From 9aebdc76b3ac72206274680886272a6aea85f497 Mon Sep 17 00:00:00 2001
From: Recep Aslantas <info@recp.me>
Date: Tue, 30 Oct 2018 09:58:11 +0300
Subject: [PATCH] avx: implement scale matrix using AVX

---
 include/cglm/mat4.h          |  4 +++-
 include/cglm/simd/avx/mat4.h | 10 ++++++++++
 2 files changed, 13 insertions(+), 1 deletion(-)

diff --git a/include/cglm/mat4.h b/include/cglm/mat4.h
index 8fbcddd..b3af335 100644
--- a/include/cglm/mat4.h
+++ b/include/cglm/mat4.h
@@ -478,7 +478,9 @@ glm_mat4_scale_p(mat4 m, float s) {
 CGLM_INLINE
 void
 glm_mat4_scale(mat4 m, float s) {
-#if defined( __SSE__ ) || defined( __SSE2__ )
+#ifdef __AVX__
+  glm_mat4_scale_avx(m, s);
+#elif defined( __SSE__ ) || defined( __SSE2__ )
   glm_mat4_scale_sse2(m, s);
 #else
   glm_mat4_scale_p(m, s);
diff --git a/include/cglm/simd/avx/mat4.h b/include/cglm/simd/avx/mat4.h
index 3c7c397..44ed318 100644
--- a/include/cglm/simd/avx/mat4.h
+++ b/include/cglm/simd/avx/mat4.h
@@ -14,6 +14,16 @@
 
 #include <immintrin.h>
 
+CGLM_INLINE
+void
+glm_mat4_scale_avx(mat4 m, float s) {
+  __m256 y0;
+  y0 = _mm256_set1_ps(s);
+
+  glmm_store256(m[0], _mm256_mul_ps(y0, glmm_load256(m[0])));
+  glmm_store256(m[2], _mm256_mul_ps(y0, glmm_load256(m[2])));
+}
+
 CGLM_INLINE
 void
 glm_mat4_mul_avx(mat4 m1, mat4 m2, mat4 dest) {