[Beowulf] AMD Opteron Linpack Benchmark

[Mike Davis]

Jim,

I thought that I had read that you removed memory from some of the 
nodes. I must have been mistaken on that point.

Mike Davis


james pepin wrote:
> whats the question??
> 
> the effeciency we saw improved with increased memory, also improves  
> with more cores on a node...
> 
> 
> jim
> 
> On Sep 25, 2006, at 10:39 AM, Daniel Kidger wrote:
> 
>> Mike (or indeed anybody else who might know)
>>
>> Can you post a reference to the list for Jim Pepin's ppt ?
>> Google didn't come up with anything useful
>>
>> Closest might be "Building and Benchmarking a 10 TF Linux Cluster"  
>> Jim Pepin,
>> USC at Sun HPC consortium 2005 but can't find any online resources  
>> for this ?
>>
>> Daniel
>>
>> Dr. Daniel Kidger, Technical Consultant, ClearSpeed Technology plc,  
>> Bristol,
>> UK
>> E: daniel.kidger at clearspeed.com
>> T: +44 117 317 2030
>> M: +44 7738 458742
>> "Write a wise saying and your name will live forever." - Anonymous.
>>
>>
>> -----Original Message-----
>> From: beowulf-bounces at beowulf.org [mailto:beowulf- 
>> bounces at beowulf.org] On
>> Behalf Of Mike Davis
>> Sent: 24 September 2006 17:48
>> To: Eric W. Biederman
>> Cc: Lai Dragonfly; beowulf at beowulf.org
>> Subject: Re: [Beowulf] AMD Opteron Linpack Benchmark
>>
>> Also, check out the changes that Jim Pepin at USC made to his  cluster to
>> produce more teraflops. There is a Powerpoint on the web that gives  some
>> of the details. If I remember correctly he actually removed RAM  from the
>> nodes and made a few other changes.
>>
>> Mike Davis
>>
>> Eric W. Biederman wrote:
>>
>>> "Lai Dragonfly" <poknam at gmail.com> writes:
>>>
>>>
>>>
>>>> Hi,
>>>> There is a rank 367 which achieve 79.97% efficiency.
>>>>
>>>>
>>>> i'm using Infiniband to connect 2 nodes. each node has 8G memory  
>>>> and dual
>>
>> AMD
>>
>>>> opteron 250.
>>>> i just tried Intel Compiler + GOTO, it just got a 60% efficiency  
>>>> which have
>>
>> a
>>
>>>> large difference with my target.
>>>> maybe i need to try Pathscale or PGI compiler.
>>>> any suggestion?
>>>> thanks a lot.
>>>>
>>>>
>>>
>>> Start with single node runs, then work you way up from there.
>>>
>>> Eric
>>> _______________________________________________
>>> Beowulf mailing list, Beowulf at beowulf.org
>>> To change your subscription (digest mode or unsubscribe) visit
>>
>> http://www.beowulf.org/mailman/listinfo/beowulf
>>
>>>
>>>
>>
>> _______________________________________________
>> Beowulf mailing list, Beowulf at beowulf.org
>> To change your subscription (digest mode or unsubscribe) visit
>> http://www.beowulf.org/mailman/listinfo/beowulf