« Fast and Cheap, pick two: Real data for Multi-threaded S3 Transfers | Main | CycleCloud BigScience Challenge 2011 »

September 19, 2011

TrackBack

TrackBack URL for this entry:
http://www.typepad.com/services/trackback/6a00e0097e0bd58833015391b8f602970b

Listed below are links to weblogs that reference New CycleCloud HPC Cluster Is a Triple Threat: 30000 cores, $1279/Hour, & Grill monitoring GUI for Chef:

Comments

Gordon Shephard

What were your concerns with regards to balancing IO/Memory/DiskStorage with the cluster - did amazon's large amounts of memory provide you some benefits that you needed versus going with a more traditional GPU solution? Did you need/utilize the extra memory/disk storage you got from all those instances - or was this a purely CPU bound process, and did the memory and disk storage for each instance go (mostly) untouched?

mrb

Molecular modeling is known to be generally very well suited to GPUs. Why not run this workload on 1/10th the number of machines, eg. by using EC2's GPU instances?

Cycle Computing

Hi Gordon & mrb:
We wish there was such a thing as a "more traditional GPU solution". Although they aren't new, GPUs are not supported by all software algorithms. Many still need to be ported. This is one of those cases, where the preferred molecular dynamics model was (1) proprietary and (2) had no GPU support.

mrb

I am surprised that this "top 5 pharma" company doesn't seem to have an in-house team of GPU developers already working on GPU-accelerated molecular modeling for them. Or, do they?

They are billion dollar businesses. It seems like they could gain an edge over the competition, instead of being locked into third-party proprietary CPU-only apps...

Adam

Why standard extra large and not hi-cpu extra large instances? For the same price, you get 47% the RAM, but 250% the compute units.

Jason

The Top 5 Pharma has a vendor with a valuable molecular dynamics model, and this model isn't available on GPU from that vendor, so in-house programmers don't matter...

CycleComputing

Adam: We did use high-cpu xlarge. It is the best CPU/$ if RAM isn't a limitation.

Verify your Comment

Previewing your Comment

This is only a preview. Your comment has not yet been posted.

Working...
Your comment could not be posted. Error type:
Your comment has been saved. Comments are moderated and will not appear until approved by the author. Post another comment

The letters and numbers you entered did not match the image. Please try again.

As a final step before posting your comment, enter the letters and numbers you see in the image below. This prevents automated programs from posting comments.

Having trouble reading this image? View an alternate.

Working...

Post a comment

Comments are moderated, and will not appear until the author has approved them.