[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Discuss-gnuradio] 3DNow fir test failing under debian with latest u
From: |
Krzysztof Kamieniecki |
Subject: |
Re: [Discuss-gnuradio] 3DNow fir test failing under debian with latest unstable updates |
Date: |
Mon, 22 Aug 2005 08:35:08 -0400 |
User-agent: |
Internet Messaging Program (IMP) 3.2.2 |
Quoting Eric Blossom <address@hidden>:
> On Mon, Aug 22, 2005 at 12:44:58AM -0400, Krzysztof Kamieniecki wrote:
> > Hello All,
> >
> > It seems like something may have broken in the 3Dnow fir code (on Debian
> > Unstable), it may be GNURadio or it may be GCC. Has anybody seen this?
> >
> >
> > SSE seems to be faster on this processor anyways, is there a recommended
> > way of selecting SSE over 3Dnow! ?
>
> The test code has an alignment problem under gcc 4.x (that we should fix...)
> When run within the normal gnu radio framework, the alignment
> constraint is met and the right answer is computed.
>
> It's safe to ignore the error.
>
> I'm open for suggestions as to how we should select SSE vs 3DNow.
> Shall we benchmark the first time through and store the answer?
>
I would be happy with a manual selection in a ~/.gnuradio.rc file.
simd_instruction_set={auto,none,sse,3dnow,altivec,...}
//auto would have the same behaviour as the current method
> Regarding:
>
> > model name : AMD Athlon(tm) 64 Processor 3700+
> > stepping : 10
> > cpu MHz : 797.986
> > cache size : 1024 KB
>
> It appears that though you have an AMD64 machine, GNU Radio is
> being compiled as a 32-bit app. Can you send me the output of
> uname -a and configure?
>
I will do this when I get home tonight. I'm running a 32bit linux image. I don't
want to experiment with 64bit debian just yet, plus the "production" system I
am using for this project is a 32bit Intel processor.
> Also, is the cpu MHz output right? 800?? Seems way low, unless you're
> in some kind of power saving mode (which wouldn't be good for
> benchmarking!)
>
It's a compaq r3000z amd64 2.4GHz laptop (desktop replacement), I'll look into
forcing the processor clock to maximum before running the benchmarks.
> Thanks,
> Eric
>
Regards,
Krys