Latency vs Bandwidth over Infiniband

From: Dow_Hurst (dhurst_at_mindspring.com)
Date: Fri Aug 17 2007 - 17:37:07 CDT

I'd like to get advice on how latency affects scaling of molecular dynamics codes versus total bandwidth of the interconnect card. We use NAMD as the molecular dynamics code and right now, we have a chance to upgrade and add nodes to our cluster using Infiniband. I've found that NAMD was coded to be latency tolerant, however, I'd like to scale up to 64 cores and beyond. I'm going blind reading IB card specs, performance benchmarks, and searching Google. I'd love some advice from someone who knows whether a consistent very low latency IB card, such as the Infinipath QLE7140, is better/worse for NAMD than a higher latency but higher bandwidth card such as the QLE7240? I can tell that Lonestar at TACC has great NAMD performance but I can't tell what IB card is used. I imagine that switch performance plays a large role.

Question:
1. Will higher latency, say 3us, reduce efficiency versus 1.5us even if
   higher bandwidth is available?

2. Low latency cards don't help if the switch isn't low latency too, correct?
   (that seems pretty obvious!)

3. QLogic's whitepaper on Infiniband shows latency as very important to
   scaling. How true is this for NAMD?

Thanks for your time,
Dow

No sig.

This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:46:49 CST