Name
Abstract | ||
|---|---|---|
ABSTRACTContent-addressable memories (CAMs) can perform fast and energy-efficient search operations. Recently, ternary CAMs (TCAMs) have been utilized to measure Hamming distance for machine learning applications, where they offer significant energy savings and speed-ups. However, the binary precision of the Hamming distance can lead to severe degradation in application-level accuracies, thus mitigating the impact of gains with respect to other figures of merit. To enhance accuracy, multi-bit CAMs (MCAMs) have been proposed that offer higher density and energy savings than TCAMs by storing multiple bits in each cell. However, existing MCAMs are based on emerging nonvolatile memory technologies that are yet to be established. To this end, we propose a fast and extremely energy-efficient MCAM based on mature and widely used flash cells, called E2-MCAM. E2-MCAM can measure the Euclidean squared distance between search queries and data stored in the MCAM "in-memory", and in a single cycle. We evaluate E2-MCAM using an experimentally calibrated flash model in HSPICE with 3-bit precision for proof of concept demonstration. A 64X32 E2-MCAM array achieves a 0.34 fJ energy per bit per search and a 2.7 ns latency while operating at a 770 μW power. Fast and efficient hardware support for Euclidean squared distance is highly valuable as it is widely used in a plethora of machine learning applications. As an example, we show that E2-MCAM achieves accuracies comparable to floating-point GPU implementations with only 3-bit precision for few-shot learning tasks with the ImageNet dataset while offering improvements in energy and latency. |
Year | DOI | Venue |
|---|---|---|
2021 | 10.1109/ISLPED52811.2021.9502488 | ISLPED |
Keywords | DocType | ISSN |
2 -MCAM array,flash cells,extremely energy-efficient MCAM,fast energy-efficient MCAM,nonvolatile memory technologies,multibit CAMs,application-level accuracies,Hamming distance,ternary CAMs,energy-efficient search operations,flash-based multibit content-addressable memory,3-bit precision,machine learning applications,euclidean squared distance,time 2.7 ns,power 770.0 muW | Conference | 1533-4678 |
ISBN | Citations | PageRank |
978-1-6654-3923-7 | 0 | 0.34 |
References | Authors | |
0 | 6 | |
Name | Order | Citations | PageRank |
|---|---|---|---|
| Arman Kazemi | 1 | 1 | 2.06 |
| Shubham Sahay | 2 | 0 | 0.34 |
| Ayush Saxena | 3 | 0 | 0.34 |
| Mohammad Mehdi Sharifi | 4 | 4 | 2.48 |
| Michael Niemier | 5 | 191 | 31.85 |
| Xiaobo Sharon Hu | 6 | 2004 | 208.24 |