forked from OSchip/llvm-project
d3465e0691
A lot of vectorized code doesn't use masks so we shouldn't penalize them by not doing shuffle combining on avx512 targets. I've added support for VALIGNQ/VALIGND and 512-bit SHUF128 to prevent some regressions. I also prevented recombining 256-bit SHUF128 to PERM2X128. We may not need to add 256-bit SHUF128 support, but I don't think I found any cases requiring that in my testing. Differential Revision: https://reviews.llvm.org/D77928 |
||
|---|---|---|
| .. | ||
| broadcast-scalar-fp.ll | ||
| broadcast-scalar-int.ll | ||
| broadcast-vector-fp.ll | ||
| broadcast-vector-int.ll | ||
| duplicate-high.ll | ||
| duplicate-low.ll | ||
| in_lane_permute.ll | ||
| partial_permute.ll | ||
| permute.ll | ||
| shuffle-interleave.ll | ||
| shuffle-vec.ll | ||
| shuffle.ll | ||
| unpack.ll | ||