Mesa (master): util: Optimize _mesa_roundeven with SSE 4.1.

[Matt Turner]

Module: Mesa
Branch: master
Commit: 036e347f3c129bb547137aed955e75062fca09b8
URL:    http://cgit.freedesktop.org/mesa/mesa/commit/?id=036e347f3c129bb547137aed955e75062fca09b8

Author: Matt Turner <mattst88 at gmail.com>
Date:   Wed Mar 18 14:23:41 2015 -0700

util: Optimize _mesa_roundeven with SSE 4.1.

The SSE 4.1 ROUND instructions let us implement roundeven directly.
Otherwise we assume that the rounding mode has not been modified (as we
do in the rest of Mesa) and use rint().

glibc uses the ROUND instruction in rint() after a cpuid check. This
patch just lets us inline it directly when we're already building for
SSE 4.1.

Reviewed-by: Carl Worth <cworth at cworth.org>

---

 src/util/rounding.h |   20 ++++++++++++++++++++
 1 file changed, 20 insertions(+)

diff --git a/src/util/rounding.h b/src/util/rounding.h
index c8be450..0cbe926 100644
--- a/src/util/rounding.h
+++ b/src/util/rounding.h
@@ -23,6 +23,10 @@
 
 #include <math.h>
 
+#ifdef __SSE4_1__
+#include <smmintrin.h>
+#endif
+
 /* The C standard library has functions round()/rint()/nearbyint() that round
  * their arguments according to the rounding mode set in the floating-point
  * control register. While there are trunc()/ceil()/floor() functions that do
@@ -45,7 +49,15 @@
 static inline float
 _mesa_roundevenf(float x)
 {
+#ifdef __SSE4_1__
+   float ret;
+   __m128 m = _mm_load_ss(&x);
+   m = _mm_round_ss(m, m, _MM_FROUND_CUR_DIRECTION | _MM_FROUND_NO_EXC);
+   _mm_store_ss(&ret, m);
+   return ret;
+#else
    return rintf(x);
+#endif
 }
 
 /**
@@ -54,5 +66,13 @@ _mesa_roundevenf(float x)
 static inline double
 _mesa_roundeven(double x)
 {
+#ifdef __SSE4_1__
+   double ret;
+   __m128d m = _mm_load_sd(&x);
+   m = _mm_round_sd(m, m, _MM_FROUND_CUR_DIRECTION | _MM_FROUND_NO_EXC);
+   _mm_store_sd(&ret, m);
+   return ret;
+#else
    return rint(x);
+#endif
 }