Top 500 trends
Many of your questions may have already been answered in earlier discussions or in the FAQ. The search results page will indicate current discussions as well as past list serves, articles, and papers.
Richard Walsh rbw at ahpcrc.orgWed Nov 27 10:26:32 PST 2002
- Previous message: Top 500 trends
- Next message: Errors in getting pvm to run
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
On Wed, 27 Nov 2002 09:22:20 Greg Lindahl: >> > > If we're going to use metaphors, at least use fair ones. It has been >> > > years since any of the big iron computers could be compared to quality >> > > PCs in speed or most other useful system dimension as oxen to chickens. >> > >> > STREAM bandwidth is the main one. >> >> out of curiosity, do you mean "PCs have improved a lot", >> or "big iron is even faster now"? > >Neither. Try reading what I quoted; if you compare vector cpus vs. PCs >using the STREAM benchmark as your measure of speed... In support of the idea that sustained bandwidth to/from memory is still an advantage vector computers (read custom designed) have, a single X1 SSP processor should deliver over 600 64-bit MFLOPS sustained from memory (large arrays) for the stream triad. Back computing the bandwidth is left as an exercise. ;-) The machine has a cache to boost the number for smaller arrays. rbw
- Previous message: Top 500 trends
- Next message: Errors in getting pvm to run
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
More information about the Beowulf mailing list
