[Pixman] [PATCH v4 1/2] mmx: Use _mm_mulhi_pu16

[Matt Turner]

The pmulhuw x86 instruction is part of Extended 3DNow! and SSE1. The
equivalent ARM wmuluh instruction was available from the first iwMMXt
instrucion set.

This instruction is already used in the SSE2 code.

Reduces code size by ~5%.

amd64
  text    data     bss     dec     hex filename
 31325    2240       0   33565    831d .libs/libpixman_mmx_la-pixman-mmx.o
 29925    2240       0   32165    7da5 .libs/libpixman_mmx_la-pixman-mmx.o

x86
  text    data     bss     dec     hex filename
 29165    1792       0   30957    78ed .libs/libpixman_mmx_la-pixman-mmx.o
 27677    1792       0   29469    731d .libs/libpixman_mmx_la-pixman-mmx.o

arm
  text    data     bss     dec     hex filename
 31632    1792       0   33424    8290 .libs/libpixman_iwmmxt_la-pixman-mmx.o
 30176    1792       0   31968    7ce0 .libs/libpixman_iwmmxt_la-pixman-mmx.o

v4: use inline assembly since the intrinsic needs SSE or 3DNowA
	- we can't use -msse, since it'll cause more SSE instructions to be
	  generated that we don't want
	- there is no -m3dnowa flag (anymore?)


Signed-off-by: Matt Turner <mattst88 at gmail.com>
---
 pixman/pixman-mmx.c |   20 ++++++++++++++++++--
 1 files changed, 18 insertions(+), 2 deletions(-)

diff --git a/pixman/pixman-mmx.c b/pixman/pixman-mmx.c
index 937ce8f..a31adf8 100644
--- a/pixman/pixman-mmx.c
+++ b/pixman/pixman-mmx.c
@@ -56,6 +56,21 @@ _mm_empty (void)
 }
 #endif
 
+#ifdef USE_X86_MMX
+/* We have to compile with -msse to use xmmintrin.h, but that causes SSE
+ * instructions to be generated that we don't want. Just duplicate the
+ * functions we want to use.  */
+extern __inline __m64 __attribute__((__gnu_inline__, __always_inline__, __artificial__))
+_mm_mulhi_pu16 (__m64 __A, __m64 __B)
+{
+    asm("pmulhuw %1, %0\n\t"
+	: "+y" (__A)
+	: "y" (__B)
+    );
+    return __A;
+}
+#endif
+
 /* Notes about writing mmx code
  *
  * give memory operands as the second operand. If you give it as the
@@ -128,6 +143,7 @@ typedef struct
     mmxdatafield mmx_ffff0000ffff0000;
     mmxdatafield mmx_0000ffff00000000;
     mmxdatafield mmx_000000000000ffff;
+    mmxdatafield mmx_4x0101;
 } mmx_data_t;
 
 #if defined(_MSC_VER)
@@ -155,6 +171,7 @@ static const mmx_data_t c =
     MMXDATA_INIT (.mmx_ffff0000ffff0000,         0xffff0000ffff0000),
     MMXDATA_INIT (.mmx_0000ffff00000000,         0x0000ffff00000000),
     MMXDATA_INIT (.mmx_000000000000ffff,         0x000000000000ffff),
+    MMXDATA_INIT (.mmx_4x0101,                   0x0101010101010101),
 };
 
 #ifdef USE_CVT_INTRINSICS
@@ -218,8 +235,7 @@ pix_multiply (__m64 a, __m64 b)
 
     res = _mm_mullo_pi16 (a, b);
     res = _mm_adds_pu16 (res, MC (4x0080));
-    res = _mm_adds_pu16 (res, _mm_srli_pi16 (res, 8));
-    res = _mm_srli_pi16 (res, 8);
+    res = _mm_mulhi_pu16 (res, MC (4x0101));
 
     return res;
 }
-- 
1.7.3.4