Many moons ago, we used MDBNCH as a performance test case for new compilers and platforms. It is a molecular dynamics benchmark test. Large caches help it. So do fast instruction issue rates. And bloody fast FP execution units.
For a long time I had wondered when we would see the first crossing of 1 second to complete this benchmark. It was getting closer and closer.
So this evening, with gfortran, I built it with -O3. Ran it on our main server, an Opteron 2212 based JackRabbit deskside server.
COMPLETE BENCHMARK EXECUTION TIME : 1.770000 CP SECONDS.
Cool. What about the Shanghai box? 2.3 GHz, low power chip.
COMPLETE BENCHMARK EXECUTION TIME : 1.024063 CP SECONDS.
But what about Nehalem? Deskside and the server chip? We have an X5550 Pegasus system we are building in the lab, but the OS isn’t loaded yet, so this will have to wait. Lets try the desktop Nehalem for the moment.
COMPLETE BENCHMARK EXECUTION TIME : 0.604908 CP SECONDS.
There we have it. Beat one second. On a machine that is … oh … $1500 to put together. While 10 years ago, the $100k machines had trouble breaking 10s.