Improving direct-mapped cache performance by the addition of a small fully-associative cache and prefetch buffers
01 May 1990-Vol. 18, pp 364-373
TL;DR: In this article, a hardware technique to improve the performance of caches is presented, where a small fully-associative cache between a cache and its refill path is used to place prefetched data and not in the cache.
Abstract: Projections of computer technology forecast processors with peak performance of 1,000 MIPS in the relatively near future. These processors could easily lose half or more of their performance in the memory hierarchy if the hierarchy design is based on conventional caching techniques. This paper presents hardware techniques to improve the performance of caches.Miss caching places a small fully-associative cache between a cache and its refill path. Misses in the cache that hit in the miss cache have only a one cycle miss penalty, as opposed to a many cycle miss penalty without the miss cache. Small miss caches of 2 to 5 entries are shown to be very effective in removing mapping conflict misses in first-level direct-mapped caches.Victim caching is an improvement to miss caching that loads the small fully-associative cache with the victim of a miss and not the requested line. Small victim caches of 1 to 5 entries are even more effective at removing conflict misses than miss caching.Stream buffers prefetch cache lines starting at a cache miss address. The prefetched data is placed in the buffer and not in the cache. Stream buffers are useful in removing capacity and compulsory cache misses, as well as some instruction cache conflict misses. Stream buffers are more effective than previously investigated prefetch techniques at using the next slower level in the memory hierarchy when it is pipelined. An extension to the basic stream buffer, called multi-way stream buffers, is introduced. Multi-way stream buffers are useful for prefetching along multiple intertwined data reference streams.Together, victim caches and stream buffers reduce the miss rate of the first level in the cache hierarchy by a factor of two to three on a set of six large benchmarks.
Citations
More filters
Patent•
19 Nov 1999TL;DR: Curious caching as mentioned in this paper improves upon cache snooping by allowing a cache to insert data from snooped bus operations that is not currently in the cache and independent of any prior accesses to the associated memory location.
Abstract: Curious caching improves upon cache snooping by allowing a snooping cache to insert data from snooped bus operations that is not currently in the cache and independent of any prior accesses to the associated memory location. In addition, curious caching allows software to specify which data producing bus operations, e.g., reads and writes, result in data being inserted into the cache. This is implemented by specifying “memory regions of curiosity” and insertion and replacement policy actions for those regions. In column caching, the replacement of data can be restricted to particular regions of the cache. By also making the replacement address-dependent, column caching allows different regions of memory to be mapped to different regions of the cache. In a set-associative cache, a replacement policy specifies the particular column(s) of the set-associative cache in which a page of data can be stored. The column specification is made in page table entries in a TLB that translates between virtual and physical addresses. The TLB includes a bit vector, one bit per column, which indicates the columns of the cache that are available for replacement.
100 citations
TL;DR: CirCNN as discussed by the authors utilizes the Fast Fourier Transform (FFT)-based fast multiplication, simultaneously reducing the computational complexity (both in inference and training) and the storage complexity from O(n2) to O(nlogn) with negligible accuracy loss.
Abstract: Large-scale deep neural networks (DNNs) are both compute and memory intensive. As the size of DNNs continues to grow, it is critical to improve the energy efficiency and performance while maintaining accuracy. For DNNs, the model size is an important factor affecting performance, scalability and energy efficiency. Weight pruning achieves good compression ratios but suffers from three drawbacks: 1) the irregular network structure after pruning; 2) the increased training complexity; and 3) the lack of rigorous guarantee of compression ratio and inference accuracy. To overcome these limitations, this paper proposes CirCNN, a principled approach to represent weights and process neural networks using block-circulant matrices. CirCNN utilizes the Fast Fourier Transform (FFT)-based fast multiplication, simultaneously reducing the computational complexity (both in inference and training) from O(n2) to O(nlogn) and the storage complexity from O(n2) to O(n), with negligible accuracy loss. Compared to other approaches, CirCNN is distinct due to its mathematical rigor: it can converge to the same effectiveness as DNNs without compression. The CirCNN architecture, a universal DNN inference engine that can be implemented on various hardware/software platforms with configurable network architecture. To demonstrate the performance and energy efficiency, we test CirCNN in FPGA, ASIC and embedded processors. Our results show that CirCNN architecture achieves very high energy efficiency and performance with a small hardware footprint. Based on the FPGA implementation and ASIC synthesis results, CirCNN achieves 6-102X energy efficiency improvements compared with the best state-of-the-art results.
98 citations
Patent•
20 Jul 2004TL;DR: In this paper, data are prefetched into a cache from a prefetch region of memory, based on a program instruction reference and on compile-time information that indicates the bounds of the prefetch regions, a size of prefetch block, and a location of the block.
Abstract: Data are prefetched into a cache from a prefetch region of memory, based on a program instruction reference and on compile-time information that indicates the bounds of the prefetch region, a size of a prefetch block, and a location of the prefetch block. If the program reference address lies with the prefetch region, an offset distance is used to determine the address of the prefetch block. Prefetching is performed either from a continuous one-dimensional prefetch region, or an embedded multi-dimensional prefetch region. The prefetch block address is respectively determined in one dimension or multiple dimensions. Program-directed prefetching is implemented by a media processor or by a separate processing component in communication with the media processor. The primary components include a program-directed prefetch controller, a cache, a function unit, and a memory. Preferably, region registers store the compile-time information, and the prefetched data are stored in a cache prefetch buffer.
98 citations
Patent•
23 Dec 1994TL;DR: In this paper, a column-associative cache that reduces conflict misses, increases the hit rate and maintains a minimum hit access time is proposed, where the cache lines represent a column of sets.
Abstract: A column-associative cache that reduces conflict misses, increases the hit rate and maintains a minimum hit access time. The column-associative cache indexes data from a main memory into a plurality of cache lines according to a tag and index field through hash and rehash functions. The cache lines represent a column of sets. Each cache line contains a rehash block indicating whether the set is a rehash location. To increase the performance of the column-associative cache, a content addressable memory (CAM) is used to predict future conflict misses.
98 citations
08 May 2000
TL;DR: This paper proposes a cooperative hardware/software mechanism to reduce memory access latencies for linked data structures by identifying the load instructions that traverse the LDS, and executing them ahead of the actual computation.
Abstract: As the performance gap between the CPU and main memory continues to grow, techniques to hide memory latency are essential to deliver a high performance computer system. Prefetching can often overlap memory latency with computation for array-based numeric applications. However, prefetching for pointer-intensive applications still remains a challenging problem. Prefetching linked data structures (LDS) is difficult because the address sequence of LDS traversal does not present the same arithmetic regularity as array-based applications and the data dependence of pointer dereferences can serialize the address generation process.In this paper, we propose a cooperative hardware/software mechanism to reduce memory access latencies for linked data structures. Instead of relying on the past address history to predict future accesses, we identify the load instructions that traverse the LDS, and execute them ahead of the actual computation. To overcome the serial nature of the LDS address generation, we attach a prefetch controller to each level of the memory hierarchy and push, rather than pull, data to the CPU. Our simulations, using four pointer-intensive applications, show that the push model can achieve between 4% and 30% larger reductions in execution time compared to the pull model.
98 citations
References
More filters
[...]
TL;DR: Specific aspects of cache memories investigated include: the cache fetch algorithm (demand versus prefetch), the placement and replacement algorithms, line size, store-through versus copy-back updating of main memory, cold-start versus warm-start miss ratios, mulhcache consistency, the effect of input /output through the cache, the behavior of split data/instruction caches, and cache size.
Abstract: design issues. Specific aspects of cache memories tha t are investigated include: the cache fetch algorithm (demand versus prefetch), the placement and replacement algorithms, line size, store-through versus copy-back updating of main memory, cold-start versus warm-start miss ratios, mulhcache consistency, the effect of input /output through the cache, the behavior of split data/instruction caches, and cache size. Our discussion includes other aspects of memory system architecture, including translation lookaside buffers. Throughout the paper, we use as examples the implementation of the cache in the Amdahl 470V/6 and 470V/7, the IBM 3081, 3033, and 370/168, and the DEC VAX 11/780. An extensive bibliography is provided.
1,614 citations
01 Jan 1990
TL;DR: This note evaluates several hardware platforms and operating systems using a set of benchmarks that test memory bandwidth and various operating system features such as kernel entry/exit and file systems to conclude that operating system performance does not seem to be improving at the same rate as the base speed of the underlying hardware.
Abstract: This note evaluates several hardware platforms and operating systems using a set of benchmarks that test memory bandwidth and various operating system features such as kernel entry/exit and file systems. The overall conclusion is that operating system performance does not seem to be improving at the same rate as the base speed of the underlying hardware. Copyright 1989 Digital Equipment Corporation d i g i t a l Western Research Laboratory 100 Hamilton Avenue Palo Alto, California 94301 USA
467 citations
01 Apr 1989
TL;DR: A parameterizable code reorganization and simulation system was developed and used to measure instruction-level parallelism and the average degree of superpipelining metric is introduced, suggesting that this metric is already high for many machines.
Abstract: Superscalar machines can issue several instructions per cycle. Superpipelined machines can issue only one instruction per cycle, but they have cycle times shorter than the latency of any functional unit. In this paper these two techniques are shown to be roughly equivalent ways of exploiting instruction-level parallelism. A parameterizable code reorganization and simulation system was developed and used to measure instruction-level parallelism for a series of benchmarks. Results of these simulations in the presence of various compiler optimizations are presented. The average degree of superpipelining metric is introduced. Our simulations suggest that this metric is already high for many machines. These machines already exploit all of the instruction-level parallelism available in many non-numeric applications, even without parallel instruction issue or higher degrees of pipelining.
316 citations
TL;DR: It is shown that prefetching all memory references in very fast computers can increase the effective CPU speed by 10 to 25 percent.
Abstract: Memory transfers due to a cache miss are costly. Prefetching all memory references in very fast computers can increase the effective CPU speed by 10 to 25 percent.
315 citations
17 May 1988
TL;DR: The inclusion property is essential in reducing the cache coherence complexity for multiprocessors with multilevel cache hierarchies and a new inclusion-coherence mechanism for two-level bus-based architectures is proposed.
Abstract: The inclusion property is essential in reducing the cache coherence complexity for multiprocessors with multilevel cache hierarchies. We give some necessary and sufficient conditions for imposing the inclusion property for fully- and set-associative caches which allow different block sizes at different levels of the hierarchy. Three multiprocessor structures with a two-level cache hierarchy (single cache extension, multiport second-level cache, bus-based) are examined. The feasibility of imposing the inclusion property in these structures is discussed. This leads us to propose a new inclusion-coherence mechanism for two-level bus-based architectures.
236 citations