Energy-Efficient Approximate Computation in Topaz
We present Topaz, a new task-based language for computations that execute on approximate computing platforms that may occasionally produce arbitrarily inaccurate results. The Topaz implementation maps approximate tasks onto the approximate machine and integrates the approximate results into the main...
| Main Authors: | , |
|---|---|
| Other Authors: | |
| Published: |
2014
|
| Online Access: | http://hdl.handle.net/1721.1/88926 |
| _version_ | 1826203411246219264 |
|---|---|
| author | Achour, Sara Rinard, Martin |
| author2 | Martin Rinard |
| author_facet | Martin Rinard Achour, Sara Rinard, Martin |
| author_sort | Achour, Sara |
| collection | MIT |
| description | We present Topaz, a new task-based language for computations that execute on approximate computing platforms that may occasionally produce arbitrarily inaccurate results. The Topaz implementation maps approximate tasks onto the approximate machine and integrates the approximate results into the main computation, deploying a novel outlier detection and reliable reexecution mechanism to prevent unacceptably inaccurate results from corrupting the overall computation. Topaz therefore provides the developers of approximate hardware with substantial freedom in producing designs with little or no precision or accuracy guarantees. Experimental results from our set of benchmark applications demonstrate the effectiveness of Topaz and the Topaz implementation in enabling developers to productively exploit emerging approximate hardware platforms. |
| first_indexed | 2024-09-23T12:36:09Z |
| id | mit-1721.1/88926 |
| institution | Massachusetts Institute of Technology |
| last_indexed | 2024-09-23T12:36:09Z |
| publishDate | 2014 |
| record_format | dspace |
| spelling | mit-1721.1/889262019-04-11T13:52:51Z Energy-Efficient Approximate Computation in Topaz Achour, Sara Rinard, Martin Martin Rinard Computer Architecture We present Topaz, a new task-based language for computations that execute on approximate computing platforms that may occasionally produce arbitrarily inaccurate results. The Topaz implementation maps approximate tasks onto the approximate machine and integrates the approximate results into the main computation, deploying a novel outlier detection and reliable reexecution mechanism to prevent unacceptably inaccurate results from corrupting the overall computation. Topaz therefore provides the developers of approximate hardware with substantial freedom in producing designs with little or no precision or accuracy guarantees. Experimental results from our set of benchmark applications demonstrate the effectiveness of Topaz and the Topaz implementation in enabling developers to productively exploit emerging approximate hardware platforms. 2014-08-19T21:00:06Z 2014-08-19T21:00:06Z 2014-08-19 2014-08-19T21:00:07Z http://hdl.handle.net/1721.1/88926 MIT-CSAIL-TR-2014-016 41 p. application/pdf |
| spellingShingle | Achour, Sara Rinard, Martin Energy-Efficient Approximate Computation in Topaz |
| title | Energy-Efficient Approximate Computation in Topaz |
| title_full | Energy-Efficient Approximate Computation in Topaz |
| title_fullStr | Energy-Efficient Approximate Computation in Topaz |
| title_full_unstemmed | Energy-Efficient Approximate Computation in Topaz |
| title_short | Energy-Efficient Approximate Computation in Topaz |
| title_sort | energy efficient approximate computation in topaz |
| url | http://hdl.handle.net/1721.1/88926 |
| work_keys_str_mv | AT achoursara energyefficientapproximatecomputationintopaz AT rinardmartin energyefficientapproximatecomputationintopaz |