piwalker/dolphin
1
0
Fork 0
mirror of https://github.com/dolphin-emu/dolphin.git synced 2025-07-25 15:19:42 -06:00
Code Issues Packages Projects Releases Wiki Activity

JIT: more optimizing of float ops based on known input characteristics

Browse Source 
If the inputs are both float singles, and the top half is known to be identical
to the bottom half, we can use packed arithmetic instead of scalar to skip
the movddup.

This is slower on a few rather old CPUs, plus the Atom+Silvermont, so detect
Atom and disable it in that case.

Also avoid PPC_FP on stores if we know that the output came from a float op.
This commit is contained in:
Fiora
2014-10-11 14:22:44 -07:00
parent 4e0591cdf1
commit 72c96c20d3
9 changed files with 154 additions and 39 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
Source/Core/Common/x64CPUDetect.cpp
View File

@ -129,6 +129,12 @@ void CPUInfo::Detect()
if (max_std_fn >= 1)
{
__cpuid(cpu_id, 0x00000001);
int family = ((cpu_id[0] >> 8) & 0xf) + ((cpu_id[0] >> 20) & 0xff);
int model = ((cpu_id[0] >> 4) & 0xf) + ((cpu_id[0] >> 12) & 0xf0);
// Detect people unfortunate enough to be running Dolphin on an Atom
if (family == 6 && (model == 0x1C || model == 0x26 ||model == 0x27 || model == 0x35 || model == 0x36 ||
model == 0x37 || model == 0x4A || model == 0x4D || model == 0x5A || model == 0x5D))
bAtom = true;
logical_cpu_count = (cpu_id[1] >> 16) & 0xFF;
ht = (cpu_id[3] >> 28) & 1;