11–20 of 84 posts
Would be interesting to compare with gcc, which has a better implementation of both inlining and register allocation.
The licence doesn't help but most of this work is open source anyway
People think of a compiler as an AI when they're actually very stupid in terms of the number of decisions available to them.
Feedback is the lifeblood of intelligent performance, it is more than possible to fake that feedback using AI.
E.g. Your error callback is on the balance of probability going to be called less than the (say) core matrix multiply loop etc. Etc.
The (current) title used on HN ( “Google ML Compiler Inlining Achieves 3-7% Reduction in Size” ) is confusing. It made me think Google is using ML ( https://en.wikipedia.org/wiki/ML_(programming_language) ) It also is not the one of the referred page, which is “MLGO: A Machine Learning Framework for Compiler Optimization” . This is about an LLVM extension that uses Machine Learning. I think it would be better to chan…
Would be interesting to compare with gcc, which has a better implementation of both inlining and register allocation.
The thing is that GCC successfully shot its own head off via Stallman so no one cares about doing research with it anymore. The licence doesn't help but most of this work is open source anyway
What do you mean about its license?
... and 1-2% improvement in performance via the register allocator.
I wonder how the performance improvements compare to PGO.
(Sincerely, this should be the baseline for every bit of research.)
The (current) title used on HN ( “Google ML Compiler Inlining Achieves 3-7% Reduction in Size” ) is confusing. It made me think Google is using ML ( https://en.wikipedia.org/wiki/ML_(programming_language) ) It also is not the one of the referred page, which is “MLGO: A Machine Learning Framework for Compiler Optimization” . This is about an LLVM extension that uses Machine Learning. I think it would be better to chan…
I think for >99% of folks, ML means machine learning, even if you put compiler after it.
I had the same question as the parent commenter and had to wait until the page loaded before determining whether the article was about machine learning or the programming language.
I wonder how the performance improvements compare to PGO.