diff options
author | Szabolcs Nagy <nsz@port70.net> | 2013-05-15 23:08:52 +0000 |
---|---|---|
committer | Szabolcs Nagy <nsz@port70.net> | 2013-05-15 23:08:52 +0000 |
commit | e216951f509b71da193da2fc63e25b998740d58b (patch) | |
tree | 391381eafd8b0b7eb4fdbb693608072df5c181d5 /src/math/j1f.c | |
parent | 2c184264eae3797de028403ed1e86c1f7ae5b813 (diff) | |
download | musl-e216951f509b71da193da2fc63e25b998740d58b.tar.gz musl-e216951f509b71da193da2fc63e25b998740d58b.tar.xz musl-e216951f509b71da193da2fc63e25b998740d58b.zip |
math: use double_t for temporaries to avoid stores on i386
When FLT_EVAL_METHOD!=0 (only i386 with x87 fp) the excess precision of an expression must be removed in an assignment. (gcc needs -fexcess-precision=standard or -std=c99 for this) This is done by extra load/store instructions which adds code bloat when lot of temporaries are used and it makes the result less precise in many cases. Using double_t and float_t avoids these issues on i386 and it makes no difference on other archs. For now only a few functions are modified where the excess precision is clearly beneficial (mostly polynomial evaluations with temporaries). object size differences on i386, gcc-4.8: old new __cosdf.o 123 95 __cos.o 199 169 __sindf.o 131 95 __sin.o 225 203 __tandf.o 207 151 __tan.o 605 499 erff.o 1470 1416 erf.o 1703 1649 j0f.o 1779 1745 j0.o 2308 2274 j1f.o 1602 1568 j1.o 2286 2252 tgamma.o 1431 1424 math/*.o 64164 63635
Diffstat (limited to 'src/math/j1f.c')
-rw-r--r-- | src/math/j1f.c | 4 |
1 files changed, 2 insertions, 2 deletions
diff --git a/src/math/j1f.c b/src/math/j1f.c index 5a760f71..6abde349 100644 --- a/src/math/j1f.c +++ b/src/math/j1f.c @@ -198,7 +198,7 @@ static const float ps2[5] = { static float ponef(float x) { const float *p,*q; - float z,r,s; + float_t z,r,s; uint32_t ix; GET_FLOAT_WORD(ix, x); @@ -294,7 +294,7 @@ static const float qs2[6] = { static float qonef(float x) { const float *p,*q; - float s,r,z; + float_t s,r,z; uint32_t ix; GET_FLOAT_WORD(ix, x); |