Why am I getting Tinker cores? Unfair :)

A forum just for SPCR's folding team... by request.

Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee

Post Reply
ColdFlame
Posts: 451
Joined: Wed May 21, 2003 9:39 pm
Location: Somewhere in Time

Why am I getting Tinker cores? Unfair :)

Post by ColdFlame » Fri Nov 14, 2003 9:28 pm

One of my machines is getting Tinker cores which are slow a$$es. Two others are getting exclusively Gromacs cores and as result my p3 1 GHz beats my Celeron 2.4 GHz :(

Any idea about what heuristics they use to distribute the load and types of cores? Or I just have to wait and eventually I will get the Gromacs? Maybe I can force Tinker to go to my slow p3 and Gromacs to Celeron 2.4?

I know I'm a newbie but I want to break into yellow (so far I'm only going to be green) so I'm anxious.

Thanks!

haysdb
Patron of SPCR
Posts: 2425
Joined: Fri Aug 29, 2003 11:09 pm
Location: Earth

Post by haysdb » Fri Nov 14, 2003 9:31 pm

Nobody wants the Tinkers, so I think they give all the "new guys" a couple. :)

David

ColdFlame
Posts: 451
Joined: Wed May 21, 2003 9:39 pm
Location: Somewhere in Time

Post by ColdFlame » Sat Nov 15, 2003 11:18 am

I managed to trick F@H into giving me all 3 gromacs cores now :twisted: I'll post my findings wrt performance later when I finish my experiments.

Preliminary results show that Tinker cores are up to 45% slower wrt to points than Gromacs cores on the same machine.

ColdFlame
Posts: 451
Joined: Wed May 21, 2003 9:39 pm
Location: Somewhere in Time

Post by ColdFlame » Sat Nov 15, 2003 6:11 pm

I'm leaving town tomorrow so I won't be able to continue my experiments this week but here are my findings so far:
1) On Tinker cores AMD seems to beat my Celeron because neither one is using SSE/3DNow!
2) On Gromacs cores Celeron beats AMD, slightly.

ColdFlame
Posts: 451
Joined: Wed May 21, 2003 9:39 pm
Location: Somewhere in Time

Post by ColdFlame » Sat Nov 15, 2003 9:47 pm

Here are the numbers:
1) If I get a Gromacs core on a Celeron 2.4 I produce about 48 PPD
2) If I get a Tinker core on a Celeron 2.4 I produce about 38 PPD

Now, I'm not actually very sure about the Gromacs number now, it might actually be as high as 58, but let's say it is 48.

The point is that Stanford needs to give more credits for Tinkers IMHO.

Zyzzyx
Friend of SPCR
Posts: 1063
Joined: Mon Dec 23, 2002 12:55 pm
Location: Richland, WA
Contact:

Post by Zyzzyx » Sat Nov 15, 2003 9:56 pm

Nah.

Ya just gotta believe that everyone eventually gets the same distribution of Tinkers/Gromac ratio.

ColdFlame
Posts: 451
Joined: Wed May 21, 2003 9:39 pm
Location: Somewhere in Time

Post by ColdFlame » Sat Nov 15, 2003 11:40 pm

My faith is based on the fact that you must be right :)

Still it is extremely suspicious that out of 3 machines that I have, 2 get exclusively Gromacs and 1 gets exclusively Tinkers. I do admit that I've only folded 10 WUs so my statistics isn't complete.

Post Reply