tpp-mlir
tpp-mlir copied to clipboard
AVX512 test runs on machine without AVX512
Machine: AMD EPYC 7742 64-Core Processor
Extension: avx2
Test: BF16/Integration/vector-contract-to-bf16dp.mlir
Error:
SplitVectorResult #0: t40: v16f32 = llvm.x86.avx512bf16.dpbf16ps.512 TargetConstant:i64<14225>, t39, t15, t32
LLVM ERROR: Do not know how to split the result of this operator!
Some error while detecting AVX512 to enable the pass. It should return matchFailure if there's no AVX512 on the target specification.
I could not reproduce this issue on compile node.
I think #1031 solved it