Server Fault is a question and answer site for system and network administrators. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

Our department has $16k to buy computer hardware for high performance computation. The purpose is to do simulation and data analysis using multithreaded applications written with MPI or GridGain, so we want low latency between the cores. Storage is not the highest priority.

We've no experience of server grade hardware before, but are familiar with clustering normal PCs running Debian.

People are keen on blades, but I'm not sure they will provide the maximum bang for buck. We are not a data centre and this $16k is likely to be the most money we will have to spend on server hardware for a long while. What other options should we be considering?

share|improve this question
up vote 1 down vote accepted

Blades suck price wise.... problem is the blades are more expensive than the normal server AND you need to buy the center on top.


They have multi node computers. 1 rack unit high = 2 separate computers, 2 rack units high = 4 separate nodes. Node = separate motherboard, discs etc. A LOT cheaper than blades.

For cross-Node-Communication your only real "low latency" technology is Infiniband (Supermicro has boards with Infiniband on board). Everything else is a number of manitudes higher latency.

share|improve this answer
Excellent answer, with lots to think about. Thanks – user56620 Oct 10 '10 at 19:30

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.