The new Hexus Teams - Setting up Folding at Home

[med2003]

Just joined.

[burtie]

Joined as burtie

[joshwa]

Updated to include instructions for Dual Core / Quad Core SMP setup.

[Dorza]

Originally Posted by joshwa

Updated to include instructions for Dual Core / Quad Core SMP setup.

Just to clear something up, do you just run fah.exe and let it run in the command window each time you want to fold and by doing so it will automatically run on both or 4 cores?

[joshwa]

Originally Posted by Dorza

Just to clear something up, do you just run fah.exe and let it run in the command window each time you want to fold and by doing so it will automatically run on both or 4 cores?

Yes, that's what I do

[Dorza]

OK cool, thanks.

[schmunk]

Right, I'm in, on my work laptop (C2D). I tried installing the SMP core, but it wouldn't work, so I'm just using the graphical standard core.

[directhex]

funny thing is, our users use GROMACS (which f@h uses for its cpu core these days) for real protein folding work

[SiM]

For those who can't run smp client get this file:
http://www.stanford.edu/group/pandeg...04-Console.exe

Make 2/4 copies of the file (I renamed to core 1.exe etc)
and then make shortcuts, right click add -local to the Target
"C:\Program Files\Folding@Home\Core1\Core 1.exe" -local
"C:\Program Files\Folding@Home\Core 2\Core 2.exe" -local

Make sure you put the Machine ID as 1 for "Core 1.exe" and 2 for "Core 2.exe" in the options

(Stolen and summarised from here)

Edit: link blocked out... if you need help just reply

[Webby]

Just joined will leave it running over night and see how my x1950 xt handles it!

[directhex]

note: the "SMP" client runs totally different types of WU to the regular client - it's not the same thing if you run multiple single-core clients

second note: as per the FAQ, performance of the "SMP" client is significantly worse on windows than anything else

third note: i'm saying "SMP" instead of SMP because it's not actually SMP - it's a cluster code being run on one machine. most problems people encounter are due to issues with the cluster communication library used by f@h for local execution

fourth note: the standard "a1" "SMP" core is locked to 4-way. if you're on dual-core, then resources are shared, and performance suffers. if you're on an 8 or more core box, you'll see no benefits in f@h performance

[Dorza]

So is it just better to just run the regular client multiple times? I've had issues with the SMP, it just crashes half way through a WU or just appears to have communication problems when fetching or sending data.

[directhex]

Originally Posted by Dorza

So is it just better to just run the regular client multiple times? I've had issues with the SMP, it just crashes half way through a WU or just appears to have communication problems when fetching or sending data.

PPD is higher for SMP, so it depends on whether you want to get the extra points for running SMP-only WUs

[Webby]

OKay so its been running for 2 days now and i have completed 2 work units.

One question which i hope somebody will be able to answer more graphics cardy than f@h, according to catalyst control centre (CCC) my GPU and memory clocks are 500 and 594 MHz (should be 625 and 900 MHz) now I understand this is to save energy when the card is not doing anything strenuous and when you load full screen 3D it jumps up to the higher values. Now I cant help thinking that by increasing my clocks to the 3D values I would get more work done! In the ATI overdrive section of CCC the is a tick box (ticked) which says enable ati overdrive settings clocks for 3D apps only, however this is greyed out so i cant untick it. Anybody know how to make the card run at the higher clocks all the time? and if it will make a difference to my folding productivity?

Thanks in advance

[directhex]

Originally Posted by directhex

PPD is higher for SMP, so it depends on whether you want to get the extra points for running SMP-only WUs

big SMP WUs are worth a *LOT*

10.25, 12am 3,283 2

[Zak33]

Originally Posted by Webby

OKay so its been running for 2 days now and i have completed 2 work units.

One question which i hope somebody will be able to answer more graphics cardy than f@h, according to catalyst control centre (CCC) my GPU and memory clocks are 500 and 594 MHz (should be 625 and 900 MHz) now I understand this is to save energy when the card is not doing anything strenuous and when you load full screen 3D it jumps up to the higher values. Now I cant help thinking that by increasing my clocks to the 3D values I would get more work done! In the ATI overdrive section of CCC the is a tick box (ticked) which says enable ati overdrive settings clocks for 3D apps only, however this is greyed out so i cant untick it. Anybody know how to make the card run at the higher clocks all the time? and if it will make a difference to my folding productivity?

Thanks in advance

first...thanks for joining

And second....I have NO IDEA!!!! Good thinking but I think it;s best not to fry your card.