[Bug 272742] Fixes for bugs in sinpi/cospi/tanpi

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=272742

--- Comment #1 from Steve Kargl <kargl@FreeBSD.org> ---
All, Paul supplied me with his test framework.  Exhaustive testing of the float
implementations in round-to-nearest yields

Checking cospi
MPFR library: 4.2.0-p9    
MPFR header:  4.2.0-p9 (based on 4.2.0)
Checking function cospif with MPFR_RNDN
libm wrong by up to 5.01e-01 ulp(s) [1] for x=0x1.112f4ep-4
cospi      gives 0x1.f4cd4cp-1
mpfr_cospi gives 0x1.f4cd4ap-1
Total: errors=110478 (0.00%) errors2=0 maxerr=5.01e-01 ulp(s)

Checking sinpi
MPFR library: 4.2.0-p9    
MPFR header:  4.2.0-p9 (based on 4.2.0)
Checking function sinpif with MPFR_RNDN
libm wrong by up to 7.50e-01 ulp(s) [1] for x=0x1.72681p-129
sinpi      gives 0x1.22eaa8p-127
mpfr_sinpi gives 0x1.22eaa4p-127
Total: errors=4614478 (0.11%) errors2=0 maxerr=7.50e-01 ulp(s)

Checking tanpi
MPFR library: 4.2.0-p9    
MPFR header:  4.2.0-p9 (based on 4.2.0)
Checking function tanpif with MPFR_RNDN
libm wrong by up to 8.00e-01 ulp(s) [1] for x=0x1.4442bep-5
tanpi      gives 0x1.fffd42p-4
mpfr_tanpi gives 0x1.fffd44p-4
Total: errors=63422266 (1.48%) errors2=0 maxerr=8.00e-01 ulp(s)

-- 
You are receiving this mail because:
You are the assignee for the bug.