Details Emerge in Microsoft's Top 500 HPC Setback
Published: November 29, 2006
by Alex Woodie
Two weeks ago, after we told you how Windows Compute Cluster Server 2003 mysteriously fell off the Top 500 list of the world's largest supercomputers following a strong showing in June, Microsoft contacted IT Jungle to let us know we got the story wrong, that Windows CCS was indeed running on one of the 30 biggest supercomputers in the world. As it turned out, we didn't get the story wrong, but there was more to it than that.
When the Top 500 organization updated its list in June 2006, Microsoft's Windows Compute Cluster Server 2003 owned two spots. The bigger of the two systems was a new Dell PowerEdge 1855 cluster run by the National Center for Supercomputing Applications (NCSA) at the at the University of Illinois that was equipped with 900 processors. The cluster, which goes by the name "Lincoln," delivered 4.1 teraflops of sustained processing capacity, and 5.7 teraflops of peak processing capacity, which made it the 130th biggest supercomputer in the world at the time. The other Windows CCS server on the list was another Dell 1855 cluster with 660 processors at the Cornell Theory Center that was capable of delivering 2.1 teraflops of sustained capacity and was ranked at 470.
When the November Top 500 list was made public earlier this month, Windows Compute Cluster Server 2003 was nowhere to be found. As it turns out, Lincoln is a dual-boot machine, and the NCSA decided to give Linux a run on a beefed-up Lincoln. When it was all said and done, the Dell cluster of 2,048 processors had scored a sustained capacity of 16.5 teraflops and a peak performance of 21.3 teraflops, which gave it a ranking of 27 on the Top 500 list. The other Windows CCS machine was dropped from the November list; a system needed to deliver 2.7 teraflops of sustained capacity to make it into the top 500.
The Microsoft spokesperson erroneously attributed Lincoln's ranking of 27 on the November list to Windows CCS. When we pointed out that number 27 was a Linux machine, the spokesperson changed gears and said the reason Windows CCS wasn't on the list was because Microsoft hadn't re-submitted the application to the Top 500 organization. A representative with the Top 500 group, Erich Strohmaier, says that's not the way it works. Generally, once a system is on the list, they stay on the list until Top 500 receives some information that the system was removed or changed.
Microsoft will undoubtedly gain spots on the Top 500 list in the future as Windows CCS gradually gains acceptance in the high performance computing community. But trying to hide the inevitable failures in the near-term is not going to earn the software giant any bonus points for good citizenship.
Windows CCS Falls Off Top 500 Supercomputer List