system/freebsd-src
system/freebsd-src
1
0
Fork 0
mirror of https://github.com/freebsd/freebsd-src synced 2024-10-05 08:00:30 +00:00
Code Issues Packages Projects Releases Wiki Activity

Use double precision to simplify and optimize a long division.

Browse source 
On athlons, this gives a speedup of 10-20% for tanf() on uniformly
distributed args in [-2Pi, 2Pi].  (It only directly applies for 43%
of the args and gives a 16-20% speedup for these (more for AXP than
A64) and this gives an overall speedup of 10-12% which is all that it
should; however, it gives an overall speedup of 17-20% with gcc-3.3
on AXP-A64 by mysteriously effected cases where it isn't executed.)

I originally intended to use double precision for all internals of
float trig functions and will probably still do this, but benchmarking
showed that converting to double precision and back is a pessimization
in cases where a simple float precision calculation works, so it may
be optimal to switch precisions only when using extra precision is
much simpler.
This commit is contained in:
Bruce Evans 2005-11-21 00:38:21 +00:00
parent 23f6483e0a
commit 58652034e8
Notes: svn2git 2020-12-20 02:59:44 +00:00 
svn path=/head/; revision=152641
1 changed files with 1 additions and 15 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

16
lib/msun/src/k_tanf.c
View file

@ -63,19 +63,5 @@ __kernel_tanf(float x, float y, int iy)
return (float)(1-((hx>>30)&2))*(v-(float)2.0*(x-(w*w/(w+v)-r)));
}
if(iy==1) return w;
else { /* if allow error up to 2 ulp,
simply return -1.0/(x+r) here */
/* compute -1.0/(x+r) accurately */
float a,t;
int32_t i;
z = w;
GET_FLOAT_WORD(i,z);
SET_FLOAT_WORD(z,i&0xfffff000);
v = r-(z - x); /* z+v = r+x */
t = a = -(float)1.0/w; /* a = -1.0/w */
GET_FLOAT_WORD(i,t);
SET_FLOAT_WORD(t,i&0xfffff000);
s = (float)1.0+t*z;
return t+a*(s+t*v);
}
else return -1.0/((double)x+r);
}