Break out SSE functions into new media_sse target.

It turns out Chrome doesn't have a minimum requirement of SSE yet,
so we can't rely on __SSE__ being set at compile time.  To use SSE
code we need to put it in a separate GYP target compiled with -msse.

This patch set does exactly that for vector_fmac::FMAC_SSE() and
SincResampler::Convolve_SSE().  Doing so required some slight
rearrangements of constants for SincResampler.

Given all of our bots should have SSE I've made it a requirement
for passing the tests when run on X86.

BUG=none
TEST=media_unittests

Review URL: https://codereview.chromium.org/12478002

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@186285 0039d316-1c4b-4281-b951-d872f2087c98

1 files changed, 5 insertions, 21 deletions

diff --git a/media/base/vector_math.cc b/media/base/vector_math.cc
index edd95cd..96f94d9 100644
--- a/media/base/vector_math.cc
+++ b/media/base/vector_math.cc
@@ -7,11 +7,6 @@
 
 #include "base/cpu.h"
 #include "base/logging.h"
-#include "build/build_config.h"
-
-#if defined(ARCH_CPU_X86_FAMILY) && defined(__SSE__)
-#include <xmmintrin.h>
-#endif
 
 namespace media {
 namespace vector_math {
@@ -25,9 +20,13 @@ void FMAC(const float src[], float scale, int len, float dest[]) {
   // selection thread safe.
   typedef void (*VectorFMACProc)(const float src[], float scale, int len,
                                  float dest[]);
-#if defined(ARCH_CPU_X86_FAMILY) && defined(__SSE__)
+#if defined(ARCH_CPU_X86_FAMILY)
+#if defined(__SSE__)
+  static const VectorFMACProc kVectorFMACProc = FMAC_SSE;
+#else
   static const VectorFMACProc kVectorFMACProc =
       base::CPU().has_sse() ? FMAC_SSE : FMAC_C;
+#endif
 #else
   static const VectorFMACProc kVectorFMACProc = FMAC_C;
 #endif
@@ -40,20 +39,5 @@ void FMAC_C(const float src[], float scale, int len, float dest[]) {
     dest[i] += src[i] * scale;
 }
 
-#if defined(ARCH_CPU_X86_FAMILY) && defined(__SSE__)
-void FMAC_SSE(const float src[], float scale, int len, float dest[]) {
-  __m128 m_scale = _mm_set_ps1(scale);
-  int rem = len % 4;
-  for (int i = 0; i < len - rem; i += 4) {
-    _mm_store_ps(dest + i, _mm_add_ps(_mm_load_ps(dest + i),
-                 _mm_mul_ps(_mm_load_ps(src + i), m_scale)));
-  }
-
-  // Handle any remaining values that wouldn't fit in an SSE pass.
-  if (rem)
-    FMAC_C(src + len - rem, scale, rem, dest + len - rem);
-}
-#endif
-
 }  // namespace vector_math
 }  // namespace media