archive/rpcs3
1
0
Fork 0
mirror of https://github.com/RPCS3/rpcs3.git synced 2025-07-09 08:21:29 +12:00
Code Issues Projects Releases Packages Wiki Activity Actions

PPU/SPU LLVM: Allow Zen4 cpus to use VPERMI2B/VPERMT2B instead of the vperm2b256to128 path

Browse source 
- Zen4 based cpus can process VPERM2B in a single uop, unlike intel where it is 3 uops.
This commit is contained in:
Malcolm Jestadt 2022-09-30 17:39:24 -04:00 • committed by kd-11
parent 7d32dc312f
commit d8897c585d
5 changed files with 27 additions and 6 deletions
Download patch file Download diff file Expand all files Collapse all files

13
rpcs3/util/sysinfo.cpp
View file

@ -227,6 +227,19 @@ bool utils::has_fma4()
#endif
}
// The Zen4 based CPUs support VPERMI2B/VPERMT2B in a single uop.
// Current Intel cpus (as of 2022) need 3 uops to execute these instructions.
// Check for SSE4A (which intel doesn't doesn't support) as well as VBMI.
bool utils::has_fast_vperm2b()
{
#if defined(ARCH_X64)
static const bool g_value = has_avx512() && (get_cpuid(7, 0)[2] & 0x2) == 0x2 && get_cpuid(0, 0)[0] >= 0x7 && (get_cpuid(0x80000001, 0)[2] & 0x20) == 0x20;
return g_value;
#else
return false;
#endif
}
bool utils::has_erms()
{
#if defined(ARCH_X64)