Pop fast-math at the end of the file
so that it doesn't affect the rest of the code in Griffin builds. This is unfortunately GCC-only, but Clang will miss only 1 vectorization.
This commit is contained in:
parent
63487e8ed8
commit
2e868fc00b
|
@ -22,7 +22,8 @@
|
|||
|
||||
/* Bog-standard windowed SINC implementation. */
|
||||
|
||||
#if defined(__GNUC__) && defined(__OPTIMIZE__)
|
||||
#if defined(__GNUC__) && defined(__OPTIMIZE__) && !defined(__clang__)
|
||||
#pragma GCC push_options
|
||||
#pragma GCC optimize ("fast-math")
|
||||
#endif
|
||||
|
||||
|
@ -1027,3 +1028,7 @@ retro_resampler_t sinc_resampler = {
|
|||
"sinc",
|
||||
"sinc"
|
||||
};
|
||||
|
||||
#if defined(__GNUC__) && defined(__OPTIMIZE__) && !defined(__clang__)
|
||||
#pragma GCC pop_options
|
||||
#endif
|
||||
|
|
Loading…
Reference in New Issue