Jack Morgenstein wrote, >We tested out IntelMPI over uDapl (from OFED 1.0-pre1) using the PALLAS test >suite, and it worked fine -- no problems. Evidently, you are trying to use
>these new (and absent/unsupported) features. >We do appreciate that these features are very important for scalability, and >we plan to include them in the 1.1 release which will follow shortly. These new options are needed to allow Intel MPI to scale up to larger clusters, 128+. If you did not run on a large cluster you would not have seen the problems. >As I posted to ipoib-general on June 7 >( http://openib.org/pipermail/openib-general/2006-June/022725.html ) Unfortunately, due to a problem with our email server, I was not receiving openib-general emails for the last week and missed this thread or would have spoken up then. Anyway, What is the criteria/decision process for deciding that something will or will not be included. I think that we have an equal say as to what should go in and what should not. Seems like there are double standards here. You are including last minute fixes for things like the Pathscale driver after RC6, but will not allow a fix that is needed by our product. woody _______________________________________________ openib-general mailing list [email protected] http://openib.org/mailman/listinfo/openib-general To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general
