Efficiency near the edge: increasing the energy efficiency of FFTs on GPUs for real-time edge computing
The Square Kilometer Array (SKA) is an international initiative for developing the world's largest radio telescope with a total collecting area of over a million square meters. The scale of the operation, combined with the remote location of the telescope, requires the use of energy-efficient c...
| المؤلفون الرئيسيون: | , , , |
|---|---|
| التنسيق: | Journal article |
| اللغة: | English |
| منشور في: |
Institute of Electrical and Electronics Engineers
2021
|
| _version_ | 1826297666185723904 |
|---|---|
| author | Adamek, K Novotny, J Thiyagalingam, J Armour, WG |
| author_facet | Adamek, K Novotny, J Thiyagalingam, J Armour, WG |
| author_sort | Adamek, K |
| collection | OXFORD |
| description | The Square Kilometer Array (SKA) is an international initiative for developing the world's largest radio telescope with a total collecting area of over a million square meters. The scale of the operation, combined with the remote location of the telescope, requires the use of energy-efficient computational algorithms. This, along with the extreme data rates that will be produced by the SKA and the requirement for a real-time observing capability, necessitates in-situ data processing in an edge style computing solution. More generally, energy efficiency in the modern computing landscape is becoming of paramount concern. Whether it be the power budget that can limit some of the world's largest supercomputers, or the limited power available to the smallest Internet-of-Things devices. In this article, we study the impact of hardware frequency scaling on the energy consumption and execution time of the Fast Fourier Transform (FFT) on NVIDIA GPUs using the cuFFT library. The FFT is used in many areas of science and it is one of the key algorithms used in radio astronomy data processing pipelines. Through the use of frequency scaling, we show that we can lower the power consumption of the NVIDIA A100 GPU when computing the FFT by up to 47% compared to the boost clock frequency, with less than a 10% increase in the execution time. Furthermore, using one common core clock frequency for all tested FFT lengths, we show on average a 43% reduction in power consumption compared to the boost core clock frequency with an increase in the execution time still below 10%. We demonstrate how these results can be used to lower the power consumption of existing data processing pipelines. These savings, when considered over years of operation, can yield significant financial savings, but can also lead to a significant reduction of greenhouse gas emissions. |
| first_indexed | 2024-03-07T04:35:08Z |
| format | Journal article |
| id | oxford-uuid:cfad76d8-05bb-4014-ac71-ff22c437435a |
| institution | University of Oxford |
| language | English |
| last_indexed | 2024-03-07T04:35:08Z |
| publishDate | 2021 |
| publisher | Institute of Electrical and Electronics Engineers |
| record_format | dspace |
| spelling | oxford-uuid:cfad76d8-05bb-4014-ac71-ff22c437435a2022-03-27T07:44:21ZEfficiency near the edge: increasing the energy efficiency of FFTs on GPUs for real-time edge computing Journal articlehttp://purl.org/coar/resource_type/c_dcae04bcuuid:cfad76d8-05bb-4014-ac71-ff22c437435aEnglishSymplectic ElementsInstitute of Electrical and Electronics Engineers2021Adamek, KNovotny, JThiyagalingam, JArmour, WGThe Square Kilometer Array (SKA) is an international initiative for developing the world's largest radio telescope with a total collecting area of over a million square meters. The scale of the operation, combined with the remote location of the telescope, requires the use of energy-efficient computational algorithms. This, along with the extreme data rates that will be produced by the SKA and the requirement for a real-time observing capability, necessitates in-situ data processing in an edge style computing solution. More generally, energy efficiency in the modern computing landscape is becoming of paramount concern. Whether it be the power budget that can limit some of the world's largest supercomputers, or the limited power available to the smallest Internet-of-Things devices. In this article, we study the impact of hardware frequency scaling on the energy consumption and execution time of the Fast Fourier Transform (FFT) on NVIDIA GPUs using the cuFFT library. The FFT is used in many areas of science and it is one of the key algorithms used in radio astronomy data processing pipelines. Through the use of frequency scaling, we show that we can lower the power consumption of the NVIDIA A100 GPU when computing the FFT by up to 47% compared to the boost clock frequency, with less than a 10% increase in the execution time. Furthermore, using one common core clock frequency for all tested FFT lengths, we show on average a 43% reduction in power consumption compared to the boost core clock frequency with an increase in the execution time still below 10%. We demonstrate how these results can be used to lower the power consumption of existing data processing pipelines. These savings, when considered over years of operation, can yield significant financial savings, but can also lead to a significant reduction of greenhouse gas emissions. |
| spellingShingle | Adamek, K Novotny, J Thiyagalingam, J Armour, WG Efficiency near the edge: increasing the energy efficiency of FFTs on GPUs for real-time edge computing |
| title | Efficiency near the edge: increasing the energy efficiency of FFTs on GPUs for real-time edge computing |
| title_full | Efficiency near the edge: increasing the energy efficiency of FFTs on GPUs for real-time edge computing |
| title_fullStr | Efficiency near the edge: increasing the energy efficiency of FFTs on GPUs for real-time edge computing |
| title_full_unstemmed | Efficiency near the edge: increasing the energy efficiency of FFTs on GPUs for real-time edge computing |
| title_short | Efficiency near the edge: increasing the energy efficiency of FFTs on GPUs for real-time edge computing |
| title_sort | efficiency near the edge increasing the energy efficiency of ffts on gpus for real time edge computing |
| work_keys_str_mv | AT adamekk efficiencyneartheedgeincreasingtheenergyefficiencyoffftsongpusforrealtimeedgecomputing AT novotnyj efficiencyneartheedgeincreasingtheenergyefficiencyoffftsongpusforrealtimeedgecomputing AT thiyagalingamj efficiencyneartheedgeincreasingtheenergyefficiencyoffftsongpusforrealtimeedgecomputing AT armourwg efficiencyneartheedgeincreasingtheenergyefficiencyoffftsongpusforrealtimeedgecomputing |