P and the two memory interface dies 14-Q are connected to each other by the GPIF 23-P and the GPIF 23-Q.
- the number of cores of the CPU core 22 of the sub die 13-P is 64, and the accelerator core 21 does not exist.
- the number of CPU cores 22 of the memory interface die 14-Q is 16, and the accelerator core 21 does not exist.
- the ratio of the number of cores of the accelerator core 21 and the CPU core 22 after the connection between the sub die 13-P and the two memory interface dies 14-Q in FIG. 8D is 0:96.
- the die and the package are manufactured from 1 according to the optimum ratio of the number of the cores of the accelerator core 21 and the CPU core 22 which are different for each type of computer, the cost becomes high and inefficient.
- the GPIF 23 is used to connect a plurality of dies to each other, so that an accelerator core 21 that is optimal for all computers can be obtained without individually manufacturing packages according to the type of computer.
- a package having a ratio of the number of cores to the CPU core 22 can be provided.
- FIG. 9 is a diagram illustrating manufacturing a set of four types of dies using one exposure mask 60.
- circuit patterns are first designed, and what kind of circuit is arranged in a small die and how efficiently they are arranged are examined. Next, based on the examination results, an exposure mask necessary for baking a circuit pattern on the surface of a wafer (a thin plate made by processing a semiconductor material into a thin disk) is created. Specifically, the exposure mask 60 shown on the right side of FIG. 9 is created.
- the exposure mask 60 When the exposure mask 60 is created, exposure is performed through the exposure mask 60, and the pattern of the exposure mask 60 is printed on the surface of the wafer for development. Thereafter, an aluminum metal film for electrode wiring is formed on the surface of the wafer. Through these steps, many circuit patterns are created on one wafer. Specifically, four types of dies shown on the left side of FIG. 9 are created.
- the created four types of dies are tested for each die to determine whether the product is good or defective.
- the yield rate decreases as the size of the die increases, the yield increases as the size of the die decreases.
- the main die 11 is larger in size than the sub dies 12 to 13 and the memory interface die 14, the yield rate is low.
- the sub die 12 is larger in size than the sub die 13 and the memory interface die 14, the yield rate is low.
- the relationship between the sub die 13 and the memory interface die 14 is such that the sub die 13 having a larger size has a lower non-defective rate.
- the wafer is cut for each die and the finish is checked. After that, through a mounting process to fix the die so that it does not deviate from the predetermined position of the lead frame, and a wire bonding process to connect the die and the lead frame with a bonding wire, a ceramic to suppress damage and impact to the die And packaged with mold resin.
- FIG. 9 shows an example in which a die set is manufactured using an exposure mask 60 having a size of 26 mm ⁇ 32 mm.
- the die set is configured to include a main die 11, a sub die 12, a sub die 13, and a memory interface die 14.
- the four types of dies constituting the die set are designed to have different sizes for the purpose of making maximum use of the exposure mask 60 having a size of 26 mm ⁇ 32 mm without any gap.
- the accelerator core 21 and the CPU core 22 are employed as the cores included in the die, but the present invention is not particularly limited thereto. That is, the die may include at least one of the first core composed of the CPU core 22 or the Latency core and the second core composed of the accelerator core 22 or the Throughput core.
- the first core is a large core for performing complicated arithmetic processing with low latency, and is a core for managing OS execution, network control / load adjustment, accelerator control / load distribution adjustment, etc. Means.
- Typical examples of the first core include an x86 general-purpose CPU core mounted in the Intel (registered trademark) Core series and the Xeon (registered trademark) series.
- the second core means a core having a small many-core configuration that has a high latency but can obtain a large amount of calculation results with high throughput.
- Typical examples of the second core include a SIMD core of GPGPU (General-purpose computing on graphics processing units) and a MIMD core of PEZY (registered trademark) -SC (Super Computing).
- n types (n is an arbitrary integer value of 1 or more) dies can be used as a set of dies. .
- the configuration of various interfaces and the ratio of the number of cores between the accelerator core and the CPU core are merely examples for achieving the object of the present invention, and are not particularly limited.
- the package to which the present invention is applied only needs to have the following configuration, and can take various embodiments including the above-described embodiment.
- the die set to which the present invention is applied only needs to have the following configuration, and can take various embodiments including the above-described embodiment.
- the set of dies to which the present invention is applied is At least one of a first core (for example, the CPU core 22 in FIG. 1) composed of a CPU core or a Latency core and a second core (for example, the accelerator core 21 in FIG. 1) composed of an accelerator core or a throughput core,
- a first core for example, the CPU core 22 in FIG. 1
- a second core for example, the accelerator core 21 in FIG. 1
- An external interface eg PCI Express
- a memory interface eg, TCI / MIF in FIG. 1
- a die interface eg, GPIF in FIG.
- a die set comprising: The die is Including a first type die and a second type die including both the first core and the second core; The ratio of the number of cores of the first core and the second core is different between the first type die and the second type die, A die set is sufficient.
- a TCI DRAM can be connected to a TCI-compliant interface, a large die set for high-speed memory communication can be easily manufactured.
- the HBM DRAM can be connected to the interface conforming to the HBM, a wide-band small die set can be easily manufactured.
- TCI DRAM and HBM DRAM can be connected to TCI compliant interface and HBM compliant interface, respectively.
- a die set that can realize high-speed, large-capacity, and wide-band memory communication can be easily manufactured.
- a TCI DRAM, an HBM DRAM 40, a DDR4 DIMM 50, a TCI compliant interface, an HBM compliant interface, and a memory interface die having an DDR4 compliant interface can be connected to each other, so that a large-capacity set of large dies can be easily manufactured.
- the core of the entire package is maintained while maintaining the ratio of the number of cores of the accelerator core 21 and the CPU core 22.
- a set of dies for increasing the number can be easily manufactured.
- a set of dies for connecting different types of independent dies to each other can be easily manufactured. Can do.
- a set of dies for connecting different types of dies with different numbers can be easily manufactured using the GPIF 23. This makes it easy to manufacture a set of dies that make up a package with the ratio of the number of cores of the accelerator core 21 and the CPU core 22 that are optimal for all computers without individually manufacturing packages according to the type of computer. can do.
- the package to which the present invention is applied is At least one of a first core (for example, the CPU core 22 in FIG. 1) composed of a CPU core or Latency core and a second core (for example, the accelerator core 21 in FIG. 1) composed of an accelerator core or a throughput core
- a first core for example, the CPU core 22 in FIG. 1
- a second core for example, the accelerator core 21 in FIG. 1
- An external interface eg PCI Express
- a memory interface eg, TCI / MIF in FIG. 1
- a die interface eg, GPIF in FIG. 1 that connects to other dies
- a package comprising at least one die comprising: The die is Including a first type die and a second type die including both the first core and the second core;
- the ratio of the number of cores of the first core and the second core includes at least one die that is different between the first type die and the second type die, respectively.
- a package is enough.
- the memory interface die 14 -M is connected to the main die 11. Since the yield is higher, the package is efficiently produced without causing a situation where only the main die 11 has an increased inventory.
- a TCI DRAM can be connected to a TCI-compliant interface, a large package for high-speed memory communication can be easily manufactured.
- TCI DRAM and HBM DRAM can be connected to TCI compliant interface and HBM compliant interface, respectively. It is possible to easily manufacture a package that can realize high-speed, large-capacity, and wide-band memory communication utilizing the above.
- a TCI DRAM, an HBM DRAM 40, a DDR4 DIMM 50, a TCI compliant interface, an HBM compliant interface, and a memory interface die having an DDR4 compliant interface respectively. Therefore, a large-capacity large package can be easily manufactured.
- the core of the entire package is maintained while maintaining the ratio of the number of cores of the accelerator core 21 and the CPU core 22. You can increase the number.
Abstract
Description
CPUコア若しくはLatencyコアからなる第1コアと、Acceleratorコア若しくはThroughputコアからなる第2コアとのうち少なくとも一方を備え、
外部インターフェースと、
メモリインターフェースと、
他のダイと接続するダイインターフェースと、
を備える前記ダイのセットであって、
前記ダイは、
前記第1コアと前記第2コアとの両方を含む第1種類のダイと第2種類のダイとを含み、
前記第1コアと前記第2コアとのコア数の比率は、前記第1種類のダイと前記第2種類のダイとでそれぞれ異なっている。
前記ダイのセットを少なくとも1つ含む。
電気的に非接触に通信を行う仕様に準拠したインターフェースを含む。
TCIに準拠したインターフェースを含む。
電気的に接触して通信を行う3次元積層の次世代高速メモリに準拠したインターフェースを含む。
HBMに準拠したインターフェースを含む、
電気的に接触して通信を行う汎用メモリ、及びDIMM(Dual Inline Memory Module)に準拠したインターフェースを含む。
DDR4に準拠したインターフェースを含む。
夫々の前記ダイインターフェースで相互に接続されている2つの前記ダイを含む。
相互に接続されている前記2つのダイが同種である。
相互に接続されている前記2つのダイが異種である。
相互に接続されている前記ダイインターフェースで別の前記ダイと接続されている。
TCI/MIF24は、磁界結合を用いた近接ダイ間無線通信によりメモリと非接触に通信を行うメモリインターフェースである。
サブダイ12は、メインコア11と同様に、アクセラレーターコア21と、CPUコア22と、GPIF23と、TCI/MIF24と、HBM/MIF25と、を備える。
ただし、アクセラレーターコア21とCPUコア22とのコア数の比率が、メインダイ21では2,048対64であるのに対し、サブダイ12では、256対256であり異なる。また、TCI/MIF24の個数も、メインダイ21では4個であるのに対し、サブダイ12では2個であり異なる。
サブダイ13は、CPUコア22と、GPIF23と、HBM/MIF25と、を備える。サブダイ13では、CPUコア22のコア数は64であり、アクセラレーターコア21は存在しない。換言すると、図1(C)のサブダイ13のアクセラレーターコア21とCPUコア22とのコア数の比率は、0対64であり、図1(A)のメインコア11とも図1(B)のサブコア12とも異なる。また、サブダイ13では、TCI/MIF24が存在しない(0個)である点も、メインコア11ともサブコア12とも異なる。
メモリインターフェース用ダイ14は、CPUコア22と、GPIF23と、DDR4/MIF(Double Data Rate 4/Memory Interface)26と、を備える。
ここで、パッケージとは、ダイ及びダイに接続されたメモリへの傷や衝撃を抑えるために、ダイ及びダイに接続されたメモリのセットをセラミックやモールド樹脂によってパッケージしたものをいう。
このように、メインダイ11とTCI DRAM30とがTCI/MIF24により非接触で接続したパッケージP1を適用することで、高速かつ大容量のメモリ通信が実現できる。
このように、メインダイ11と、HBM DRAM40とがHBM/MIF24により接続したパッケージP2を適用することで、パッケージの小型化と広帯域のメモリ通信とが実現できる。
サブダイ13-Eの2つのGPIF23-Eの夫々に、サブダイ13-Fの2つのGPIF23-Fの夫々が接続されている。
第1のコアの代表的な例としては、Intel(登録商標)のCoreシリーズやXeon(登録商標)シリーズに搭載されるx86系の汎用CPUコアなどがある。
第2のコアの代表的な例としては、GPGPU(General-purpose computing on graphics processing units)のSIMDコアやPEZY(登録商標)-SC(Super Computing)のMIMDコアなどがある。
CPUコア若しくはLatencyコアからなる第1コア(例えば図1のCPUコア22)と、Acceleratorコア若しくはThroughputコアからなる第2コア(例えば図1のアクセラレーターコア21)とのうち少なくとも一方を備え、
外部インターフェース(例えばPCIエクスプレス)と、
メモリインターフェース(例えば図1のTCI/MIF)と、
他のダイと接続するダイインターフェース(例えば図1のGPIF)と、
を備えるダイのセットであって、
前記ダイは、
前記第1コアと前記第2コアとの両方を含む第1種類のダイと第2種類のダイとを含み、
前記第1コアと前記第2コアとのコア数の比率は、前記第1種類のダイと前記第2種類のダイとでそれぞれ異なっている、
ダイのセットであれば足りる。
これにより、コンピュータの種類に応じたパッケージを個別に製造することなく、全てのコンピュータにとって最適なアクセラレーターコア21とCPUコア22とのコア数の比率のダイのセットを提供することができる。
さらに、図8に示すように、GPIF23を用いて、数が異なる異種類のダイを相互に接続させるためのダイのセットを容易に製造することができる。
これにより、コンピュータの種類に応じたパッケージを個別に製造することなく、全てのコンピュータにとって最適なアクセラレーターコア21とCPUコア22とのコア数の比率のパッケージを構成するダイのセットを容易に製造することができる。
CPUコア若しくはLatencyコアからなる第1コア(例えば図1のCPUコア22)と、Acceleratorコア若しくはThroughputコアからなる第2コア(例えば図1のアクセラレーターコア21)とのうち少なくとも一方を備え、
外部インターフェース(例えばPCIエクスプレス)と、
メモリインターフェース(例えば図1のTCI/MIF)と、
他のダイと接続するダイインターフェース(例えば図1のGPIF)と、
を備えるダイを少なくとも1つ含むパッケージであって、
前記ダイは、
前記第1コアと前記第2コアとの両方を含む第1種類のダイと第2種類のダイとを含み、
前記第1コアと前記第2コアとのコア数の比率は、前記第1種類のダイと前記第2種類のダイとでそれぞれ異なっているダイを少なくとも1つ含む、
パッケージであれば足りる。
これにより、コンピュータの種類に応じたパッケージを個別に製造することなく、全てのコンピュータにとって最適なアクセラレーターコア21とCPUコア22とのコア数の比率のパッケージを提供することができる。
さらに、図8に示すように、GPIF23を用いて、数が異なる異種類のダイを相互に接続させることができる。
これにより、コンピュータの種類に応じたパッケージを個別に製造することなく、全てのコンピュータにとって最適なアクセラレーターコア21とCPUコア22とのコア数の比率となるパッケージを容易に製造することができる。
12,12-C,D,I,R,T サブダイ
13,13-E,F,P,S サブダイ
14,14-G,H,M,Q,U メモリインターフェース用ダイ
21 アクセラレーターコア
22 CPUコア
23,23-A~U GPIF
24 TCI/MIF
25 HBM/MIF
26 DDR4/MIF
30 TCI DRAM
40 HBM DRAM
50 DDR4 DIMM
60 露光マスク
P1 パッケージ
P2 パッケージ
P3 パッケージ
P4 パッケージ
Claims (14)
- CPUコア若しくはLatencyコアからなる第1コアと、Acceleratorコア若しくはThroughputコアからなる第2コアとのうち少なくとも一方を備え、
外部インターフェースと、
メモリインターフェースと、
他のダイと接続するダイインターフェースと、
を備えるダイのセットであって、
前記ダイは、
前記第1コアと前記第2コアとの両方を含む第1種類のダイと第2種類のダイとを含み、
前記第1コアと前記第2コアとのコア数の比率は、前記第1種類のダイと前記第2種類のダイとでそれぞれ異なっている、
ダイのセット。 - 請求項1に記載の前記ダイのセットを少なくとも1つ含む、
パッケージ。 - 前記メモリインターフェースは、
電気的に非接触に通信を行う仕様に準拠したインターフェースを含む、
請求項2に記載のパッケージ。 - 前記メモリインターフェースは、
TCIに準拠したインターフェースを含む
請求項3に記載のパッケージ。 - 前記メモリインターフェースは、
電気的に接触して通信を行う3次元積層の次世代高速メモリに準拠したインターフェースをさらに含む、
請求項2乃至4のうち何れか1項に記載のパッケージ。 - 前記メモリインターフェースは、
HBMに準拠したインターフェースをさらに含む、
請求項5に記載のパッケージ。 - 前記メモリインターフェースは、
電気的に接触して通信を行う汎用メモリ、及びDIMM(Dual Inline Memory Module)に準拠したインターフェースをさらに含む、
請求項2乃至6のうち何れか1項に記載のパッケージ。 - 前記メモリインターフェースは、
DDR4に準拠したインターフェースをさらに含む、
請求項7に記載のパッケージ。 - 前記パッケージは、
夫々の前記ダイインターフェースで相互に接続されている2つの前記ダイを含む、
請求項2乃至8のうち何れか1項に記載のパッケージ。 - 前記パッケージは、
相互に接続されている前記2つのダイが同種である、
請求項9に記載のパッケージ。 - 前記パッケージは、
相互に接続されている前記2つのダイが異種である、
請求項9に記載のパッケージ。 - 前記2つのダイのうち、少なくとも1つは、さらに、
相互に接続されている前記ダイインターフェースで別の前記ダイと接続されている、
請求項2乃至11のうち何れか1項に記載のパッケージ。 - CPUコア若しくはLatencyコアからなる第1コアと、Acceleratorコア若しくはThroughputコアからなる第2コアとのうち少なくとも一方を備え、
外部インターフェースと、
メモリインターフェースと、
他のダイと接続するダイインターフェースと、
を備える複数種類のダイを生成するための露光マスクを用意し、
当該露光マスクを用いて前記複数種類のダイのセットを製造する、
ダイのセットの製造方法。 - CPUコア若しくはLatencyコアからなる第1コアと、Acceleratorコア若しくはThroughputコアからなる第2コアとのうち少なくとも一方を備え、
外部インターフェースと、
メモリインターフェースと、
他のダイと接続するダイインターフェースと、
を備えるダイを少なくとも1つ含むように、
パッケージを製造する、
パッケージ製造方法。
Priority Applications (6)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
PCT/JP2015/083669 WO2017094091A1 (ja) | 2015-11-30 | 2015-11-30 | ダイ及びパッケージ、並びに、ダイの製造方法及びパッケージの生成方法 |
EP15909721.1A EP3385857A4 (en) | 2015-11-30 | 2015-11-30 | Die and package, and manufacturing method for die and producing method for package |
CN201580084961.8A CN108292292A (zh) | 2015-11-30 | 2015-11-30 | 管芯和封装件、以及管芯的制造方法和封装件的生成方法 |
JP2016517011A JP5956708B1 (ja) | 2015-11-30 | 2015-11-30 | ダイ及びパッケージ、並びに、ダイの製造方法及びパッケージの生成方法 |
KR1020187018353A KR20180088437A (ko) | 2015-11-30 | 2015-11-30 | 다이 및 패키지, 및 다이의 제조 방법 및 패키지의 생성 방법 |
US15/780,061 US10818638B2 (en) | 2015-11-30 | 2015-11-30 | Die and package |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
PCT/JP2015/083669 WO2017094091A1 (ja) | 2015-11-30 | 2015-11-30 | ダイ及びパッケージ、並びに、ダイの製造方法及びパッケージの生成方法 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2017094091A1 true WO2017094091A1 (ja) | 2017-06-08 |
Family
ID=56513722
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/JP2015/083669 WO2017094091A1 (ja) | 2015-11-30 | 2015-11-30 | ダイ及びパッケージ、並びに、ダイの製造方法及びパッケージの生成方法 |
Country Status (6)
Country | Link |
---|---|
US (1) | US10818638B2 (ja) |
EP (1) | EP3385857A4 (ja) |
JP (1) | JP5956708B1 (ja) |
KR (1) | KR20180088437A (ja) |
CN (1) | CN108292292A (ja) |
WO (1) | WO2017094091A1 (ja) |
Families Citing this family (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US10936942B2 (en) * | 2017-11-21 | 2021-03-02 | Google Llc | Apparatus and mechanism for processing neural network tasks using a single chip package with multiple identical dies |
US11171115B2 (en) | 2019-03-18 | 2021-11-09 | Kepler Computing Inc. | Artificial intelligence processor with three-dimensional stacked memory |
US11836102B1 (en) | 2019-03-20 | 2023-12-05 | Kepler Computing Inc. | Low latency and high bandwidth artificial intelligence processor |
US11043472B1 (en) | 2019-05-31 | 2021-06-22 | Kepler Compute Inc. | 3D integrated ultra high-bandwidth memory |
US11844223B1 (en) | 2019-05-31 | 2023-12-12 | Kepler Computing Inc. | Ferroelectric memory chiplet as unified memory in a multi-dimensional packaging |
US11289497B2 (en) | 2019-12-27 | 2022-03-29 | Kepler Computing Inc. | Integration method of ferroelectric memory array |
US11791233B1 (en) | 2021-08-06 | 2023-10-17 | Kepler Computing Inc. | Ferroelectric or paraelectric memory and logic chiplet with thermal management in a multi-dimensional packaging |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPS5618439A (en) * | 1979-07-25 | 1981-02-21 | Fujitsu Ltd | Semiconductor device consisting of different ic |
JPH05190758A (ja) * | 1992-01-09 | 1993-07-30 | Sharp Corp | 半導体装置及びその製造方法 |
Family Cites Families (29)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO1998025271A1 (fr) * | 1996-12-04 | 1998-06-11 | Hitachi, Ltd. | Dispositif comprenant un circuit integre a semi-conducteur |
JP3942198B2 (ja) * | 1996-12-04 | 2007-07-11 | 株式会社ルネサステクノロジ | 半導体集積回路装置 |
JP2001100395A (ja) * | 1999-09-30 | 2001-04-13 | Toshiba Corp | 露光用マスク及びその製造方法 |
EP1496435A1 (en) * | 2003-07-11 | 2005-01-12 | Yogitech Spa | Dependable microcontroller, method for designing a dependable microcontroller and computer program product therefor |
JP4592413B2 (ja) * | 2004-12-27 | 2010-12-01 | 三洋電機株式会社 | 回路装置 |
US7566971B2 (en) * | 2005-05-27 | 2009-07-28 | Semiconductor Energy Laboratory Co., Ltd. | Semiconductor device and manufacturing method thereof |
GB2441726B (en) * | 2005-06-24 | 2010-08-11 | Metaram Inc | An integrated memory core and memory interface circuit |
JP4936517B2 (ja) * | 2006-06-06 | 2012-05-23 | 学校法人早稲田大学 | ヘテロジニアス・マルチプロセッサシステムの制御方法及びマルチグレイン並列化コンパイラ |
EP3540736B1 (en) * | 2006-12-14 | 2023-07-26 | Rambus Inc. | Multi-die memory device |
US7962771B2 (en) * | 2007-12-31 | 2011-06-14 | Intel Corporation | Method, system, and apparatus for rerouting interrupts in a multi-core processor |
WO2010035315A1 (ja) * | 2008-09-24 | 2010-04-01 | 富士通株式会社 | マルチコアcpuにおける消費電力制御方法,消費電力制御プログラム及び情報処理システム |
US8796740B1 (en) | 2009-01-13 | 2014-08-05 | Altera Corporation | Using a single mask for various design configurations |
US8330489B2 (en) * | 2009-04-28 | 2012-12-11 | International Business Machines Corporation | Universal inter-layer interconnect for multi-layer semiconductor stacks |
US8719547B2 (en) * | 2009-09-18 | 2014-05-06 | Intel Corporation | Providing hardware support for shared virtual memory between local and remote physical memory |
JP2011108140A (ja) | 2009-11-20 | 2011-06-02 | Seiko Epson Corp | データ処理装置 |
US8683243B2 (en) * | 2011-03-11 | 2014-03-25 | Intel Corporation | Dynamic core selection for heterogeneous multi-core systems |
US9405550B2 (en) * | 2011-03-31 | 2016-08-02 | International Business Machines Corporation | Methods for the transmission of accelerator commands and corresponding command structure to remote hardware accelerator engines over an interconnect link |
US9432298B1 (en) | 2011-12-09 | 2016-08-30 | P4tents1, LLC | System, method, and computer program product for improving memory systems |
WO2013078536A1 (en) * | 2011-12-01 | 2013-06-06 | Mosaid Technologies Incorporated | Cpu with stacked memory |
US20130141442A1 (en) | 2011-12-06 | 2013-06-06 | John W. Brothers | Method and apparatus for multi-chip processing |
US8984228B2 (en) * | 2011-12-13 | 2015-03-17 | Intel Corporation | Providing common caching agent for core and integrated input/output (IO) module |
US9129158B1 (en) * | 2012-03-05 | 2015-09-08 | Hrl Laboratories, Llc | Method and system for embedding visual intelligence |
US9396020B2 (en) * | 2012-03-30 | 2016-07-19 | Intel Corporation | Context switching mechanism for a processing core having a general purpose CPU core and a tightly coupled accelerator |
US8922243B2 (en) | 2012-12-23 | 2014-12-30 | Advanced Micro Devices, Inc. | Die-stacked memory device with reconfigurable logic |
US9465432B2 (en) * | 2013-08-28 | 2016-10-11 | Via Technologies, Inc. | Multi-core synchronization mechanism |
US9032099B1 (en) | 2013-12-12 | 2015-05-12 | Intel Corporation | Writeback mechanisms for improving far memory utilization in multi-level memory architectures |
US9740617B2 (en) * | 2014-12-23 | 2017-08-22 | Intel Corporation | Hardware apparatuses and methods to control cache line coherence |
US9734103B2 (en) * | 2015-01-25 | 2017-08-15 | Dell Products, L.P. | Systems and methods for transforming a central processing unit (CPU) socket into a memory and/or input/output (I/O) expander |
US10091295B1 (en) * | 2015-09-23 | 2018-10-02 | EMC IP Holding Company LLC | Converged infrastructure implemented with distributed compute elements |
-
2015
- 2015-11-30 JP JP2016517011A patent/JP5956708B1/ja active Active
- 2015-11-30 US US15/780,061 patent/US10818638B2/en active Active
- 2015-11-30 WO PCT/JP2015/083669 patent/WO2017094091A1/ja active Application Filing
- 2015-11-30 EP EP15909721.1A patent/EP3385857A4/en not_active Ceased
- 2015-11-30 CN CN201580084961.8A patent/CN108292292A/zh active Pending
- 2015-11-30 KR KR1020187018353A patent/KR20180088437A/ko not_active Application Discontinuation
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPS5618439A (en) * | 1979-07-25 | 1981-02-21 | Fujitsu Ltd | Semiconductor device consisting of different ic |
JPH05190758A (ja) * | 1992-01-09 | 1993-07-30 | Sharp Corp | 半導体装置及びその製造方法 |
Non-Patent Citations (4)
Title |
---|
HIROSHIGE GOTO: "Multi Die eno Kaji o Kitta ATI GPU no Henka ga Motarasu Mirai", DOS/V POWER REPORT, vol. 18, 1 January 2008 (2008-01-01), pages 188 - 192, XP009507572 * |
NAOKI ASAKAWA: "Data Bunseki Kiban wa HPC e Google, IBM ga Chushi suru Jisedai Core Gijutsu", NIKKEI COMPUTER, vol. 865, 24 July 2014 (2014-07-24), pages 58 - 63, XP009507491 * |
NORIYUKI MIURA ET AL.: "A 1 TB/s 1 pJ/b 6.4mm2/TB/s QDR Inductive- Coupling Interface Between 65-nm CMOS Logic and Emulated 100-nm DRAM", IEEE JOURNAL ON EMERGING AND SELECTED TOPICS IN CIRCUITS AND SYSTEMS, vol. 2, no. 2, 7 June 2012 (2012-06-07), pages 249 - 256, XP011446365, ISSN: 2156-3357, DOI: doi:10.1109/JETCAS.2012.2193836 * |
See also references of EP3385857A4 * |
Also Published As
Publication number | Publication date |
---|---|
EP3385857A1 (en) | 2018-10-10 |
EP3385857A4 (en) | 2018-12-26 |
US10818638B2 (en) | 2020-10-27 |
US20180350773A1 (en) | 2018-12-06 |
CN108292292A (zh) | 2018-07-17 |
JP5956708B1 (ja) | 2016-07-27 |
JPWO2017094091A1 (ja) | 2017-12-14 |
KR20180088437A (ko) | 2018-08-03 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
JP5956708B1 (ja) | ダイ及びパッケージ、並びに、ダイの製造方法及びパッケージの生成方法 | |
US9361254B2 (en) | Memory device formed with a semiconductor interposer | |
DE112012006625T5 (de) | Mehrchiplagenhalbleiterstruktur mit vertikalem Zwischenseitenchip und Halbleiterpaket dafür | |
TW201515176A (zh) | 具有一控制器及一記憶體堆疊之彈性記憶體系統 | |
US20200083194A1 (en) | Stacked dice systems | |
US11742293B2 (en) | Multiple die package using an embedded bridge connecting dies | |
WO2017094092A1 (ja) | ダイ及びパッケージ | |
US20210288034A1 (en) | Multi-die, vertical-wire package-in-package apparatus, and methods of making same | |
US10833020B2 (en) | High density interconnect structures configured for manufacturing and performance | |
US11133261B2 (en) | Electronic device packaging | |
US9019738B2 (en) | Memory device having sequentially cascading dices | |
US20190252321A1 (en) | Interconnector with bundled interconnects | |
US10903155B2 (en) | Vertical modular stiffeners for stacked multi-device packages | |
US20210028084A1 (en) | Variable-thickness integrated heat spreader (ihs) | |
WO2020107447A1 (zh) | 芯片制造方法及芯片结构 | |
US11562954B2 (en) | Frame-array interconnects for integrated-circuit packages | |
US10497669B2 (en) | Hybrid die stacking | |
US20210263880A1 (en) | Disaggregated die with input/output (i/o) tiles | |
JP2010109111A (ja) | 半導体集積回路 | |
US20220100692A1 (en) | Electrical and optical interfaces at different heights along an edge of a package to increase bandwidth along the edge | |
JP2018097716A (ja) | 電子回路装置 | |
DE112022000550T5 (de) | Komplementäre chip-zu-chip-schnittstelle |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
ENP | Entry into the national phase |
Ref document number: 2016517011 Country of ref document: JP Kind code of ref document: A |
|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 15909721 Country of ref document: EP Kind code of ref document: A1 |
|
NENP | Non-entry into the national phase |
Ref country code: DE |
|
ENP | Entry into the national phase |
Ref document number: 20187018353 Country of ref document: KR Kind code of ref document: A |
|
WWE | Wipo information: entry into national phase |
Ref document number: 2015909721 Country of ref document: EP |
|
ENP | Entry into the national phase |
Ref document number: 2015909721 Country of ref document: EP Effective date: 20180702 |