[gmx-developers] mdrun not working with VERSION 3.3_beta_20050202

[Erik Lindahl]

Hi,

Raw performance (ps/day or hour) is fine, and include both the clock 
frequency and the amount of cache (cat /proc/cpuinfo). If it seems to 
be significantly slower than it should for the CPU I might get back and 
ask about more details :-)


On Feb 15, 2005, at 4:13 PM, Florian Haberl wrote:

> Hi,
>
>
>> David is currently running some tests on the Thunder system at LLNL,
>> but since both compilers and CPUs might differ a bit I'm certainly
>> interested in both single and double precision results to make sure
>> we're not having performance problems anywhere (e.g. due to different
>> cache latencies, the ia64 code is pretty sensitive to that).
>>
>
> Do you want all output data or some special debug flag to run?
> Or are the benchmark results enough.
>
>
> Are there than soon any news about performance on supercluster as i 
> asked
> here: 
> http://www.gromacs.org/pipermail/gmx-users/2005-January/013903.html
>

We're still working on improving the parallelization together with some 
people from the Garchingen supercomputing center, but it's looking 
promising - it should start to show up in CVS as soon as 3.3 is forked 
off.

Cheers,

Erik

-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/pkcs7-signature
Size: 2363 bytes
Desc: not available
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-developers/attachments/20050215/f119e0d5/attachment.p7s>