Classifications

• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06F—ELECTRIC DIGITAL DATA PROCESSING
  • G06F12/00—Accessing, addressing or allocating within memory systems or architectures
  • G06F12/02—Addressing or allocation; Relocation
  • G06F12/08—Addressing or allocation; Relocation in hierarchically structured memory systems, e.g. virtual memory systems
  • G06F12/0802—Addressing of a memory level in which the access to the desired data or data block requires associative addressing means, e.g. caches
  • G06F12/0893—Caches characterised by their organisation or structure
• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06F—ELECTRIC DIGITAL DATA PROCESSING
  • G06F12/00—Accessing, addressing or allocating within memory systems or architectures
  • G06F12/02—Addressing or allocation; Relocation
  • G06F12/08—Addressing or allocation; Relocation in hierarchically structured memory systems, e.g. virtual memory systems
  • G06F12/0802—Addressing of a memory level in which the access to the desired data or data block requires associative addressing means, e.g. caches
  • G06F12/0806—Multiuser, multiprocessor or multiprocessing cache systems
• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06F—ELECTRIC DIGITAL DATA PROCESSING
  • G06F12/00—Accessing, addressing or allocating within memory systems or architectures
  • G06F12/02—Addressing or allocation; Relocation
  • G06F12/08—Addressing or allocation; Relocation in hierarchically structured memory systems, e.g. virtual memory systems
  • G06F12/0802—Addressing of a memory level in which the access to the desired data or data block requires associative addressing means, e.g. caches
  • G06F12/0806—Multiuser, multiprocessor or multiprocessing cache systems
  • G06F12/0811—Multiuser, multiprocessor or multiprocessing cache systems with multilevel cache hierarchies
• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; 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
• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06F—ELECTRIC DIGITAL DATA PROCESSING
  • G06F2212/00—Indexing scheme relating to accessing, addressing or allocation within memory systems or architectures
  • G06F2212/22—Employing cache memory using specific memory technology
  • G06F2212/222—Non-volatile memory

Definitions

• the present disclosure relates generally to spin transfer torque
• STT-MRAM magnetoresistive random-access memory
• multi-core processers such as the Intel ® CoreTM i7 Processor
• multi-tier memory hierarchy Each core in the multi-core processor may have a dedicated hierarchy of cache (Level 1 (LI) - Level 2 (L2)) and may also share a lower level cache (Level 3 (L3)).
• L2 Level 1
• L3 Level 3
• the lower level cache tends to have an increased chip area, and therefore, the lower level cache tends to increase the costs associated with multi-core processors.
• the embedded cache e.g., LI - L3
• SRAM static random access memory
• SRAM may be desirable because it is logic compatible and fast.
• SRAM has a large size area that results in a high cost, and furthermore, SRAM tends to leak power. Therefore, the use of SRAM for the lower level cache of a multi-core processor may be undesirable due to, at least, an increased cost and power consumption.
• Hybrid cache refers to a system that utilizes different types of memory in the cache levels.
• the hybrid cache may be configured as an inter-level hybrid cache, intra-level hybrid cache, or three-dimensional (3D) hybrid cache.
• FIGURE 1 illustrates an example of a prior art inter- level hybrid cache 100.
• the core 102 which includes LI
• L2 cache 104 may use SRAM
• the L3 cache 106 may use a different type of memory, such as embedded dynamic random access memory (eDRAM), MRAM, or phase-change random access memory (PRAM).
• the inter- level hybrid cache 100 is not limited to the memory types discussed in FIGURE 1 and may utilize various memory types for each cache layer 102-106.
• FIGURE 2 illustrates an example of a prior art intra- level hybrid cache 200. As illustrated in FIGURE 2, the core 202 (with LI) and a first L2 cache 204 may use SRAM.
• the first L2 cache 204 may be configured for high speed memory, and therefore, the first L2 cache 204 may utilize SRAM.
• a second L2 cache 206 may be configured for slow memory, and therefore, the second L2 cache 206 may be a type of memory that is different from the type of memory used in the first L2 cache 204.
• the second L2 cache 206 may use eDRAM, MRAM, or PRAM.
• the intra-level hybrid cache 200 is not limited to the memory types illustrated in FIGURE 2 and may utilize various memory types for each cache layer 202-206.
• the inter-level or intra-level hybrid cache may use various memory technologies. Still, the inter-level cache and intra-level hybrid cache are complex and have increased fabrication costs due to the monolithic integration of various memory technologies. Therefore, the use of an inter-level or intra-level hybrid cache for the lower level cache of a multi-core processor is undesirable due to the increased costs and complexity.
• FIGURE 3 illustrates an example of a prior art 3D hybrid cache 300.
• the core 302 (with LI) and a first L2 cache 304 may use SRAM.
• a second L2 cache 306 may be configured to utilize slow memory, and therefore, the second L2 cache 306 may utilize a different type of memory, such as eDRAM, MRAM, or PRAM.
• the core 302, first L2 cache 304, and second L2 cache 306 may all be defined on a single chip.
• the L3 cache 308, may use a memory such as PRAM, and may be defined on a separate die that is connected to the chip including the core 302, first L2 cache 304, and second L2 cache 306. That is, the L3 cache 308 is on a layer (e.g., chip) that is different from the layer including the core 302, first L2 cache 304, and second L2 cache 306.
• the 3D hybrid cache may be desirable due to the use of heterogeneous memories in multiple layers and high-density memories in known good die. Still, a 3D hybrid cache specifies a multi-die solution that is stacked. Accordingly, the 3D hybrid cache may increase costs due to the stacked die. Furthermore, the stacking of the die also increases the overhead (e.g., cost, reliability). SUMMARY
• a multi-core processor includes a first spin transfer torque magnetoresistive random-access memory (STT-MRAM) cache associated with a first core of the multi-core processor and tuned according to first attributes.
• the multi-core processor further includes a second STT-MRAM cache associated with a second core of the multi-core processor and tuned according to second attributes.
• STT-MRAM spin transfer torque magnetoresistive random-access memory
• a multi-core processor includes a first storage means associated with a first core of the multi- core processor and tuned according to first attributes.
• the multi-core processor also includes a second storage means associated with a second core of the multi-core processor and tuned according to second attributes.
• a method for associating caches in a multi-core processor includes associating a first spin transfer torque magnetoresistive random-access memory (STT-MRAM) cache with a first core of the multi-core processor and tuned according to first attributes.
• the method also includes associating a second STT-MRAM cache with a second core of the multi-core processor and tuned according to second attributes.
• STT-MRAM spin transfer torque magnetoresistive random-access memory
• a method for fabricating caches for a multi-core processor includes tuning a first spin transfer torque magnetoresistive random-access memory (STT-MRAM) cache according to first attributes.
• the method also includes tuning a second STT-MRAM cache according to second attributes.
• STT-MRAM spin transfer torque magnetoresistive random-access memory
• FIGURES 1-3 illustrate prior art memory systems.
• FIGURES 4-5 illustrate a multi-tiered multi-core processor utilizing an STT- MRAM cache according to an aspect of the present disclosure.
• FIGURE 6 illustrates a block diagram for a method of associating and tuning an STT-MRAM cache in a multi-core processor according to an aspect of the present disclosure.
• FIGURE 7 illustrates an exemplary wireless communication system in which an embodiment of the disclosure may be advantageously employed.
• FIGURE 8 is a block diagram illustrating a design workstation used for circuit, layout, and logic design of a semiconductor component according to one aspect of the present disclosure.
• One aspect of the disclosure provides a lower level cache that may be tunable for speed, power, and density, while decreasing costs and complexity.
• a multi-tiered STT-MRAM cache for a multi-core processor is described.
• MRAM magnetoresistive RAM
• the storage elements are formed from two ferromagnetic layers separated by a tunneling layer.
• One of the two ferromagnetic layers which is referred to as the fixed layer or pinned layer, has a magnetization that is fixed in a particular direction.
• the other ferromagnetic magnetic layer which is referred to as the free layer, has a magnitization direction that can be altered to represent either "1" when the free layer magnetization is anti-parallel to the fixed layer magnitization or "0" when the free layer magnetization is parallel to the fixed layer magnetization.
• One such device having a fixed layer, a tunneling layer, and a free layer is a magnetic tunnel junction (MTJ).
• the electrical resistance of an MTJ depends on whether the free layer magnitization and fixed layer magnitization are parallel or anti-parallel with each other.
• a memory device such as MRAM is built from an array of individually addressable MTJs.
• STT-MRAM is a type of MRAM.
• the free layer magnetization of STT- MRAM may be switched by an electrical current that passes through an MTJ.
• STT- MRAM is differentiated from conventional MRAM which uses a magnetic field.
• STT- MRAM is tunable for speed, power, density, and cost.
• STT-MRAM may be tailored as an alternative to memories utilized for lower level cache.
• STT-MRAM cells and macros can be fabricated in multiple configurations (e.g., multi-tiered) on a monolithic die without incurring extra process steps and cost.
• the STT-MRAM is not limited to the lower level (L3) cache and may be utilized for other cache layers (e.g., LI or L2) as desired.
• each processor core may process different workloads, and therefore, each processor core may specify a different type of memory. For example, one processor core may specify a memory tuned for high performance while a second processor core may specify a memory tuned for low power.
• the lower level cache is the same circuitry and is not tuned for each core.
• the multi-tiered STT-MRAM cache may be tuned (e.g., customized) for each processor core so that the tailored multi-tiered STT-MRAM cache may improve the efficiency of each processor core, and thereby, deliver improved chip-level power-performance.
• each multi-tiered STT-MRAM may be shared by at least two processor cores.
• the term processor core may sometimes be referred to as a core.
• FIGURE 4 illustrates a multi-tiered multi-core processor 400 utilizing an STT-MRAM cache according to an aspect of the present disclosure. As illustrated in
• each core 408-412 (Core 1-Core X) of a multi-core processor is associated with tiers of a cache (L1-L3).
• the LI and L2 caches may be dedicated to each core.
• X is equal to the number of cores and is equal to or greater than the number of STT-MRAM types that may be utilized for the cache of each core.
• Y is equal to the number to STT-MRAM cache types and Y is equal to or less than X.
• each lower level STT-MRAM cache 402-406 may be tuned according to specific attributes (Type 1-Type Y).
• the Type 1 STT-MRAM cache 402 may be tuned with a first set of attributes for fast access (e.g., increased performance). That is, the Type 1 STT-MRAM cache 402 may be tuned to include a short latency and a fast cycle time.
• the size of the Type 1 STT-MRAM cache 402 cell may also be tuned to be greater (e.g., 2 transistors (T)/l MTJ) than a typical cell size of other types of an STT-MRAM cache.
• the size of the MTJ of the Type 1 STT-MRAM cache 402 may be smaller than the size of the MTJ of other types of STT- MRAM cache.
• the decreased size of the MTJ may increase the speed of a STT-MRAM cache due to an increased current density.
• the Type Y STT-MRAM cache 404 may be tuned with a second set of attributes.
• the Type Y STT-MRAM cache 404 may be tuned for a higher density. That is, the Type Y STT-MRAM cache 404 may be tuned to have a density that is greater than a density of other types of STT-MRAM cache.
• the Type 1 STT-MRAM cache 402 may have a density of 8 megabits (Mb) and the Type Y STT- MRAM cache 404 may have a density of 16-32 Mb.
• Mb 8 megabits
• the increased density of the Type Y STT-MRAM cache 404 increases the cache size, thereby, increasing a cache hit rate and decreasing a cache miss rate.
• the cell size of the Type Y STT-MRAM cache 404 may be less than the size of other types of STT-MRAM.
• the cell size of the Type Y STT-MRAM cache 404 may be 1 transistor- 1 MTJ (1T- 1J). Because the Type Y STT-MRAM cache 404 has an increased density, the Type Y STT-MRAM cache 404 may store data such as programming code for power-down and instant-on (e.g., portions of the operating system (OS)).
• OS operating system
• the attributes may be similar for all or some STT-MRAM cache types.
• the Type 2 STT-MRAM cache 406 may have the same attributes as the Type 1 STT-MRAM cache 402 or the Type Y STT-MRAM cache 404.
• the attributes may be different for each type of STT-MRAM.
• each of the attributes of the Type 1 STT-MRAM cache 402, Type 2 STT-MRAM cache 406, and Type Y STT-MRAM cache 404 may be different.
• the tuning of the types of STT-MRAM cache is not limited to the configurations described for FIGURE 4. That is, the STT-MRAM may be tuned for various configurations, such as, for example, security, performance, reliability, data type, power, usage, density, and other configurations associated with various memory types.
• FIGURE 5 illustrates a multi-tiered multi-core processor 500 utilizing STT- MRAM cache according to another aspect of the present disclosure. As illustrated in
• each core 506-510 (Core 1-Core X) of a multi-core processor is associated with a high level cache (LI) and mid-level cache (L2), furthermore, a lower level cache (L3) is shared between the cores.
• LI high level cache
• L2 mid-level cache
• L3 lower level cache
• X is equal to the number of cores and the STT-MRAM cache may be shared between the cores when X is greater than two.
• F is equal to the number to STT-MRAM cache types and F is equal to or less than X.
• each lower level STT-MRAM cache may be tuned according to specific attributes (Type 1-Type F).
• the Type 1 STT- MRAM cache 502 may be a shared cache for Core 1 506 and Core 2 510 and the Type F STT-MRAM cache 504 may be a shared cache for all other cores.
• the Type 1 STT-MRAM cache 502 may be tuned according to a first set of attributes and the Type F STT-MRAM cache 504 may be tuned according to a second set of attributes.
• the sharing of the STT-MRAM cache is not limited to the aspect illustrated in FIGURE 5, the STT-MRAM cache may be shared by more than two cores and some cores may not share an STT-MRAM cache in the shared configuration. That is, some types of STT-MRAM cache may only be associated with one core while other types of STT-MRAM cache may be shared amongst cores.
• the types of STT-MRAM may be tuned according to specific attributes.
• the attributes may include latency, cache hit/miss rate, energy, energy- delay product (e.g., combination of performance and energy consumption), data utilization (e.g., duty factor: read, write, or hold/store), and data reliability and security (e.g., strong data retention and protection).
• the aforementioned attributes are examples of various attributes that may be configured to tune the types of STT-MRAM cache, the present disclosure is not limited to the aforementioned attributes and is contemplated for other attributes. It should be noted that in the present disclosure attributes include parameters.
• each STT-MRAM cache may be tailored for performance (e.g., latency, cache hit/miss rate), data type usage, reliability, power, and density.
• the STT-MRAM cache utilizes a homogeneous lower level process integration resulting in no extra process overhead because the different tiers of the lower level cache are simultaneously fabricated. That is, the same fabrication processes may be used for each type of STT-MRAM cache with only slight variation in the overall process flow. For example, different mask layouts may be used within the same process flow to create the differently tuned caches.
• the use of an STT-MRAM cache may decrease chip costs. That is, the STT-MRAM cache may use less space in comparison to the SRAM cache. For example, according to some aspects, the STT-MRAM cache may save approximately two to three times the area in comparison to an SRAM lower level cache of the same density.
• FIGURE 6 illustrates a block diagram of a method for associating and tuning STT-MRAM caches 600.
• a first STT- MRAM cache is associated with a first core.
• the first STT-MRAM is tuned according to first attributes.
• a second STT- MRAM cache is associated with a second core.
• the second STT-MRAM is tuned according to second attributes, as shown in block 608.
• the method may also include associating at least one other core with the first STT-MRAM cache or the second STT- MRAM, as shown in block 610.
• FIGURE 7 shows an exemplary wireless communication system 700 in which an embodiment of the disclosure may be advantageously employed.
• FIGURE 7 shows three remote units 720, 730, and 750 and two base stations 740. It will be recognized that wireless communication systems may have many more remote units and base stations.
• Remote units 720, 730, and 750 include multi-core processors with lower level STT-MRAM cache 725A, 725B, 725C.
• FIGURE 7 shows forward link signals 770 from the base stations 740 and the remote units 720, 730, and 750 and reverse link signals 710 from the remote units 720, 730, and 750 to base stations 740.
• the remote unit 720 is shown as a mobile telephone
• remote unit 730 is shown as a portable computer
• remote unit 750 is shown as a fixed location remote unit in a wireless local loop system.
• the remote units may be cell phones, hand-held personal communication systems (PCS) units, a set top box, a music player, a video player, an entertainment unit, a navigation device, portable data units, such as personal data assistants, or fixed location data units such as meter reading equipment.
• PCS personal communication systems
• FIGURE 7 illustrates remote units, which may employ multi-core processors with lower level STT-MRAM cache 725A, 725B, 725C according to the teachings of the disclosure, the disclosure is not limited to these exemplary illustrated units.
• multi-core processors with lower level STT-MRAM cache according to aspects of the present disclosure may be suitably employed in any device.
• FIGURE 8 is a block diagram illustrating a design workstation used for circuit, layout, and logic design of a semiconductor component, such as the multi-core processor with lower level STT-MRAM cache disclosed above.
• a design workstation 800 includes a hard disk 801 containing operating system software, support files, and design software such as Cadence or OrCAD.
• the design workstation 800 also includes a display 802 to facilitate design of a circuit 810 or a semiconductor component 812 such as the multi- core processor with lower level STT-MRAM.
• a storage medium 804 is provided for tangibly storing the circuit design 810 or the semiconductor component 812.
• the circuit design 810 or the semiconductor component 812 may be stored on the storage medium 804 in a file format such as GDSII or GERBER.
• the storage medium 804 may be a CD-ROM, DVD, hard disk, flash memory, or other appropriate device. Furthermore, the design workstation 800 includes a drive apparatus 803 for accepting input from or writing output to the storage medium 804. [0039] Data recorded on the storage medium 804 may specify logic circuit configurations, pattern data for photolithography masks, or mask pattern data for serial write tools such as electron beam lithography. The data may further include logic verification data such as timing diagrams or net circuits associated with logic simulations. Providing data on the storage medium 804 facilitates the design of the circuit design 810 or the semiconductor component 812 by decreasing the number of processes for designing semiconductor wafers.
• the memory apparatus includes a first storage means associated with a first core of the multi-core processor and tuned according to first attributes.
• the memory apparatus also includes a second storage means associated with a second core of the multi-core processor and tuned according to second attributes.
• the storage means may be a lower level STT-MRAM cache 402-406 502-504, a level two (L2) cache, and/or other memory types configured to perform the functions recited by the identifying means.
• the methodologies described herein may be implemented by various means depending upon the application. For example, these methodologies may be implemented in hardware, firmware, software, or any combination thereof.
• the processing units may be implemented within one or more application specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), processors, controllers, micro-controllers, microprocessors, electronic devices, other electronic units designed to perform the functions described herein, or a combination thereof.
• ASICs application specific integrated circuits
• DSPs digital signal processors
• DSPDs digital signal processing devices
• PLDs programmable logic devices
• FPGAs field programmable gate arrays
• processors controllers, micro-controllers, microprocessors, electronic devices, other electronic units designed to perform the functions described herein, or a combination thereof.
• the methodologies may be implemented with modules (e.g., procedures, functions, and so on) that perform the functions described herein.
• Any machine or computer readable medium tangibly embodying instructions may be used in implementing the methodologies described herein.
• software code may be stored in a memory and executed by a processor. When executed by the processor, the executing software code generates the operational environment that implements the various methodologies and functionalities of the different aspects of the teachings presented herein.
• Memory may be implemented within the processor or external to the processor.
• the term "memory" refers to any type of long term, short term, volatile, nonvolatile, or other memory and is not to be limited to any particular type of memory or number of memories, or type of media upon which memory is stored.
• the machine or computer readable medium that stores the software code defining the methodologies and functions described herein includes physical computer storage media.
• a storage medium may be any available medium that can be accessed by a computer.
• such computer-readable media can comprise RAM, ROM, EEPROM, CD-ROM or other optical disk storage, magnetic disk storage or other magnetic storage devices, or any other medium that can be used to store desired program code in the form of instructions or data structures and that can be accessed by a computer.
• disk and/or disc includes compact disc (CD), laser disc, optical disc, digital versatile disc (DVD), floppy disk and blu-ray disc where disks usually reproduce data magnetically, while discs reproduce data optically with lasers. Combinations of the above should also be included within the scope of computer readable media.
• instructions and/or data may be provided as signals on transmission media included in a communication apparatus.
• a communication apparatus may include a transceiver having signals indicative of instructions and data. The instructions and data are configured to cause one or more processors to implement the functions outlined in the claims.

Landscapes

• Engineering & Computer Science (AREA)
• Theoretical Computer Science (AREA)
• General Engineering & Computer Science (AREA)
• Physics & Mathematics (AREA)
• General Physics & Mathematics (AREA)
• Computer Hardware Design (AREA)
• Computing Systems (AREA)
• Microelectronics & Electronic Packaging (AREA)
• Memory System Of A Hierarchy Structure (AREA)
• Mram Or Spin Memory Techniques (AREA)

Priority Applications (5)

Applications Claiming Priority (2)

Publications (1)

Family

ID=49004012

Family Applications (1)

Country Status (7)

Families Citing this family (12)

Family Cites Families (14)

• 2012
  • 2012-08-10 US US13/571,426 patent/US9244853B2/en active Active
• 2013
  • 2013-08-07 EP EP13751006.1A patent/EP2883151B1/en active Active
  • 2013-08-07 CN CN201380042021.3A patent/CN104520838B/zh active Active
  • 2013-08-07 JP JP2015526678A patent/JP6196305B2/ja not_active Expired - Fee Related
  • 2013-08-07 IN IN76MUN2015 patent/IN2015MN00076A/en unknown
  • 2013-08-07 KR KR20157005965A patent/KR20150041092A/ko not_active Withdrawn
  • 2013-08-07 WO PCT/US2013/054004 patent/WO2014025920A1/en not_active Ceased

Non-Patent Citations (3)

Also Published As

Similar Documents

Legal Events