This commit is contained in:
Jay D Dee
2019-05-30 16:59:49 -04:00
parent eb3f57bfc7
commit 77c5ae80ab
82 changed files with 6906 additions and 3706 deletions

View File

@@ -49,6 +49,7 @@
* no slowdown from the prefixes is generally observed on AMD CPUs supporting
* XOP, some slowdown is sometimes observed on Intel CPUs with AVX.
*/
/*
#ifdef __XOP__
#warning "Note: XOP is enabled. That's great."
#elif defined(__AVX__)
@@ -60,7 +61,7 @@
#else
#warning "Note: building generic code for non-x86. That's OK."
#endif
*/
/*
* The SSE4 code version has fewer instructions than the generic SSE2 version,
* but all of the instructions are SIMD, thereby wasting the scalar execution