[clang] Add _MM_FROUND_TO_NEAREST_TIES_EVEN to avx512fintrin.h (PR #99691)
via cfe-commits
cfe-commits at lists.llvm.org
Thu Jul 25 17:06:24 PDT 2024
hpkfft wrote:
I see that both phoebewang and KanRobert work for Intel, so Intel's LLVM compiler `icx` is sure to get this patch.
Intel itself is both a customer and partner of Microsoft, so MSVC ought not be a problem. After all, it is in both Intel's and Microsoft's interest to delight their developer community by supplying a macro that is suitably named for what it does.
But it needs to start here (to demonstrate some measure of community review and support).
Regarding the use of `NEAREST_INT` for instructions that do not round to integer, I am reminded of the line spoken by IƱigo Montoya:
> You keep using that word. I do not think it means what you think it means.
Yes, it will take some time for this change to percolate though to other code bases and documents. But as in the proverb:
> Society grows great when old men plant trees in whose shade they know they shall never sit.
https://github.com/llvm/llvm-project/pull/99691
More information about the cfe-commits
mailing list