charmen
Re:Compiler Auto-Vectorization with Imitation Learning [pdf]
on Tuesday January 07, 2020 @11:55AM (#21981765)
The MIT news article is about the original performance prediction work, which is an important piece for future reinforcement learning-based approaches. (https://github.com/ithemal/Ithemal http://proceedings.mlr.press/v97/mendis19a.html). The imitation learning approach is a step towards how to automate the development of compiler algorithms. Note that we are not merely tuning heuristics, we learn the entire optimization algorithm.
Re:Compiler Auto-Vectorization with Imitation Learning [pdf]
on Tuesday January 07, 2020 @11:50AM (#21981706)
The MIT news article is about the original performance prediction work, which is an important piece for future reinforcement learning-based approaches. https://github.com/ithemal/Ithemal
http://proceedings.mlr.press/v97/mendis19a.html
How fast will your code run on the latest Intel chip? Find out with Ithemal
With only black-box access to a processor, we use machine learning to answer the question with half the error of Intel's own tools.Demo: http://3.18.198.23/predict Code: https://github.com/psg-mit/Ithemal
Re:Rejuvenating Performance of Hand-Vectorized Programs
on Wednesday February 20, 2019 @10:37PM (#19214318)
Github: https://github.com/revec
Rejuvenating Performance of Hand-Vectorized Programs
Hi Hacker News,Recently we developed a compiler pass (https://www.nextgenvec.org/#revec) in LLVM to retarget hand-vectorized codes to use newer vector instructions. We reinstate Performance Portability back to hand-vectorized codes. I am happy to answer any questions regarding it!