ts-7000
[Top] [All Lists]

[ts-7000] Re: New version cross toolchain for Cygwin

To:
Subject: [ts-7000] Re: New version cross toolchain for Cygwin
From: "charliem_1216" <>
Date: Tue, 23 Jan 2007 18:44:51 -0000
--- In  Paulo Marques <> wrote:
>
> > --- In  "charliem_1216" <charliem_1216@> 
> > wrote:
> >[...]
> >> Sorry, that did not come out right in html:
> >>
> >> 2.4.26ts11      926 KIPS
> >> 2.6.20rc5       406 KIPS
> >> 2.6.20rc5-eabi  5900 KIPS (no crunch)
> >> 2.6.20rc5-eabi  8100 KIPS (with -mfpu=maverick -mfloat-abi=softfp
> >>                           -mcpu=ep9312)
> 
> Isn't this result not to flattering for maverick crunch because of 
> benchmarking arithmetic divisions too?

Yes, quite likely.  I did not know about the lack of hw FP divide (I'm
just starting to look into crunch).

> 
>  From the benchmarks posted by cirrus:
> 
> soft_float:
> ADD: calculation took        187 ms,   test rv=2.2556635255738473e+06
> SUB: calculation took        191 ms,   test rv=-2.2556635255738473e+06
> MUL: calculation took        310 ms,   test rv=7.0863759618797886e+06
> DIV: calculation took       1037 ms,   test rv=1.9758917663375622e+06
> 
> crunch compiled:
> ADD: calculation took          1 ms,   test rv=2.2556635255738473e+06
> SUB: calculation took          1 ms,   test rv=-2.2556635255738473e+06
> MUL: calculation took         25 ms,   test rv=7.0863759618797889e+06
> NRD: calculation took        670 ms,   test rv=1.9758917663375624e+06
> DIV: calculation took       1020 ms,   test rv=1.9758917663375624e+06

I will look at this code and see what I get with it.  At first glance,
their soft-float numbers look strange (ADD taking 60% of a MUL).

> 
> So even if you improve soft by a factor of 6, additions, subtractions 
> and multiplications would still be much much better using crunch. And 
> these are the basic operations you need to do matrix multiplication.

Yes, with the DIV handicap, some workloads will be much more improved
than others.  I was very happy to see the big improvment using
softfloat, without crunch.  Could that improvement be brought to 2.4
with the right libs & toolchain?  I guess it could, since crunch would
not be involved.

Regards, ...... Charlie

> -- 
> Paulo Marques
> Software Development Department - Grupo PIE, S.A.
> Phone: +351 252 290600, Fax: +351 252 290601
> Web: www.grupopie.com
> 
> "The face of a child can say it all, especially the
> mouth part of the face."
>




 
Yahoo! Groups Links

<*> To visit your group on the web, go to:
    http://groups.yahoo.com/group/ts-7000/

<*> Your email settings:
    Individual Email | Traditional

<*> To change settings online go to:
    http://groups.yahoo.com/group/ts-7000/join
    (Yahoo! ID required)

<*> To change settings via email:
     
    

<*> To unsubscribe from this group, send an email to:
    

<*> Your use of Yahoo! Groups is subject to:
    http://docs.yahoo.com/info/terms/
 

<Prev in Thread] Current Thread [Next in Thread>
Admin

Disclaimer: Neither Andrew Taylor nor the University of NSW School of Computer and Engineering take any responsibility for the contents of this archive. It is purely a compilation of material sent by many people to the birding-aus mailing list. It has not been checked for accuracy nor its content verified in any way. If you wish to get material removed from the archive or have other queries about the archive e-mail Andrew Taylor at this address: andrewt@cse.unsw.EDU.AU