Just when I thought I understood things … Ran the original test case that we ran previously, but with the rebuilt GAMESS with a modern compiler. Ran one on the 2.66 GHz Woodcrest, one on the 2.2 GHz Opteron. Both are dual core, I don’t have a 2.4 or 2.6 GHz dual core set of Opterons to put into a machine. Used 4 way parallel on shared memory machine. Woodcrest has a 2x cache size advantage, has a 30% faster memory system, and about a 20% clock speed advantage.

So by now you know I ran an old binary and an old test on the Woodcrest and Opteron. I wasn’t impressed with the results, the hype was out of proportion with the reality. Lets assume that Intel suggests we recompile our code. I pulled down a new GAMESS (the 2-2006 variant), built it with the PGI compiler, though I had to do some option tweaking, it was otherwise OK, and ran it.

for testing/development and other purposes. Config is reasonable, with an upgrade later today. 2 x 2.66 GHz (5130) processors, 4 GB ram, nice video card (Quadro FX/4500). Installed SuSE 10.1 with updates/patches. My expectations were that this machine would positively blow the doors off of a similarly clocked Opteron (252). Given the massive hype around Woodcrest, this is what one might expect. If you are going to hype like mad, you need to be able to deliver on the hype.

We are working on some benchmarks for a customer. This is a commercial code, closed source, MPI based. Cluster in question is an Infinipath based system. I cannot say enough good things about the HTX based Infinipath systems, they are very fast, very low latency. And they come with an MPI stack. Ok, let me give you a hint where this is going. The benchmark could not run, as the code could not run on the nice super fancy Infinipath system.

About a week has passed, and a friend said something quite interesting about this. Michigan has decided not to invest in Michigan’s future. Amusing. The commercialization areas that we submitted for were to be scored on Scientific and Technical basis, Personnel, Commercialization Merit, an fund leverage. The first sentance of the review sets the tone. “The technical merit of this proposal could be phenomenal, but there is a lack of explanation of the important underlying science.

Results are in and we did not advance. Good luck to the advancers. Update: Looking over the results, not a single advanced computing project was advanced. HPC was completely ignored which runs counter to what they said they would do. Lots of automotive bits were advanced. The message for a small HPC company located in Michigan is sadly getting clearer.

We have been patiently waiting to hear whether or not we advanced to the next round. We already know the fate of one proposal which we helped on, but was not one of our own submissions. That was discarded on a technicality, basically an index entry reporting on confidential material and color pages was not pointing to the right pages according to the report back from the compliance tester. Not screened for content, as it did not pass screening for format.

I have opined here that I do not believe that Solaris will overtake Linux’s lead in HPC clusters. This does not mean that I don’t think it can have a role. Basically, imagine you have to deliver a service. Something like Google. The end user of Google doesn’t care what OS the underlying software runs on. They care about their usage. Same with the end user of Yahoo, Slashdot, Digg, … .

Folks, we are getting spammed. If something gets through, please note that I want to apologise in advance and I will handle it as soon as possible. If I miss something, fire me a note.

The machine is up. If you don’t know about Blastwave, it is a helpful resource. Once you set up your machine, mozy over there, and get all the tools you would otherwise be missing. nVidia graphics, pulled down the Solaris nVidia binary, installed it, rebooted and up it came. nVidia makes great chips and great drivers. We are lucky they are still supporting Solaris. We have an Itanium2 linux box here where the nVidia drivers are built for the old XFree86 versus xorg used in Centos 4.