MDR Benchmarks

I have taken a bit of a break from blogging this summer and promise to be more vigilant the rest of the year. Hopefully this will the first of many new posts over the coming weeks.

We just received some new desktop computers with the new Intel Quad Core Xeon chips and wanted to compare their performance with MDR to one of our old machines that is three years old now. The task was to run MDR 1.1 from the command line on the MDR sample data using the default settings. We expected the newer machines to run MDR faster than the older machines. It is important to remember that the MDR software automatically detects multiple processors and hyperthreading and will run in parallel to speed up the analysis. Here are the results for the two machines both running Linux.

1) Old machine (circa Sept. 2004)

Specs: 2 x Intel(R) Xeon(TM) CPU 3.00GHz, hyperthreaded
Speed: 9.326 seconds

2) New machine (circa Sept. 2007)

Specs: 4 x Intel(R) Xeon(R) CPU X5365 @ 3.00GHz
Speed: 5.764 seconds

Both machines had MDR running in parallel on four threads but the Quad Core machine was much faster even though the chip speed is the same. I highly recommend running MDR on a multi-processor machine.