ShuriZma
/
visualboyadvance-m
mirror of https://github.com/visualboyadvance-m/visualboyadvance-m.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

Revert "Use MSVC intrinsic sqrt() on x86 for XBRZ" 

This reverts commit 55a0a52a39.

Does not make any difference, the performance issue is elsewhere.
This commit is contained in:
Rafael Kitover 2022-02-03 07:57:51 +00:00
parent 55a0a52a39
commit 8bd09c9012
No known key found for this signature in database
GPG Key ID: 08AB596679D86240
1 changed files with 7 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

7
src/filters/xBRZ/xbrz.cpp
View File

@ -69,6 +69,13 @@ inline double fastSqrt(double n)
#if (defined(__GNUC__) || defined(__clang__)) && (defined(__x86_64__) || defined(__i386__)) #if (defined(__GNUC__) || defined(__clang__)) && (defined(__x86_64__) || defined(__i386__))
__asm__ ("fsqrt" : "+t" (n)); __asm__ ("fsqrt" : "+t" (n));
return n; return n;
#elif defined(_MSC_VER) && defined(_M_IX86)
// speeds up xBRZ by about 9% compared to std::sqrt which internally uses
// the same assembler instructions but adds some "fluff"
__asm {
fld n
fsqrt
}
#elif defined(_MSC_VER) #elif defined(_MSC_VER)
// On MSVC x64 use intrinsic with /Oi and /fp:fast // On MSVC x64 use intrinsic with /Oi and /fp:fast
return sqrt(n); return sqrt(n);