[macstl-dev] Fwd: running on intel emt64

[Glen Low]

All:

FYI, macstl benchmarks on an Intel EM64T.

Begin forwarded message:

> From: "Damien DOLIMIER"
> Date: 3 October 2005 1:34:35 AM
> To: "'MacSTL'"
> Subject: RE: running on intel emt64
>
>
> The clock speed is 3.2Ghz with HT on and 1MB cache, of course! you  
> may use
> this benchmark for your mailing list.
>
>
> -----Original Message-----
> From: MacSTL
> Sent: Sunday, October 02, 2005 2:28 AM
> To: Damien Dolimier
> Subject: Re: running on intel emt64
>
> Damien:
>
> The benchmarks look good. As expected only the complex multiply-add,
> predicate (?), slicing and power are not optimized with SSE so their
> multiplier over raw is close to 1.
>
> May I ask what the clock speed of your Pentium 840? Can I have  
> permission to
> post this to our public mailing list: http://
> www.pixelglow.com/lists/listinfo/macstl-dev?
>
> On 28/09/2005, at 3:47 PM, Damien Dolimier wrote:
>
>
>> Hi,
>> thanks for your prompt reply,
>> the only modification I have made is -march=athlon64, the option
>> -fno-unit-at-time is correct
>>
>> FYI, on my pentium 840:
>> Benching multiply add:  555555/sec  1.55556*std  1.5*raw.
>> Benching inner product:  666666/sec  1.73333*std  1.66667*raw.
>> Benching polynomial:  238095/sec  1.35714*std  1.07143*raw.
>> Benching hypotenuse:  227272/sec  2.68182*std  2.61364*raw.
>> Benching complex multiply add:  50505/sec  1.13131*std  0.388889*raw.
>> Benching predicate:  2500000/sec  16*std  5.75*raw.
>> Benching slicing:  250000/sec  0.925*std  1*raw.
>> Benching power:  5099/sec  1.00561*std  0.99745*raw.
>> Benching trigonometric:  16474/sec  31.8386*std  37.6804*raw.
>>
>> thanks for your help,
>> Damien


Cheers, Glen Low


---
pixelglow software | simply brilliant stuff
www.pixelglow.com
aim: pixglen

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.pixelglow.com/lists/archive/macstl-dev/attachments/20051003/6fc77c33/attachment.html