This is the mail archive of the
`gcc-patches@gcc.gnu.org`
mailing list for the GCC project.

Index Nav: | [Date Index] [Subject Index] [Author Index] [Thread Index] | |
---|---|---|

Message Nav: | [Date Prev] [Date Next] | [Thread Prev] [Thread Next] |

Other format: | [Raw text] |

*From*: Giuliano Augusto Faulin Belinassi <giuliano dot belinassi at usp dot br>*To*: Jeff Law <law at redhat dot com>*Cc*: Richard Biener <richard dot guenther at gmail dot com>, GCC Patches <gcc-patches at gcc dot gnu dot org>*Date*: Tue, 21 Aug 2018 17:08:20 -0300*Subject*: Re: patch to bug #86829*References*: <CAEFO=4AiqFvHH5sb1xWguEoSY2osH+NZJzWELkHqefEbgTd_6g@mail.gmail.com> <5c2ff99f-f23a-97d6-4bbe-a1cc20e649fa@redhat.com> <CAFiYyc32AagkKsYfA-EvPOWEr3oryDu0GmfDvBybSHUsLeaSbA@mail.gmail.com> <5036aefe-ad98-15c4-51cd-b3122d38faec@redhat.com>

> Just as an example, compare the results for > x = 0x1.fffffffffffffp1023 Thank you for your answer and the counterexample. :-) > If we had useful range info on floats we might conditionalize such > transforms appropriately. Or we can enable it on floats and do > the sqrt (x*x + 1) in double. I think I managed to find a bound were the transformation can be done without overflow harm, however I don't know about rounding problems, however Suppose we are handling double precision floats for now. The function x/sqrt(1 + x*x) approaches 1 when x is big enough. How big must be x for the function be 1? Since sqrt(1 + x*x) > x when x > 1, then we must find a value to x that x/sqrt(1 + x*x) < eps, where eps is the biggest double smaller than 1. Such eps must be around 1 - 2^-53 in ieee double because the mantissa has 52 bits. Solving for x yields that x must be somewhat bigger than 6.7e7, so let's take 1e8. Therefore if abs(x) > 1e8, it is enough to return copysign(1, x). Notice that this arguments is also valid for x = +-inf (if target supports that) because sin(atan(+-inf)) = +-1, and it can be extended to other floating point formats.The following test code illustrates my point: https://pastebin.com/M4G4neLQ This might still be faster than calculating sin(atan(x)) explicitly. Please let me know if this is unfeasible. :-) Giuliano. On Tue, Aug 21, 2018 at 11:27 AM, Jeff Law <law@redhat.com> wrote: > On 08/21/2018 02:02 AM, Richard Biener wrote: >> On Mon, Aug 20, 2018 at 9:40 PM Jeff Law <law@redhat.com> wrote: >>> >>> On 08/04/2018 07:22 AM, Giuliano Augusto Faulin Belinassi wrote: >>>> Closes bug #86829 >>>> >>>> Description: Adds substitution rules for both sin(atan(x)) and >>>> cos(atan(x)). These formulas are replaced by x / sqrt(x*x + 1) and 1 / >>>> sqrt(x*x + 1) respectively, providing up to 10x speedup. This identity >>>> can be proved mathematically. >>>> >>>> Changelog: >>>> >>>> 2018-08-03 Giuliano Belinassi <giuliano.belinassi@usp.br> >>>> >>>> * match.pd: add simplification rules to sin(atan(x)) and cos(atan(x)). >>>> >>>> Bootstrap and Testing: >>>> There were no unexpected failures in a proper testing in GCC 8.1.0 >>>> under a x86_64 running Ubuntu 18.04. >>> I understand these are mathematical identities. But floating point >>> arthmetic in a compiler isn't nearly that clean :-) We have to worry >>> about overflows, underflows, rounding, and the simple fact that many >>> floating point numbers can't be exactly represented. >>> >>> Just as an example, compare the results for >>> x = 0x1.fffffffffffffp1023 >>> >>> I think sin(atan (x)) is well defined in that case. But the x*x isn't >>> because it overflows. >>> >>> So I think this has to be somewhere under the -ffast-math umbrella. >>> And the testing requirements for that are painful -- you have to verify >>> it doesn't break the spec benchmark. >>> >>> I know Richi acked in the PR, but that might have been premature. >> >> It's under the flag_unsafe_math_optimizations umbrella, but sure, >> a "proper" way to optimize this would be to further expand >> sqrt (x*x + 1) to fabs(x) + ... (extra terms) that are precise enough >> and not have this overflow issue. >> >> But yes, I do not find (quickly skimming) other simplifications that >> have this kind of overflow issue (in fact I do remember raising >> overflow/underflow issues for other patches). >> >> Thus approval withdrawn. > At least until we can do some testing around spec. There's also a patch > for logarithm addition/subtraction from MCC CS and another from Giuliano > for hyperbolics that need testing with spec. I think that getting that > testing done anytime between now and stage1 close is sufficient -- none > of the 3 patches is particularly complex. > > >> >> If we had useful range info on floats we might conditionalize such >> transforms appropriately. Or we can enable it on floats and do >> the sqrt (x*x + 1) in double. > Yea. I keep thinking about what it might take to start doing some light > VRP of floating point objects. I'd originally been thinking to just > track 0.0 and exceptional value state. But the more I ponder the more I > think we could use the range information to allow transformations that > are currently guarded by the -ffast-math family of options. > > jeff >

**Follow-Ups**:**Re: patch to bug #86829***From:*Jeff Law

**References**:**patch to bug #86829***From:*Giuliano Augusto Faulin Belinassi

**Re: patch to bug #86829***From:*Jeff Law

**Re: patch to bug #86829***From:*Richard Biener

**Re: patch to bug #86829***From:*Jeff Law

Index Nav: | [Date Index] [Subject Index] [Author Index] [Thread Index] | |
---|---|---|

Message Nav: | [Date Prev] [Date Next] | [Thread Prev] [Thread Next] |