Messages in this thread | | | Date | Thu, 25 Jun 1998 16:25:51 -0400 (EDT) | From | "Robert G. Brown" <> | Subject | Re: quad ppro Compaq proliant 5000 problems with 2.1.106-ac4 was: Re: Any SMP people out there with SCSI CD ROMs? |
| |
On Thu, 25 Jun 1998, Lauri Tischler wrote:
> Robert HYATT wrote: > > > > I ran the matrix multiply benchmark... with 1, 2, 3 and 4 > > processors. > > > > The numbers I got were 52 seconds, 59 seconds, 66 seconds and 73 seconds > > Hmmm... with four cpus the benchmark took 11 sec more then with 1 cpu ?? > > One might suppose you meant the times to be : > > 52/1 1 cpu > 59/2 2 cpu > 66/3 3 cpu > 73/4 4 cpu > > Cheers... >
Actually, Robert probably meant just what he said. The CPU-mem benchmark multiplies great big matrices (much bigger than cache). The four same-size (and big!) jobs complete in PARALLEL on the four processors, testing CPU-to-memory-bus bandwidth and how well it is shared among the processors. What the numbers mean is that the benchmark is "almost" CPU bound on his (Orion chipset?) quad with its interleaved memory. I would guess that the memory bus is really only saturated when the fourth processor kicks in, as the time takes a These numbers are almost identical to those reported more than a year ago by Tom Quinn, see:
http://www.phy.duke.edu/brahma/benchmark.results
I added Qiru Zhou's latest results to this page as well. This page is kind of messy, for which I apologize. I'll try to condense the results (with any further additions that occur) into a neat table, just as soon as I have time. In this form, though, one CAN see who did what, and even approximately when.
BTW, I still have the source if anybody wants to bench new systems...
rgb
Robert G. Brown http://www.phy.duke.edu/~rgb/ Duke University Dept. of Physics, Box 90305 Durham, N.C. 27708-0305 Phone: 1-919-660-2567 Fax: 919-660-2525 email:rgb@phy.duke.edu
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu
| |