PII Benchmarks

**cygnussphere** · 06-17-2003, 04:05 PM

Will PII benchmarks mean anything until multiple generations are buffered?

**Welnic** · 06-17-2003, 04:32 PM

I think the whole point of the benchmark is that it is always the same test of the cpu. With the last client you could get a good idea of what was going on with any one protein if you ran it for a little while. With PII all of the generations past 0 are highly variable according to the protein that they are dealing with from past generations.

That said:

Usr time Sys time
-------- --------
Maketrj 7.31 .61
Foldtraj 83.03 16.17

Athlon 2100XP running linux 2.4

**pfb** · 06-17-2003, 05:00 PM

hmm...on my windows box, running the -bench command gives me:

Code:

========================[ Jun 17, 2003  9:54 PM ]========================
FATAL ERROR: [067:001] {mmdbapi1.c, line 1190} FindPath failed in LoadDict

my Mandrake box (Barton 2500+ at 2100MHz) gives me:

Code:

		Usr Time	Sys Time
		
Make...		2.870		0.710
Fold...		37.100		7.920

nice to check on my PCs but as stated probably not useful for cross-platform/cpu benching

**pointwood** · 06-17-2003, 05:04 PM

YEP - getting the same error on this WinXP box:

C:\distribfold>foldtrajlite -bench
[NULL_Caption] FATAL ERROR: [067:001] FindPath failed in LoadDict
Hit Return

EDIT: it creates a .lock file (which it should), but it doesn't delete it again.

Besides that, it updated perfectly here and it's crunching

**Brian the Fist** · 06-17-2003, 06:18 PM

My bad, I wrote the instructions in a hurry and forgot to metion:

Due to circumstances beyond our control, you must run '.\foldtrajlite -bench' on Windows.

(note the .\ before the command). I will add this to the readme for future.

**pfb** · 06-17-2003, 06:32 PM

ah - I remember that from when you added it during the BETA testing...

anyway, my P4 2.7, Windows XP gets:

Code:

Summary
-------
          Usr time  Sys time
          --------  --------
Maketrj      9.813     0.797
Foldtraj    38.109    13.313

and so a P4 is slower than an AMD Barton

I know that they're not comparable but it's still 'interesting' to see

**pointwood** · 06-18-2003, 03:07 AM

P4 2Ghz WinXP, 512MB mem:

Code:

Summary
-------
          Usr time  Sys time
          --------  --------
Maketrj     14.047     2.109
Foldtraj    57.328    20.375

**Tateman** · 06-18-2003, 12:02 PM

Here are my machines

P4 2.4 533fsb 512mb ram WinXP:

Code:

          Usr time  Sys time
          --------  --------
Maketrj      9.250     1.609
Foldtraj    38.859    14.156

XP1800+ @1627 512mb ram WinXP:

Code:

          Usr time  Sys time
          --------  --------
Maketrj      8.391     1.516
Foldtraj    41.188     7.188

p4 2.4 533fsb 384mb ram WinXP:

Code:

          Usr time  Sys time
          --------  --------
Maketrj     10.344     1.828
Foldtraj    51.844    19.953

P4 2.26 533fsb 256mb ram WinXP:

Code:

          Usr time  Sys time
          --------  --------
Maketrj     11.719     0.672
Foldtraj    48.641    18.391

P3 Dual 600 1gb ram WinXP:

Code:

          Usr time  Sys time
          --------  --------
Maketrj     22.875     1.438
Foldtraj   140.344    15.609

**HaloJones** · 06-18-2003, 12:18 PM

Code:

One moment, opening rotamer library...
Predicting secondary structure and generating trajectory distribution...
Folding protein...
Benchmark complete.

Summary
-------
          Usr time  Sys time
          --------  --------
Maketrj      6.250     0.328
Foldtraj    30.250     5.438

AMD XP at 2.405GHz

**jonnyw** · 06-24-2003, 06:03 AM

P4 2.53 Ghz - 512MB Ram

nothing running (i.e, no virus protection, firewall etc)

Code:

Summary
-------
               Usr time  Sys time
                 --------  --------
Maketrj      8.938     0.547
Foldtraj    36.906    12.891

evertything running (virus protection, firewall, internet connection manager, notepad, winamp, 3 IE windows)

Code:

Summary
-------
                Usr time  Sys time
                   --------  --------
Maketrj     10.641     0.734
Foldtraj     42.813    15.500

Which one is the most important, if its the 1st it seems to quite good, if its the second then obviously not as good?

**erk** · 06-24-2003, 09:17 PM

Mac G4 1GHz OSX 10.2.6 with browser,terminal, and a couple of small things open:

Usr time Sys time
-------- --------
Maketrj 11.500 0.000
Foldtraj 118.140 0.000

Celeron 2.0Ghz FreeBSD 4.8-RELEASE

Usr time Sys time
-------- --------
Maketrj 11.008 0.422
Foldtraj 58.062 2.117

Celeron 900Mhz FreeBSD 4.8-RELEASE

Usr time Sys time
-------- --------
Maketrj 15.984 0.602
Foldtraj 142.156 6.164

**markhl** · 06-24-2003, 11:53 PM

Usr time Sys time
Maketrj 19.879 3.245
Foldtrj 134.323 28.831

Intel Pentium III, WinXP SP1, 866 MHz, 256 MB RAM

**dtsang** · 06-30-2003, 01:01 PM

Usr time Sys time
-------- --------
Maketrj 15.390 0.000
Foldtraj 191.130 0.000

Power Mac G4 466 MHz, 384 MB RAM, Mac OS X 10.1.5

**rsbriggs** · 06-30-2003, 01:30 PM

Dell Dimension - P4 @ 2.66 Ghz 1Gig Ram 533 Mhz FSB

usr sys
8.563 0.531
34.859 13.00

Dell Poweredge 600 P4 @ 2.4 Ghz 768 Mb Ram 400 Mhz FSB

usr sys
9.250 0.672
37.750 13.250

----------------------------------------------------------------------------------
"Average" time to complete a set of 50 structures varies from a "couple of minutes" to more than an hour.

For all of that - I'm still encountering individual structures within a generation that get stuck for between 8 and 10 hours while the laxness levels slowly creep up to 43%/59%/100%. Are we completely certain that there isn't something wrong there? That's an ENORMOUS number of CPU cycles to spend in one spot of one structure....

**Grumpy** · 07-15-2003, 07:05 PM

Erm, has anyone got this working on Win98..I have 3 Win98 computers, and while it runs, the results are always 0.00 0.00 0.00 0.00

**Brian the Fist** · 07-16-2003, 10:19 AM

The Benchmark has not been tested on Win98 but SHOULD work... Anyone else get it to work? Is this Win98SE?

**Grumpy** · 07-16-2003, 11:35 AM

Yes, Win98 SE..seems to run ok, but says zeros for the four values. Three different computers, all different MBs and Cpus and the like..my 2 Win 2000 runs it ok, as does my Win XP Pro. Have run it from CLI and DFGUI 3.1, no difference in result.

**Brian the Fist** · 07-16-2003, 11:38 AM

Yup, I checked and you're right. Win98/ME does not support GetProcessTimes which is used to time the benchmark accurately. All the more reason to upgrade to Linux!

**spldart** · 07-16-2003, 12:40 PM

This is a bench on my dually. The bench was run on my second client as my first client was running normally in order to simulate two cpu's hard at work at the same time.
MSI master L, dual OC 1700+ at 2400+, on liquid, 512 meg samsung pc2700, FSB at 140, Win XP pro

Code:

Summary
-------
          Usr time  Sys time
          --------  --------
Maketrj      9.172     1.578
Foldtraj    51.484    11.531

**spldart** · 07-16-2003, 12:48 PM

Oh...
And I wasn't able to run a bench on that new test foldtraj (got an error in the command prompt box)... I had to put the old one back in for a minute to run the bench then switch back out.

**Grumpy** · 07-16-2003, 12:52 PM

What do we do with cheeky Project Leaders

And yest, the Benchmark gives a Fatal Error message with the uodate

The results for my dual 2400 MP are interesting:

Single Client running

Code:

   
Summary
-------
          Usr time  Sys time
          --------  --------

Maketrj..  9.156     0.797 
Foldtraj.. 56.047    12.172

Dual Clients

1st Instance

Code:

Summary
-------
          Usr time  Sys time
          --------  --------
Maketrj      9.625     0.750
Foldtraj    65.219    18.078

2nd Instance

Code:

Summary
-------
          Usr time  Sys time
          --------  --------
Maketrj      9.605     0.710
Foldtraj    65.267    18.143

The new Client really seems to have a lot of overhead on Duallies. Linux is not quite this bad, but still ugly

I think your MSI owns my Tyan bad..what is your FSB & Ram Timings

Got the old Barton up to 12*185 2220.42 Mhz 2,2,2,6

Code:

Summary
-------
          Usr time  Sys time
          --------  --------
Maketrj      6.297     0.172
Foldtraj    33.016     6.609

Thread: PII Benchmarks

Thread Tools

Rate This Thread

Display

PII Benchmarks

A few more

Posting Permissions