OpenE2K
/
gcc
2
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

xmmintrin.h (_MM_TRANSPOSE4_PS): Rewrite using high/low moves and unpack to speed up. 

2005-11-29  Evan Cheng  <evan.cheng@apple.com>

        * config/i386/xmmintrin.h (_MM_TRANSPOSE4_PS): Rewrite using high/low
        moves and unpack to speed up.

From-SVN: r107700
This commit is contained in:
Evan Cheng 2005-11-30 06:29:07 +00:00 committed by Eric Christopher
parent 7e04157d14
commit be7724ed74
2 changed files with 14 additions and 9 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

7
gcc/ChangeLog
View File

@ -1,3 +1,8 @@
2005-11-29 Evan Cheng <evan.cheng@apple.com>
* config/i386/xmmintrin.h (_MM_TRANSPOSE4_PS): Rewrite using high/low
moves and unpack to speed up.
2005-11-29 David S. Miller <davem@sunset.davemloft.net>
* config/sparc/sparc.c (gen_compare_reg): Kill 2nd and 3rd
@ -107,7 +112,7 @@
Uros Bizjak <uros@kss-loka.si>
PR middle-end/20219
* fold-const.c (fold binary) <RDIV_EXPR>: Optimize
* fold-const.c (fold binary) <RDIV_EXPR>: Optimize
sin(x)/tan(x) as cos(x) and tan(x)/sin(x) as 1.0/cos(x)
when flag_unsafe_math_optimizations is set and
we don't care about NaNs or Infinities.

16
gcc/config/i386/xmmintrin.h
View File

@ -1197,14 +1197,14 @@ _mm_pause (void)
#define _MM_TRANSPOSE4_PS(row0, row1, row2, row3) \
do { \
__v4sf __r0 = (row0), __r1 = (row1), __r2 = (row2), __r3 = (row3); \
__v4sf __t0 = __builtin_ia32_shufps (__r0, __r1, 0x44); \
__v4sf __t2 = __builtin_ia32_shufps (__r0, __r1, 0xEE); \
__v4sf __t1 = __builtin_ia32_shufps (__r2, __r3, 0x44); \
__v4sf __t3 = __builtin_ia32_shufps (__r2, __r3, 0xEE); \
(row0) = __builtin_ia32_shufps (__t0, __t1, 0x88); \
(row1) = __builtin_ia32_shufps (__t0, __t1, 0xDD); \
(row2) = __builtin_ia32_shufps (__t2, __t3, 0x88); \
(row3) = __builtin_ia32_shufps (__t2, __t3, 0xDD); \
__v4sf __t0 = __builtin_ia32_unpcklps (__r0, __r1); \
__v4sf __t2 = __builtin_ia32_unpcklps (__r2, __r3); \
__v4sf __t1 = __builtin_ia32_unpckhps (__r0, __r1); \
__v4sf __t3 = __builtin_ia32_unpckhps (__r2, __r3); \
(row0) = __builtin_ia32_movlhps (__t0, __t1); \
(row1) = __builtin_ia32_movhlps (__t1, __t0); \
(row2) = __builtin_ia32_movlhps (__t2, __t3); \
(row3) = __builtin_ia32_movhlps (__t3, __t2); \
} while (0)
/* For backward source compatibility. */