From 599524dacf973a96a0e7d6ae7159ad10aa8457a4 Mon Sep 17 00:00:00 2001
From: Recep Aslantas <info@recp.me>
Date: Thu, 10 May 2018 16:27:40 +0300
Subject: [PATCH] docs: add new option to docs

---
 docs/source/opt.rst           | 8 +++++++-
 include/cglm/simd/intrin.h    | 2 +-
 include/cglm/simd/sse2/mat4.h | 6 +++---
 3 files changed, 11 insertions(+), 5 deletions(-)

diff --git a/docs/source/opt.rst b/docs/source/opt.rst
index c614e42..4acb972 100644
--- a/docs/source/opt.rst
+++ b/docs/source/opt.rst
@@ -5,7 +5,7 @@ Options
 
 A few options are provided via macros.
 
-Alignment
+Alignment Option
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 
 As default, cglm requires types to be aligned. Alignment requirements:
@@ -34,3 +34,9 @@ have to compile cglm with **CGLM_ALL_UNALIGNED** macro.
  ALWAYS USE SAME CONFIGURATION / OPTION for **cglm** if you have multiple projects.
 
  For instance if you set CGLM_ALL_UNALIGNED in a project then set it in other projects too
+
+SSE and SSE2 Shuffle Option
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+**_mm_shuffle_ps** generates **shufps** instruction even if registers are same.
+You can force it to generate **pshufd** instruction by defining
+**CGLM_USE_INT_DOMAIN** macro. As default it is not defined.
diff --git a/include/cglm/simd/intrin.h b/include/cglm/simd/intrin.h
index 2f79b01..f4854bd 100644
--- a/include/cglm/simd/intrin.h
+++ b/include/cglm/simd/intrin.h
@@ -32,7 +32,7 @@
 #ifdef CGLM_USE_INT_DOMAIN
 #  define glmm_shuff1(xmm, z, y, x, w)                                        \
      _mm_castsi128_ps(_mm_shuffle_epi32(_mm_castps_si128(xmm),                \
-                      _MM_SHUFFLE(z, y, x, w)))
+                                        _MM_SHUFFLE(z, y, x, w)))
 #else
 #  define glmm_shuff1(xmm, z, y, x, w)                                        \
      _mm_shuffle_ps(xmm, xmm, _MM_SHUFFLE(z, y, x, w))
diff --git a/include/cglm/simd/sse2/mat4.h b/include/cglm/simd/sse2/mat4.h
index 20afeae..c2acb9f 100644
--- a/include/cglm/simd/sse2/mat4.h
+++ b/include/cglm/simd/sse2/mat4.h
@@ -168,8 +168,8 @@ glm_mat4_inv_fast_sse2(mat4 mat, mat4 dest) {
   r3 = glmm_load(mat[3]); /* p o n m */
 
   x0 = _mm_shuffle_ps(r2, r3, _MM_SHUFFLE(3, 2, 3, 2));  /* p o l k */
-  x1 = glmm_shuff1(x0, 1, 3, 3, 3);                  /* l p p p */
-  x2 = glmm_shuff1(x0, 0, 2, 2, 2);                  /* k o o o */
+  x1 = glmm_shuff1(x0, 1, 3, 3, 3);                      /* l p p p */
+  x2 = glmm_shuff1(x0, 0, 2, 2, 2);                      /* k o o o */
   x0 = _mm_shuffle_ps(r2, r1, _MM_SHUFFLE(3, 3, 3, 3));  /* h h l l */
   x3 = _mm_shuffle_ps(r2, r1, _MM_SHUFFLE(2, 2, 2, 2));  /* g g k k */
 
@@ -180,7 +180,7 @@ glm_mat4_inv_fast_sse2(mat4 mat, mat4 dest) {
   t0 = _mm_sub_ps(_mm_mul_ps(x3, x1), _mm_mul_ps(x2, x0));
 
   x4 = _mm_shuffle_ps(r2, r3, _MM_SHUFFLE(2, 1, 2, 1)); /* o n k j */
-  x4 = glmm_shuff1(x4, 0, 2, 2, 2);                 /* j n n n */
+  x4 = glmm_shuff1(x4, 0, 2, 2, 2);                     /* j n n n */
   x5 = _mm_shuffle_ps(r2, r1, _MM_SHUFFLE(1, 1, 1, 1)); /* f f j j */
 
   /* t1[1] = j * p - n * l;