Files
clang-p2996/llvm/test/tools/llvm-mca
Simon Pilgrim d073b19dbf [X86] Fix SLM FP<->INT throughputs.
Noticed while trying to clean up the shift costs model for SSE4 targets using the script in D10369 - SLM double-pumps all the 128-bit vector conversion ops and only use FP0 pipe - numbers taken from Intel AOM + Agner.
2021-07-22 19:39:04 +01:00
..
2021-07-22 19:39:04 +01:00