ReD-LUT: Reconfigurable In-DRAM LUTs Enabling Massive Parallel Computation

ReD-LUT: Reconfigurable In-DRAM LUTs Enabling Massive Parallel Computation
复制标题

ReD-LUT:可重新配置的 DRAM 内 LUT,支持大规模并行计算

DOI:
10.1145/3508352.3549469
复制
发表时间:
2022
期刊:
2022 IEEE/ACM International Conference On Computer Aided Design (ICCAD)
影响因子:
--
通讯作者:
Shaahin Angizi
Shaahin Angizi
中科院分区:
--
文献类型:
--
作者:
Ranyang Zhou;A. Roohi;Durga Misra;Shaahin Angizi

文献摘要

参考文献

被引文献

相似文献

在本文中,我们提出了一种名为ReD-LUT的可重构dram处理架构,利用商品主存储器的高密度来实现灵活、通用和大规模并行计算。ReD-LUT支持查找表(LUT)查询,仅通过内存读取操作有效地执行复杂的算术运算(例如,乘法,除法等)。此外,ReD-LUT通过提升DRAM子阵列的模拟操作来实现存储在同一位线上的操作数之间的布尔函数,从而实现内存中按位的批量逻辑,超出了先前基于DRAM的建议的范围。我们探讨了ReD-LUT在两个计算密集型应用中的有效性,即低精度深度学习加速和高级加密标准(AES)计算。我们的电路到架构仿真结果表明,对于量化的深度学习工作负载,与GPU相比,ReD-LUT将每个图像的能耗降低了21.4倍,并且比最佳的dram位加速器实现了~37.8倍的加速和2.1倍的能效。对于AES数据加密,与ASIC实现相比,其能耗降低了约2.2倍。
In this paper, we propose a reconfigurable processing-in-DRAM architecture named ReD-LUT leveraging the high density of commodity main memory to enable a flexible, general-purpose, and massively parallel computation. ReD-LUT supports lookup table (LUT) queries to efficiently execute complex arithmetic operations (e.g., multiplication, division, etc.) via only memory read operation. In addition, ReD-LUT enables bulk bit-wise in-memory logic by elevating the analog operation of the DRAM sub-array to implement Boolean functions between operands stored in the same bit-line beyond the scope of prior DRAM-based proposals. We explore the efficacy of ReD-LUT in two computationally-intensive applications, i.e., low-precision deep learning acceleration, and the Advanced Encryption Standard (AES) computation. Our circuit-to-architecture simulation results show that for a quantized deep learning workload, ReD-LUT reduces the energy consumption per image by a factor of 21.4× compared with the GPU and achieves ~37.8× speedup and 2.1× energy-efficiency over the best in-DRAM bit-wise accelerators. As for AES data-encryption, it reduces energy consumption by a factor of ~2.2× compared to an ASIC implementation.
FlexiDRAM:一种灵活的 DRAM 框架,可实现并行通用计算
DOI: 10.1145/3531437.3539721
发表时间: 2022
期刊: 2022.
影响因子: --
作者:
Zhou, R;Roohi, A;Misra, D;Angizi, S
通讯作者: Angizi, S
DOI: 10.1109/iccad45719.2019.8942101
发表时间: 2019-11
期刊: 2019 IEEE/ACM International Conference on Computer-Aided Design (ICCAD)
影响因子: --
作者:
Shaahin Angizi;Deliang Fan
通讯作者: Shaahin Angizi;Deliang Fan
DOI: 10.1145/3299874.3317984
发表时间: 2019-05
期刊: Proceedings of the 2019 Great Lakes Symposium on VLSI
影响因子: --
作者:
Shaahin Angizi;Deliang Fan
通讯作者: Shaahin Angizi;Deliang Fan