Topic
Memory architecture
About: Memory architecture is a research topic. Over the lifetime, 4804 publications have been published within this topic receiving 102444 citations.
Papers published on a yearly basis
Papers
24 Jun 2017
TL;DR: The Tensor Processing Unit (TPU) as discussed by the authors is a custom ASIC deployed in datacenters since 2015 that accelerates the inference phase of neural networks (NN) using a 65,536 8-bit MAC matrix multiply unit that offers a peak throughput of 92 TeraOps/second (TOPS).
Abstract: Many architects believe that major improvements in cost-energy-performance must now come from domain-specific hardware. This paper evaluates a custom ASIC---called a Tensor Processing Unit (TPU) --- deployed in datacenters since 2015 that accelerates the inference phase of neural networks (NN). The heart of the TPU is a 65,536 8-bit MAC matrix multiply unit that offers a peak throughput of 92 TeraOps/second (TOPS) and a large (28 MiB) software-managed on-chip memory. The TPU's deterministic execution model is a better match to the 99th-percentile response-time requirement of our NN applications than are the time-varying optimizations of CPUs and GPUs that help average throughput more than guaranteed latency. The lack of such features helps explain why, despite having myriad MACs and a big memory, the TPU is relatively small and low power. We compare the TPU to a server-class Intel Haswell CPU and an Nvidia K80 GPU, which are contemporaries deployed in the same datacenters. Our workload, written in the high-level TensorFlow framework, uses production NN applications (MLPs, CNNs, and LSTMs) that represent 95% of our datacenters' NN inference demand. Despite low utilization for some applications, the TPU is on average about 15X -- 30X faster than its contemporary GPU or CPU, with TOPS/Watt about 30X -- 80X higher. Moreover, using the CPU's GDDR5 memory in the TPU would triple achieved TOPS and raise TOPS/Watt to nearly 70X the GPU and 200X the CPU.
3,848 citations
18 Jun 2016
TL;DR: This work proposes a novel PIM architecture, called PRIME, to accelerate NN applications in ReRAM based main memory, and distinguishes itself from prior work on NN acceleration, with significant performance improvement and energy saving.
Abstract: Processing-in-memory (PIM) is a promising solution to address the "memory wall" challenges for future computer systems. Prior proposed PIM architectures put additional computation logic in or near memory. The emerging metal-oxide resistive random access memory (ReRAM) has showed its potential to be used for main memory. Moreover, with its crossbar array structure, ReRAM can perform matrix-vector multiplication efficiently, and has been widely studied to accelerate neural network (NN) applications. In this work, we propose a novel PIM architecture, called PRIME, to accelerate NN applications in ReRAM based main memory. In PRIME, a portion of ReRAM crossbar arrays can be configured as accelerators for NN applications or as normal memory for a larger memory space. We provide microarchitecture and circuit designs to enable the morphable functions with an insignificant area overhead. We also design a software/hardware interface for software developers to implement various NNs on PRIME. Benefiting from both the PIM architecture and the efficiency of using ReRAM for NN computation, PRIME distinguishes itself from prior work on NN acceleration, with significant performance improvement and energy saving. Our experimental results show that, compared with a state-of-the-art neural processing unit design, PRIME improves the performance by ~2360× and the energy consumption by ~895×, across the evaluated machine learning benchmarks.
1,500 citations
5 Nov 2016
TL;DR: This paper propose a machine reading simulator which processes text incrementally from left to right and performs shallow reasoning with memory and attention, enabling adaptive memory usage during recurrence with neural attention, offering a way to weakly induce relations among tokens.
Abstract: In this paper we address the question of how to render sequence-level networks better at handling structured input. We propose a machine reading simulator which processes text incrementally from left to right and performs shallow reasoning with memory and attention. The reader extends the Long Short-Term Memory architecture with a memory network in place of a single memory cell. This enables adaptive memory usage during recurrence with neural attention, offering a way to weakly induce relations among tokens. The system is initially designed to process a single sequence but we also demonstrate how to integrate it with an encoder-decoder architecture. Experiments on language modeling, sentiment analysis, and natural language inference show that our model matches or outperforms the state of the art.
1,275 citations
TL;DR: The directory architecture for shared memory (Dash) as discussed by the authors allows shared data to be cached, significantly reducing the latency of memory accesses and yielding higher processor utilization and higher overall performance, and a distributed directory-based protocol that provides cache coherence without compromising scalability.
Abstract: The overall goals and major features of the directory architecture for shared memory (Dash) are presented. The fundamental premise behind the architecture is that it is possible to build a scalable high-performance machine with a single address space and coherent caches. The Dash architecture is scalable in that it achieves linear or near-linear performance growth as the number of processors increases from a few to a few thousand. This performance results from distributing the memory among processing nodes and using a network with scalable bandwidth to connect the nodes. The architecture allows shared data to be cached, significantly reducing the latency of memory accesses and yielding higher processor utilization and higher overall performance. A distributed directory-based protocol that provides cache coherence without compromising scalability is discussed in detail. The Dash prototype machine and the corresponding software support are described. >
1,060 citations
TL;DR: The process of validating DRAMSim2 timing against manufacturer Verilog models in an effort to prove the accuracy of simulation results is described.
Abstract: In this paper we present DRAMSim2, a cycle accurate memory system simulator. The goal of DRAMSim2 is to be an accurate and publicly available DDR2/3 memory system model which can be used in both full system and trace-based simulations. We describe the process of validating DRAMSim2 timing against manufacturer Verilog models in an effort to prove the accuracy of simulation results. We outline the combination of DRAMSim2 with a cycle-accurate x86 simulator that can be used to perform full system simulations. Finally, we discuss DRAMVis, a visualization tool that can be used to graph and compare the results of DRAMSim2 simulations.
983 citations
Related Topics (5)
Performance Metrics
| Year | Papers |
|---|---|
| 2024 | 2 |
| 2023 | 4 |
| 2022 | 14 |
| 2021 | 118 |
| 2020 | 148 |
| 2019 | 171 |