End-to-end programmable computing systems
Abstract Recent technological advances have contributed to the rapid increase in algorithmic complexity of applications, ranging from signal processing to autonomous systems. To control this complexity and endow heterogeneous computing systems with autonomous programming and optimization capabilitie...
| Main Authors: | , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Nature Portfolio
2023-11-01
|
| Series: | Communications Engineering |
| Online Access: | https://doi.org/10.1038/s44172-023-00127-7 |
| _version_ | 1797452548273078272 |
|---|---|
| author | Yao Xiao Guixiang Ma Nesreen K. Ahmed Mihai Capotă Theodore L. Willke Shahin Nazarian Paul Bogdan |
| author_facet | Yao Xiao Guixiang Ma Nesreen K. Ahmed Mihai Capotă Theodore L. Willke Shahin Nazarian Paul Bogdan |
| author_sort | Yao Xiao |
| collection | DOAJ |
| description | Abstract Recent technological advances have contributed to the rapid increase in algorithmic complexity of applications, ranging from signal processing to autonomous systems. To control this complexity and endow heterogeneous computing systems with autonomous programming and optimization capabilities, we propose a unified, end-to-end, programmable graph representation learning (PGL) framework that mines the complexity of high-level programs down to low-level virtual machine intermediate representation, extracts specific computational patterns, and predicts which code segments run best on a core in heterogeneous hardware. PGL extracts multifractal features from code graphs and exploits graph representation learning strategies for automatic parallelization and correct assignment to heterogeneous processors. The comprehensive evaluation of PGL on existing and emerging complex software demonstrates a 6.42x and 2.02x speedup compared to thread-based execution and state-of-the-art techniques, respectively. Our PGL framework leads to higher processing efficiency, which is crucial for future AI and high-performance computing applications such as autonomous vehicles and machine vision. |
| first_indexed | 2024-03-09T15:09:16Z |
| format | Article |
| id | doaj.art-dce5624382d842e2b61228caa8395f1b |
| institution | Directory Open Access Journal |
| issn | 2731-3395 |
| language | English |
| last_indexed | 2024-03-09T15:09:16Z |
| publishDate | 2023-11-01 |
| publisher | Nature Portfolio |
| record_format | Article |
| series | Communications Engineering |
| spelling | doaj.art-dce5624382d842e2b61228caa8395f1b2023-11-26T13:29:09ZengNature PortfolioCommunications Engineering2731-33952023-11-012111510.1038/s44172-023-00127-7End-to-end programmable computing systemsYao Xiao0Guixiang Ma1Nesreen K. Ahmed2Mihai Capotă3Theodore L. Willke4Shahin Nazarian5Paul Bogdan6University of Southern CaliforniaIntel LabsIntel LabsIntel LabsIntel LabsUniversity of Southern CaliforniaUniversity of Southern CaliforniaAbstract Recent technological advances have contributed to the rapid increase in algorithmic complexity of applications, ranging from signal processing to autonomous systems. To control this complexity and endow heterogeneous computing systems with autonomous programming and optimization capabilities, we propose a unified, end-to-end, programmable graph representation learning (PGL) framework that mines the complexity of high-level programs down to low-level virtual machine intermediate representation, extracts specific computational patterns, and predicts which code segments run best on a core in heterogeneous hardware. PGL extracts multifractal features from code graphs and exploits graph representation learning strategies for automatic parallelization and correct assignment to heterogeneous processors. The comprehensive evaluation of PGL on existing and emerging complex software demonstrates a 6.42x and 2.02x speedup compared to thread-based execution and state-of-the-art techniques, respectively. Our PGL framework leads to higher processing efficiency, which is crucial for future AI and high-performance computing applications such as autonomous vehicles and machine vision.https://doi.org/10.1038/s44172-023-00127-7 |
| spellingShingle | Yao Xiao Guixiang Ma Nesreen K. Ahmed Mihai Capotă Theodore L. Willke Shahin Nazarian Paul Bogdan End-to-end programmable computing systems Communications Engineering |
| title | End-to-end programmable computing systems |
| title_full | End-to-end programmable computing systems |
| title_fullStr | End-to-end programmable computing systems |
| title_full_unstemmed | End-to-end programmable computing systems |
| title_short | End-to-end programmable computing systems |
| title_sort | end to end programmable computing systems |
| url | https://doi.org/10.1038/s44172-023-00127-7 |
| work_keys_str_mv | AT yaoxiao endtoendprogrammablecomputingsystems AT guixiangma endtoendprogrammablecomputingsystems AT nesreenkahmed endtoendprogrammablecomputingsystems AT mihaicapota endtoendprogrammablecomputingsystems AT theodorelwillke endtoendprogrammablecomputingsystems AT shahinnazarian endtoendprogrammablecomputingsystems AT paulbogdan endtoendprogrammablecomputingsystems |