External/cglm
1
0
Fork 0
mirror of https://github.com/recp/cglm.git synced 2025-12-24 20:34:58 +00:00
Code Issues Packages Projects Releases Wiki Activity

vec4: optimize muladd and muladds with fma

Browse Source
This commit is contained in:
Recep Aslantas
2021-04-18 16:24:29 +03:00
parent 0d0d22f96c
commit f3f29bd383
1 changed files with 4 additions and 16 deletions
Download Patch File Download Diff File Expand all files Collapse all files

20
include/cglm/vec4.h
View File

@@ -568,14 +568,8 @@ glm_vec4_subadd(vec4 a, vec4 b, vec4 dest) {
CGLM_INLINE
void
glm_vec4_muladd(vec4 a, vec4 b, vec4 dest) {
#if defined( __SSE__ ) || defined( __SSE2__ )
glmm_store(dest, _mm_add_ps(glmm_load(dest),
_mm_mul_ps(glmm_load(a),
glmm_load(b))));
#elif defined(CGLM_NEON_FP)
vst1q_f32(dest, vaddq_f32(vld1q_f32(dest),
vmulq_f32(vld1q_f32(a),
vld1q_f32(b))));
#if defined(CGLM_SIMD)
glmm_store(dest, glmm_fmadd(glmm_load(a), glmm_load(b), glmm_load(dest)));
#else
dest[0] += a[0] * b[0];
dest[1] += a[1] * b[1];
@@ -596,14 +590,8 @@ glm_vec4_muladd(vec4 a, vec4 b, vec4 dest) {
CGLM_INLINE
void
glm_vec4_muladds(vec4 a, float s, vec4 dest) {
#if defined( __SSE__ ) || defined( __SSE2__ )
glmm_store(dest, _mm_add_ps(glmm_load(dest),
_mm_mul_ps(glmm_load(a),
_mm_set1_ps(s))));
#elif defined(CGLM_NEON_FP)
vst1q_f32(dest, vaddq_f32(vld1q_f32(dest),
vmulq_f32(vld1q_f32(a),
vdupq_n_f32(s))));
#if defined(CGLM_SIMD)
glmm_store(dest, glmm_fmadd(glmm_load(a), _mm_set1_ps(s), glmm_load(dest)));
#else
dest[0] += a[0] * s;
dest[1] += a[1] * s;