[Bioclusters] what motherboards are people using for dual Athlon boxes?

Chris Dagdigian bioclusters@bioinformatics.org
Thu, 06 Mar 2003 12:54:46 -0500

This has been a very informative thread. I appreciate all the feedback!

I am on a speaking relationship with the client; in fact one or more of 
them are on this list I suspect.

The central issue is that the current project is for a (< 64 CPU) 'phase 
I' general engineering/technical computing resource for a university 
commmunity. The initial users will be doing physics, math, CFD, 
bioinformatics etc. etc. Lots of CPU/Memory/IO-bound apps.

Currently there are no real use-cases or even representative codes to 
benchmark. This make choosing an architecture (AMD vs Intel) hard as 
usually the choice is pretty clear once you run your local set of apps.

Actually making sure the operational stuff is nailed down with respect 
to scalability and management/operational burden is more important to 
get right the first time than the CPU selection which can vary over time.

I suspect that what will happen is that the group will deploy a 
reasonably sized cluster of 2-way SMP nodes. Whatever CPU is chosen for 
Phase I may change for the next expansion phase since by then they will 
have real usage data to work with.


William Park wrote:
> If that's the specs they want from contractors, then you have no choice.
> Tyan for dual-AMD, and Tyan/Supermicro/Intel for dual-Xeon.  But, if you
> have speaking relationship with the client, perhaps you can convince
> them that single-CPU is better route to go, ie. more motherboard choice,
> faster cpu, faster development, no memory/cache contention, etc.