Re: CPU GPU comparison

From: Axel Kohlmeyer (akohlmey_at_gmail.com)
Date: Fri Jul 16 2010 - 08:39:18 CDT

On Fri, Jul 16, 2010 at 4:13 AM, francesco oteri
<francesco.oteri_at_gmail.com> wrote:
> I refer to numerical comparison. I would like to know if there are
> significant difference between calculated energy on CPU or GPU

not really. since you'll be doing floating point math and
summing up numbers of different magnitude in different
order, trajectories will diverge eventually. same as when
using a different number or kind of processors.

that divergence is usually irrelevant, since you don't
care about perfectly exact reproduction of a specific
trajectory, but about sampling of relevant phase space.

cheers,
    axel.

> 2010/7/16 bahare bamdad <baharebamdad_at_gmail.com>
>>
>> Dear Framcesco,
>>
>> I did a test for a moderate system to check the efficiency of CUDA enabled
>> NAMD. The results are in below.
>>
>> wallclock time for 1 CPU=187895.234375
>> wallclock time for 1 GPU=20768.080078
>> wallclock time for 2 GPU=11241.884766
>> wallclock time for 3 GPU=8318.795898
>>
>> Best Regards,
>>
>> Bahareh Bamdad
>>
>>
>> On Tue, Jul 13, 2010 at 9:17 PM, francesco oteri
>> <francesco.oteri_at_gmail.com> wrote:
>>>
>>> Dear NAMD users,
>>> I would like to know if someone did a comparison between CUDA enabled
>>> NAMD and classic CPU program.
>>> I would like to know if anyone detected significant difference between
>>> the CUDA and non-CUDA.
>>>
>>> Francesco
>>
>
>
>
> --
> Cordiali saluti, Dr.Oteri Francesco
>

-- 
Dr. Axel Kohlmeyer    akohlmey_at_gmail.com
http://sites.google.com/site/akohlmey/
Institute for Computational Molecular Science
Temple University, Philadelphia PA, USA.

This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:54:19 CST