mirror of
https://github.com/RPCS3/llvm-mirror.git
synced 2024-11-22 10:42:39 +01:00
815b215830
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. |
||
---|---|---|
.. | ||
AArch64 | ||
AMDGPU | ||
ARM | ||
JSON/X86 | ||
SystemZ | ||
X86 | ||
invalid_input_file_name.test | ||
lit.local.cfg |