AMBER Archive (2000)

Subject: Re: amber6.0 timing

From: jim caldwell (caldwell_at_heimdal.compchem.ucsf.edu)
Date: Wed Nov 01 2000 - 15:13:59 CST

These are Amber6 sander/PME runs on various machines which
I did before we decided to purchase ES40's.

jim

-----------------------------------------------------------------------
Compaq

ES40 667MHz EV67

sander.out.pme.1.es40a:| Nonsetup wallclock 1839 seconds
sander.out.pme.2.es40a:| Nonsetup wallclock 1022 seconds
sander.out.pme.4.es40a:| Nonsetup wallclock 577 seconds
sander.out.pme.8.es40a:| Nonsetup wallclock 365 seconds

GS160 667MHz EV67

1.wfdemo2.eng.lkg.dec.com:| Nonsetup wallclock 2027 seconds
2.wfdemo2.eng.lkg.dec.com:| Nonsetup wallclock 1045 seconds
4.wfdemo2.eng.lkg.dec.com:| Nonsetup wallclock 543 seconds
8.wfdemo2.eng.lkg.dec.com:| Nonsetup wallclock 304 seconds
12.wfdemo2.eng.lkg.dec.com:| Nonsetup wallclock 234 seconds
16.wfdemo2.eng.lkg.dec.com:| Nonsetup wallclock 194 seconds

-----------------------------------------------------------------------
HP

L440: L-class, 4cpu 440mhz 8gb:

sander.out.pme.1.barbara:| Nonsetup wallclock 3319 seconds
sander.out.pme.2.barbara:| Nonsetup wallclock 1824 seconds
sander.out.pme.4.barbara:| Nonsetup wallclock 1099 seconds

N440: N-class, 8cpu 440mhz 16gb:

sander.out.pme.1.bo:| Nonsetup wallclock 3149 seconds
sander.out.pme.2.bo:| Nonsetup wallclock 1676 seconds
sander.out.pme.4.bo:| Nonsetup wallclock 970 seconds
sander.out.pme.8.bo:| Nonsetup wallclock 763 seconds

N552: N-class, 8cpu 552mhz 16gb:

sander.out.pme.1.pamela:| Nonsetup wallclock 2564 seconds
sander.out.pme.2.pamela:| Nonsetup wallclock 1370 seconds
sander.out.pme.4.pamela:| Nonsetup wallclock 818 seconds
sander.out.pme.8.pamela:| Nonsetup wallclock 669 seconds

-----------------------------------------------------------------------
SGI

ORIGIN 200:

        original version Modified version ucsf binary
        ----------------- ----------------- --------------
nproc time speedup(%P) time speedup(%P) time speedup

  1 2653 2186 2765
  2 1380 1.92 (95.83) 1148 1.90(94.74)
  4 821 3.23 (92.05) 686 3.19(91.54)

SGI 2100:

        original version Modified version ucsf binary
        ----------------- ----------------- --------------
nproc time speedup(%P) time speedup(%P) time speedup

  1 2670 2212 2804
  2 1394 1.92 (95.83) 1165 1.90(94.74)
  4 823 3.24 (92.18) 688 3.22(91.93)

On Wed, 1 Nov 2000, Andrew Petersen wrote:

>
> 1) Has anyone installed Amber6.0 on a Compaq ES Alpha machine?
>
> 2) Does anyone have Amber6.0 timing measurements for Compaq ES Alpha
> machines, SGI RS10000 or any Beowulf cluster?
>
>
> Andrew Petersen
>
>
>

----------------------------------------------------------------------------
James W. Caldwell (voice) 415-476-8603
Department of Pharmaceutical Chemistry (fax) 415-502-1411
Mail Stop 0446 (email) caldwell_at_heimdal.ucsf.edu
513 Parnassus Avenue (video) farbauti.compchem.ucsf.edu
University of California (netmeeting)
San Francisco, CA 94143-0446
----------------------------------------------------------------------------