Messages in this thread | | | Date | Thu, 25 Jun 1998 10:13:57 -0500 (CDT) | From | Robert HYATT <> | Subject | Re: quad ppro Compaq proliant 5000 problems with 2.1.106-ac4 was: Re: Any SMP people out there with SCSI CD ROMs? |
| |
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 if I recall correctly...
that benchmark is very memory-intensive not using cache very well at all. I ran one copy, then two copies, etc... each extra copy added about 7 seconds of cpu time, caused by memory bandwidth limits. This machine uses 4-way interleaving, and seems to scale quite well. I have a threaded application that can smoke all 4 cpus for hours and it gets enough hash hits that I don't see any degradation at all when I run 4 copies of this program, vs 1 copy, vs 1 copy with 4 threads...
So it appears to work quite well for me. I do have gross problems with the pthread mutex stuff, so I don't use them, having resorted to spinlocks for better performance.
Robert Hyatt Computer and Information Sciences hyatt@cis.uab.edu University of Alabama at Birmingham (205) 934-2213 115A Campbell Hall, UAB Station (205) 934-5473 FAX Birmingham, AL 35294-1170
On Thu, 25 Jun 1998 linker@nightshade.ml.org wrote:
> > > On Wed, 24 Jun 1998, Arnaldo Carvalho de Melo wrote: > [snip] > > [mike@sckott mike]$ uname -a > > Linux sckott 2.1.106 #1 SMP Tue Jan 19 03:14:07 2038 i686 unknown > > > > > > - Arnaldo > > Were you having fun with the date or is something wrong? > > Think you could so some bencharks on how well linux does on a 4 cpu > system? Few people use Linux on 4 cpu boxen, so performance numbers are > hard to find. Linux performs very well on two cpu boxen. > > Because of this lack of data, there is more ammo for the FUD bringers. > Several times I've heard that 'Linux only supports 2 cpus' or 'Linux > doesnt scale well past two cpus'. > >
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu
| |