summaryrefslogtreecommitdiff
path: root/src/math
AgeCommit message (Collapse)AuthorLines
2012-05-01support alternate glibc name pow10 for exp10Rich Felker-0/+9
2012-04-30add C stub for sqrtl (already implemented in asm on i386 and x86_64)Rich Felker-0/+9
2012-04-30first try at writing an efficient and "correct" exp10Rich Felker-0/+55
this is a nonstandard function so it's not clear what conditions it should satisfy. my intent is that it be fast and exact for positive integral exponents when the result fits in the destination type, and fast and correctly rounded for small negative integral exponents. otherwise we aim for at most 1ulp error; it seems to differ from pow by at most 1ulp and it's often 2-5 times faster than pow.
2012-04-29fix typo in the x86_64 rounding asmRich Felker-4/+4
2012-04-29new math asm (abs/rounding) for x86_64Rich Felker-0/+36
untested
2012-04-04math: fix x86 asin accuracynsz-5/+6
use (1-x)*(1+x) instead of (1-x*x) in asin.s the later can be inaccurate with upward rounding when x is close to 1
2012-03-29math: minor cleanups in ceil and floornsz-10/+7
2012-03-29math: remove x86 modf asmnsz-111/+0
the int part was wrong when -1 < x <= -0 (+0.0 instead of -0.0) and the size and performace gain of the asm version was negligible
2012-03-29math: rewrite modf.c and clean up modff.cnsz-72/+45
cleaner implementation with unions and unsigned arithmetic
2012-03-28math: fix modfl.c bugnsz-1/+1
modfl(+-inf) was wrong on ld80 because the explicit msb was not taken into account during inf vs nan check
2012-03-27math: fix a regression in powl and do some cleanupsnsz-12/+11
previously a division was accidentally turned into integer div (w = -i/NXT;) instead of long double div (w = -i; w /= NXT;)
2012-03-27math: add dummy tgamma and tgammaf implementationsnsz-0/+32
2012-03-27math: remove comment about aliasing lgamma as gammansz-8/+4
It is probably not worth supporting gamma. (it was already deprecated in 4.3BSD)
2012-03-27math: fix typo in i386 remquof and remquol asmnsz-5/+5
(fldl instruction was used instead of flds and fldt)
2012-03-23Merge branch 'master' of git://git.etalabs.net/muslnsz-0/+87
2012-03-23asm for hypot and hypotfRich Felker-0/+87
special care is made to avoid any inexact computations when either arg is zero (in which case the exact absolute value of the other arg should be returned) and to support the special condition that hypot(±inf,nan) yields inf. hypotl is not yet implemented since avoiding overflow is nontrivial.
2012-03-23minor rintl.c fix: remove unsupported ldbl format messagensz-2/+0
2012-03-23fix tgammal: don't set the signgam globalnsz-13/+6
(tgamma must be thread-safe, signgam is for lgamma* functions)
2012-03-22acos.s fix: use the formula acos(x) = atan2(sqrt(1-x),sqrt(1+x))nsz-6/+2
the old formula atan2(1,sqrt((1+x)/(1-x))) was faster but could give nan result at x=1 when the rounding mode is FE_DOWNWARD (so 1-1 == -0 and 2/-0 == -inf), the new formula gives -0 at x=+-1 with downward rounding.
2012-03-20x86_64 math asm, long double functions onlyRich Felker-0/+239
this has not been tested heavily, but it's known to at least assemble and run in basic usage cases. it's nearly identical to the corresponding i386 code, and thus expected to be just as correct or just as incorrect.
2012-03-20nearbyint optimization (only clear inexact when necessary)nsz-19/+29
old code saved/restored the fenv (the new code is only as slow as that when inexact is not set before the call, but some other flag is set and the rounding is inexact, which is rare) before: bench_nearbyint_exact 5000000 N 261 ns/op bench_nearbyint_inexact_set 5000000 N 262 ns/op bench_nearbyint_inexact_unset 5000000 N 261 ns/op after: bench_nearbyint_exact 10000000 N 94.99 ns/op bench_nearbyint_inexact_set 25000000 N 65.81 ns/op bench_nearbyint_inexact_unset 10000000 N 94.97 ns/op
2012-03-20remove a fixme commentnsz-3/+2
2012-03-20clean up pow.c and powf.cnsz-57/+43
fix comments about special cases
2012-03-20clean up powl.cnsz-92/+47
fix special cases, use multiplication instead of scalbnl
2012-03-20fix a cbrtl.c regression and remove x87 precision settingnsz-27/+6
2012-03-20optimize scalbn familyRich Felker-7/+46
the fscale instruction is slow everywhere, probably because it involves a costly and unnecessary integer truncation operation that ends up being a no-op in common usages. instead, construct a floating point scale value with integer arithmetic and simply multiply by it, when possible. for float and double, this is always possible by going to the next-larger type. we use some cheap but effective saturating arithmetic tricks to make sure even very large-magnitude exponents fit. for long double, if the scaling exponent is too large to fit in the exponent of a long double value, we simply fallback to the expensive fscale method. on atom cpu, these changes speed up scalbn by over 30%. (min rdtsc timing dropped from 110 cycles to 70 cycles.)
2012-03-19remquo asm: return quotient mod 8, as intended by the specRich Felker-17/+26
this is a lot more efficient and also what is generally wanted. perhaps the bit shuffling could be more efficient...
2012-03-19use alternate formula for acos asm to avoid loss of precisionRich Felker-3/+11
2012-03-19Merge remote branch 'nsz/master'Rich Felker-806/+654
2012-03-19fix exp asmRich Felker-23/+22
exponents (base 2) near 16383 were broken due to (1) wrong cutoff, and (2) inability to fit the necessary range of scalings into a long double value. as a solution, we fall back to using frndint/fscale for insanely large exponents, and also have to special-case infinities here to avoid inf-inf generating nan. thankfully the costly code never runs in normal usage cases.
2012-03-19code cleanup of named constantsnsz-623/+513
zero, one, two, half are replaced by const literals The policy was to use the f suffix for float consts (1.0f), but don't use suffix for long double consts (these consts can be exactly represented as double).
2012-03-19fix remainder*.c: remove useless long double castnsz-15/+6
2012-03-19don't try to create non-standard denormalization signalnsz-7/+4
Underflow exception is only raised when the result is invalid, but fmod is always exact. x87 has a denormalization exception, but that's nonstandard. And the superflous *1.0 will be optimized away by any compiler that does not honor signaling nans.
2012-03-19Merge remote branch 'nsz/master'Rich Felker-80/+3
2012-03-19new modff.c code, fix nan handling in modflnsz-44/+26
2012-03-19use scalbn or *2.0 instead of ldexp, fix fmalnsz-101/+102
Some code assumed ldexp(x, 1) is faster than 2.0*x, but ldexp is a wrapper around scalbn which uses multiplications inside, so this optimization is wrong. This commit also fixes fmal which accidentally used ldexp instead of ldexpl loosing precision. There are various additional changes from the work-in-progress const cleanups.
2012-03-19fix long double const workaround in cbrtlnsz-5/+1
2012-03-19don't inline __rem_pio2l so the code size is smallernsz-9/+1
2012-03-19minor fix in __tanl (get sign properly)nsz-2/+1
2012-03-19bug fix: wrong opcode for writing long longRich Felker-2/+2
2012-03-19remove long double const workaroundsnsz-79/+2
Some long double consts were stored in two doubles as a workaround for x86_64 and i386 with the following comment: /* Long double constants are slow on these arches, and broken on i386. */ This is most likely old gcc bug related to the default x87 fpu precision setting (it's double instead of double extended on BSD).
2012-03-19fix erfl wrapper for long double==double casensz-1/+1
2012-03-19asm for log1pRich Felker-0/+45
2012-03-19asm for log2Rich Felker-0/+21
2012-03-19asm for remquoRich Felker-0/+43
this could perhaps use some additional testing for corner cases, but it seems to be correct.
2012-03-19optimize exponential asm for i386Rich Felker-58/+77
up to 30% faster exp2 by avoiding slow frndint and fscale functions. expm1 also takes a much more direct path for small arguments (the expected usage case).
2012-03-19Merge remote branch 'nsz/master'Rich Felker-163/+208
2012-03-19fix broken modf family functionsRich Felker-27/+66
2012-03-19asm for modf functionsRich Felker-0/+45
2012-03-19simplify scalbn*.c implementationsnsz-152/+65
The old scalbn.c was wrong and slow, the new one is just slow. (scalbn(0x1p+1023,-2097) should give 0x1p-1074, but the old code gave 0)