projects / platform / upstream / llvm.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: 5f62156) | patch
[X86] Update some of the AVX512 intrinsic tests to avoid adds.
authorLuo, Yuanke <yuanke.luo@intel.com>
Mon, 7 Mar 2022 08:51:42 +0000 (16:51 +0800)
committerLuo, Yuanke <yuanke.luo@intel.com>
Mon, 7 Mar 2022 09:03:31 +0000 (17:03 +0800)
commitbe85f55b2dcbcdce4d96d6dc10d69ecdde72841e
tree0b9f103408b4a1a0bd3fbfe049d8abbe2671ae73tree | snapshot
parent5f62156762d45f53fa70446c718813f9f9a099e5commit | diff
[X86] Update some of the AVX512 intrinsic tests to avoid adds.

As noticed in D119654, by adding the masked intrinsics results together
we can end up with the selects being canonicalized away from the
intrinsic - this isn't what we want to test here so replace with a
insertvalue chain into a aggregate instead to retain all the results.
llvm/test/CodeGen/X86/avx512-intrinsics-upgrade.ll diff | blob | history
Domain: System / Toolchain;