Re: gcc/libm floating-point bug?

From: Dofri Jonsson (imp_at_hell.is)
Date: 05/20/03

  • Next message: Steve Kargl: "Re: XFree86 process gets stuck in *Giant" 
    To: Jon Lido <jlido@goof.com>
Date: 20 May 2003 18:19:56 +0000

    See PR bin/43299: march=pentium4 miscompiles msun/src/e_pow.c for more
    details:
    http://www.freebsd.org/cgi/query-pr.cgi?pr=43299

    On Tue, 2003-05-20 at 14:25, Jon Lido wrote:
    > I've been running 5-CURRENT on my laptop for about a week now. I believe I
    > may have found a bug in FreeBSD's gcc floating-point code generation, or
    > perhaps more likely, in the math library.
    >
    > Like most users, I don't do much with floating point, so it took me a while to
    > narrow this one down. Where I've noticed the problem is playing when lossy
    > audio. When playing MP3s with artsd, xmms, or mpg123, I can get extremely
    > noisy output (like static), with the audio distorted, but recognizable
    > underneath. Ogg Vorbis files don't play at all, and simply crash artsd and
    > xmms. I would have dismissed this as an audio problem, but I am able to play
    > wav and shn audio (which uses integer math for decompression, I believe)
    > fine. Additionally, I can decode and play MP3s fine with mpg321, which uses
    > an integer decoder.
    >
    > I know my floating point hardware is fine, since I used to run xmms and artsd
    > fine under Linux before I switched the machine to FreeBSD.
    >
    > I'd appreciate any help with further isolating and reproducing the bug. Has
    > anyone else experienced this problem?
    >
    > -Jon
    >
    > _______________________________________________
    > freebsd-current@freebsd.org mailing list
    > http://lists.freebsd.org/mailman/listinfo/freebsd-current
    > To unsubscribe, send any mail to "freebsd-current-unsubscribe@freebsd.org"

    _______________________________________________
    freebsd-current@freebsd.org mailing list
    http://lists.freebsd.org/mailman/listinfo/freebsd-current
    To unsubscribe, send any mail to "freebsd-current-unsubscribe@freebsd.org"

  • Next message: Steve Kargl: "Re: XFree86 process gets stuck in *Giant"

    Relevant Pages

    • gcc/libm floating-point bug?
      ... Like most users, I don't do much with floating point, so it took me a while to ... noisy output, with the audio distorted, but recognizable ... I would have dismissed this as an audio problem, but I am able to play ... I'd appreciate any help with further isolating and reproducing the bug. ...
      (freebsd-current)
    • Re: Audio processing -- fixed or floating point?
      ... > audio processing code that they are putting in Windows Vista (the next ... > floating point is efficient (the old int16 code dates back to when they ... then maybe you need to be careful with those floats. ...
      (comp.dsp)
    • Re: time.clock() going backwards??
      ... including printing the reprof the floating ... it so to ignore these bogus reads, and just reading off again. ... Is it possible this to be a bug in Python itself ... Are you running on a multi-core or a multi-processor machine? ...
      (comp.lang.python)
    • Re: An strange point
      ... I wrote a program which has a bug when the value of u is greater than ... The real error is that you do not understand the technical issues of floating ... otherwise know as finite precision computation. ... only interested in multiples of 0.1. ...
      (comp.lang.fortran)
    • Re: Cost of multiplication relative to addition
      ... I have never had occasion to work with a floating point DSP -- all my ... sticking to integer math for many of the reasons you discussed. ... that will ever get the component light enough! ...
      (comp.dsp)