Parallel Computing Research Laboratory & NVIDIA > 자유게시판

• 목록
• 답변
• 글쓰기
• 게시판 리스트 옵션
  • 수정
  • 삭제

Scratchpad memory (SPM), also known as scratchpad, scratchpad RAM or local retailer in laptop terminology, is an internal memory, often excessive-velocity, used for temporary storage of calculations, knowledge, and different work in progress. In reference to a microprocessor (or CPU), scratchpad refers to a special excessive-speed memory used to carry small objects of knowledge for fast retrieval. It's much like the utilization and size of a scratchpad in life: a pad of paper for preliminary notes or sketches or writings, and many others. When the scratchpad is a hidden portion of the main memory then it's sometimes referred to as bump storage. L1 cache in that it's the subsequent closest memory to the ALU after the processor MemoryWave Guide registers, with express directions to move knowledge to and from primary memory, usually using DMA-based data switch. In contrast to a system that makes use of caches, a system with scratchpads is a system with non-uniform memory access (NUMA) latencies, as a result of the memory access latencies to the different scratchpads and the main memory range.

One other distinction from a system that employs caches is that a scratchpad generally does not include a copy of data that can be stored in the main memory. Scratchpads are employed for simplification of caching logic, and to guarantee a unit can work without major memory contention in a system using multiple processors, particularly in multiprocessor system-on-chip for embedded systems. They're principally suited for storing temporary results (as it can be found in the CPU stack) that sometimes wouldn't must always be committing to the primary memory; nevertheless when fed by DMA, they will also be used in place of a cache for mirroring the state of slower important memory. The identical issues of locality of reference apply in relation to effectivity of use; although some methods enable strided DMA to entry rectangular data units. One other difference is that scratchpads are explicitly manipulated by purposes. They could also be helpful for realtime applications, where predictable timing is hindered by cache conduct.

Scratchpads aren't used in mainstream desktop processors the place generality is required for legacy software program to run from generation to generation, wherein the available on-chip memory dimension could change. They are higher carried out in embedded methods, particular-purpose processors and game consoles, where chips are often manufactured as MPSoC, and where software is usually tuned to 1 hardware configuration. Fairchild F8 of 1975 contained 64 bytes of scratchpad. Cyrix 6x86 is the only x86-suitable desktop processor to incorporate a dedicated scratchpad. SuperH, used in Sega's consoles, might lock cachelines to an address outdoors of foremost memory for use as a scratchpad. Sony's PS1's R3000 had a scratchpad as a substitute of an L1 cache. It was doable to place the CPU stack right here, an instance of the non permanent workspace usage. Adapteva's Epiphany parallel coprocessor options native-shops for every core, related by a network on a chip, with DMA possible between them and off-chip links (probably to DRAM).

The structure is much like Sony's Cell, except all cores can directly handle one another's scratchpads, producing community messages from commonplace load/store directions. Sony's PS2 Emotion Engine features a sixteen KB scratchpad, to and from which DMA transfers might be issued to its GS, and predominant memory. Cell's SPEs are restricted purely to working of their "native-store", relying on DMA for transfers from/to fundamental memory and between native stores, very like a scratchpad. On this regard, further profit is derived from the lack of hardware to test and update coherence between multiple caches: MemoryWave Guide the design takes advantage of the assumption that every processor's workspace is separate and non-public. It is predicted this benefit will become more noticeable as the variety of processors scales into the "many-core" future. Yet because of the elimination of some hardware logics, the information and instructions of purposes on SPEs must be managed by way of software if the whole process on SPE can't fit in local retailer.