MATSA: An MRAM-Based Energy-Efficient Accelerator for Time Series Analysis
<italic>Time Series Analysis</italic> (<italic>TSA</italic>) is a critical workload to extract valuable information from collections of sequential data, e.g., detecting anomalies in electrocardiograms. Subsequence Dynamic Time Warping (sDTW) is the state-of-the-art algorithm...
| Main Authors: | , , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2024-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/10458946/ |
| _version_ | 1797243376058236928 |
|---|---|
| author | Ivan Fernandez Christina Giannoula Aditya Manglik Ricardo Quislant Nika Mansouri Ghiasi Juan Gomez-Luna Eladio Gutierrez Oscar Plata Onur Mutlu |
| author_facet | Ivan Fernandez Christina Giannoula Aditya Manglik Ricardo Quislant Nika Mansouri Ghiasi Juan Gomez-Luna Eladio Gutierrez Oscar Plata Onur Mutlu |
| author_sort | Ivan Fernandez |
| collection | DOAJ |
| description | <italic>Time Series Analysis</italic> (<italic>TSA</italic>) is a critical workload to extract valuable information from collections of sequential data, e.g., detecting anomalies in electrocardiograms. Subsequence Dynamic Time Warping (sDTW) is the state-of-the-art algorithm for high-accuracy TSA. We find that the performance and energy efficiency of sDTW on conventional CPU and GPU platforms are heavily burdened by the latency and energy overheads of data movement between the compute and the memory units. sDTW exhibits low arithmetic intensity and low data reuse on conventional platforms, stemming from poor amortization of the data movement overheads. To improve the performance and energy efficiency of the sDTW algorithm, we propose MATSA, the first <underline>M</underline>agnetoresistive RAM (MRAM)-based <underline>A</underline>ccelerator for <underline>TSA</underline>. MATSA leverages Processing-Using-Memory (PUM) based on MRAM crossbars to minimize data movement overheads and exploit parallelism in sDTW. MATSA improves performance by <inline-formula> <tex-math notation="LaTeX">$7.35\times /6.15\times /6.31\times $ </tex-math></inline-formula> and energy efficiency by <inline-formula> <tex-math notation="LaTeX">$11.29\times /4.21\times /2.65\times $ </tex-math></inline-formula> over server-class CPU, GPU, and Processing-Near-Memory platforms, respectively. |
| first_indexed | 2024-04-24T18:54:08Z |
| format | Article |
| id | doaj.art-97c47f1741df4ecf9342ec359e741af0 |
| institution | Directory Open Access Journal |
| issn | 2169-3536 |
| language | English |
| last_indexed | 2024-04-24T18:54:08Z |
| publishDate | 2024-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Access |
| spelling | doaj.art-97c47f1741df4ecf9342ec359e741af02024-03-26T17:45:18ZengIEEEIEEE Access2169-35362024-01-0112367273674210.1109/ACCESS.2024.337331110458946MATSA: An MRAM-Based Energy-Efficient Accelerator for Time Series AnalysisIvan Fernandez0https://orcid.org/0000-0001-6133-5670Christina Giannoula1https://orcid.org/0000-0003-0162-4547Aditya Manglik2https://orcid.org/0000-0003-4189-8761Ricardo Quislant3https://orcid.org/0000-0002-4705-7042Nika Mansouri Ghiasi4https://orcid.org/0000-0002-0833-0042Juan Gomez-Luna5https://orcid.org/0000-0002-6514-1571Eladio Gutierrez6https://orcid.org/0000-0001-9748-9161Oscar Plata7https://orcid.org/0000-0003-2233-0011Onur Mutlu8https://orcid.org/0000-0002-0075-2312Department of Computer Architecture, University of Málaga, Málaga, SpainDepartment of Information Technology and Electrical Engineering (D-ITET), ETH Zürich, Zürich, SwitzerlandDepartment of Information Technology and Electrical Engineering (D-ITET), ETH Zürich, Zürich, SwitzerlandDepartment of Computer Architecture, University of Málaga, Málaga, SpainDepartment of Information Technology and Electrical Engineering (D-ITET), ETH Zürich, Zürich, SwitzerlandDepartment of Information Technology and Electrical Engineering (D-ITET), ETH Zürich, Zürich, SwitzerlandDepartment of Computer Architecture, University of Málaga, Málaga, SpainDepartment of Computer Architecture, University of Málaga, Málaga, SpainDepartment of Information Technology and Electrical Engineering (D-ITET), ETH Zürich, Zürich, Switzerland<italic>Time Series Analysis</italic> (<italic>TSA</italic>) is a critical workload to extract valuable information from collections of sequential data, e.g., detecting anomalies in electrocardiograms. Subsequence Dynamic Time Warping (sDTW) is the state-of-the-art algorithm for high-accuracy TSA. We find that the performance and energy efficiency of sDTW on conventional CPU and GPU platforms are heavily burdened by the latency and energy overheads of data movement between the compute and the memory units. sDTW exhibits low arithmetic intensity and low data reuse on conventional platforms, stemming from poor amortization of the data movement overheads. To improve the performance and energy efficiency of the sDTW algorithm, we propose MATSA, the first <underline>M</underline>agnetoresistive RAM (MRAM)-based <underline>A</underline>ccelerator for <underline>TSA</underline>. MATSA leverages Processing-Using-Memory (PUM) based on MRAM crossbars to minimize data movement overheads and exploit parallelism in sDTW. MATSA improves performance by <inline-formula> <tex-math notation="LaTeX">$7.35\times /6.15\times /6.31\times $ </tex-math></inline-formula> and energy efficiency by <inline-formula> <tex-math notation="LaTeX">$11.29\times /4.21\times /2.65\times $ </tex-math></inline-formula> over server-class CPU, GPU, and Processing-Near-Memory platforms, respectively.https://ieeexplore.ieee.org/document/10458946/Time series analysisprocessing-using-memorymemory-boundemerging technologies |
| spellingShingle | Ivan Fernandez Christina Giannoula Aditya Manglik Ricardo Quislant Nika Mansouri Ghiasi Juan Gomez-Luna Eladio Gutierrez Oscar Plata Onur Mutlu MATSA: An MRAM-Based Energy-Efficient Accelerator for Time Series Analysis IEEE Access Time series analysis processing-using-memory memory-bound emerging technologies |
| title | MATSA: An MRAM-Based Energy-Efficient Accelerator for Time Series Analysis |
| title_full | MATSA: An MRAM-Based Energy-Efficient Accelerator for Time Series Analysis |
| title_fullStr | MATSA: An MRAM-Based Energy-Efficient Accelerator for Time Series Analysis |
| title_full_unstemmed | MATSA: An MRAM-Based Energy-Efficient Accelerator for Time Series Analysis |
| title_short | MATSA: An MRAM-Based Energy-Efficient Accelerator for Time Series Analysis |
| title_sort | matsa an mram based energy efficient accelerator for time series analysis |
| topic | Time series analysis processing-using-memory memory-bound emerging technologies |
| url | https://ieeexplore.ieee.org/document/10458946/ |
| work_keys_str_mv | AT ivanfernandez matsaanmrambasedenergyefficientacceleratorfortimeseriesanalysis AT christinagiannoula matsaanmrambasedenergyefficientacceleratorfortimeseriesanalysis AT adityamanglik matsaanmrambasedenergyefficientacceleratorfortimeseriesanalysis AT ricardoquislant matsaanmrambasedenergyefficientacceleratorfortimeseriesanalysis AT nikamansourighiasi matsaanmrambasedenergyefficientacceleratorfortimeseriesanalysis AT juangomezluna matsaanmrambasedenergyefficientacceleratorfortimeseriesanalysis AT eladiogutierrez matsaanmrambasedenergyefficientacceleratorfortimeseriesanalysis AT oscarplata matsaanmrambasedenergyefficientacceleratorfortimeseriesanalysis AT onurmutlu matsaanmrambasedenergyefficientacceleratorfortimeseriesanalysis |