From f3f29bd383f439d0dc1f949ac3839c5b594158f7 Mon Sep 17 00:00:00 2001
From: Recep Aslantas <info@recp.me>
Date: Sun, 18 Apr 2021 16:24:29 +0300
Subject: [PATCH] vec4: optimize muladd and muladds with fma

---
 include/cglm/vec4.h | 20 ++++----------------
 1 file changed, 4 insertions(+), 16 deletions(-)

diff --git a/include/cglm/vec4.h b/include/cglm/vec4.h
index 7a4549c..54d487f 100644
--- a/include/cglm/vec4.h
+++ b/include/cglm/vec4.h
@@ -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;