TPU's F@H Team

[Steevo]

I had issues for awhile yesterday. They might be shuffling WU's around to different servers based on what server-WU you might get assigned.


I have been down some at work as were are starting the shuffling of systems around, and from what I hear now, our other store is going to move, and expand, and we are still on track to open at least one other new store.



So for now I wait. Mebey I can sell my current system to them for a grand and upgrade...........

 

[newtekie1]

Can or cannot?

It should have been cannot. It seems to be working fine now though.

 

[Polaris573]

Uninstalled the client and it still won't grab a new WU. Must be something on their end.

 

[Jess Stingray]

If you can see which server it tries to connect to, check it's status on here:

http://fah-web.stanford.edu/serverstat.html

 

[imperialreign]

back up for now - I've currently got both my 1950's cranking independently. I might throw the CPU into the mix, too - but I've got too address some cooling issues first.

to install the second GPU, I had to remove the fan on the SSB, which has left me with a passive heatsink and a nice pocket of warm air between the two vid cards, and under the bottom most one. mobo temps are getting a little too warm for my taste and it needs to be addressed.

 

[imperialreign]

odd problem here . . .


Now that I'm running to GPU consoles seperately, one for each card, both were folding at the same pace as the other - each taking ~10min to complete 1 step.

But, on occassion, one will be running ~10min/step, but the other is only folding at about twice that speed ~20min/step. Both are using the same core (10), so I don't really think there should be too much of a difference in pace . . . Although, now that I've closed them back out and re-opened them, both are running at the same pace again. They don't change pace too much once the console is actually at work, so it must be something when the core initializes, I would think. I leave xFire disabled when folding.

Could there be a potential hardware problem, or could it just be a difference in the actual WU it's trying to complete? I'm not really sure of what is causing the difference every now and then.

 

[hat]

It's probably just a difference in the WU's.

I reformatted back to my XP Home x86, and I feel that XP Pro x64 was crap. x64 just isn't ready yet I don't think.

Anyone remember me complaining about how SMP was taking 100% of my CPU regaurdless of what I set it at? Now, I have it set at 95%, and it's only using 95%, give or take 1%.

 

[imperialreign]

y'know, come to think of it . . . I think it's the difference in the PCIE bandwidth that's used. When not in xFire mode, the primary slot operates at x16, and the secondary at x8 - which is why is only seems that "Console 2" (as I've renamed it) is the only one that runs slow when both GPUs are folding. If I close out Console 1, #2 will pick up the pace.

I found a little workaround, anyhow . . . I leave Console 1 at low priority at 100% load, but I manually change the priority of Console 2 to 'high' or 'realtime' from within Task Manager - that seems to put both of them back at the same pace.

 

[hat]

SMP is being gay now.

It works great until I shut it down. If I do, when I start it again, I get a FILE_IO_ERROR, and SMP seems to forget everything it just did, meaning I lose all progress on my WU. Obviously I can't get anything done this way.

SMP is the only version of F@H I have running. I have 2 hard drives, and if it matters, I'm running it off of the non-system drive. The two drives are also different in interface: the system drive is SATA, the spare is ATA133. The ATA drive is on the same channel as my DVD burner (dvd burner is set to master), and I only have one IDE port.

What bothers me is before I went to x64 and back to x86 it worked perfectly, and I'm not doing anything different, aside from different settings in the BIOS (overclocking settings).

help!

 

[hat]

Hm, Version 5.04 text only console doesn't fuck up like SMP does. sometingwong!

 

[hat]

Hm, SMP gives me this error:
Bad work unit. Digital signatures don't match.


edit: hmm, I think I may have solved my own problem. My Crypotgraphic Services was disabled. Enabled it, will see how things go from here.

 

[hat]

sigh, no go there either. Cryptographic Services set to Auto, still giving me a bad signature error.

 

[Athlon2K15]

are you having fun talknig to yourself? lol when's the gpu client come out for the 2k's

 

[hat]

trying to get help here but nobody's listening :/
I don't get this error in the 6.xx beta client (single core).

 

[newtekie1]

Lower you overclock as see if it fixes the problem.

The SMP client is a lot more sensitive to instability. My Pentium D 805 would give the same problem of restarting at the beginning every time I stopped it and restarted it. I eventually fixed it by bumping the core voltage a notch in the BIOS.

 

[hat]

Well, I'm back at stock with a fresh install of SMP. If it doesn't work its probably a disabled service (which doesn't make sense if the other clients work).

 

[hat]

Lower you overclock as see if it fixes the problem.

The SMP client is a lot more sensitive to instability. My Pentium D 805 would give the same problem of restarting at the beginning every time I stopped it and restarted it. I eventually fixed it by bumping the core voltage a notch in the BIOS.

I'm back at my old OC settings. 250x11 and RAM at 500MHz (250MHz real). SMP seems to be working fine. Who would have thought? I thought the other OC settings were more stable.

 

[hat]

Hm, it happens when I rebooted without shutting F@H down first and having F@H in the startup folder. Eh..

Yup. Not shutting F@H down at reboot is what boogers it up, at least as far as I can tell.

 

[imperialreign]

Alright, I think I've gotten everything sorted out, now.

I lost a couple of WUs trying to figure out how best to run the GPUs. It's a shame, too - it only happened having the priority set to 'realtime' instead of 'high'. So, I've learned me lesson there.

One GPU steps at ~11min, the other at ~9min. I can live with that. I'm completely forgoing the CPU console, as my P4 (even OC) just turns too slow. I'd rather just leave the CPU console out of it so the CPU itself can attend to the 2 GPUs, which only put the sys at about 70% load.

TPU is now up to position 344

 

[[I.R.A]_FBi]

hd38XX client out yet?

 

[hat]

Alright, I think I've gotten everything sorted out, now.

I lost a couple of WUs trying to figure out how best to run the GPUs. It's a shame, too - it only happened having the priority set to 'realtime' instead of 'high'. So, I've learned me lesson there.

One GPU steps at ~11min, the other at ~9min. I can live with that. I'm completely forgoing the CPU console, as my P4 (even OC) just turns too slow. I'd rather just leave the CPU console out of it so the CPU itself can attend to the 2 GPUs, which only put the sys at about 70% load.

TPU is now up to position 344

just use the Xfire console and be done with it

 

[newtekie1]

Alright, I think I've gotten everything sorted out, now.

I lost a couple of WUs trying to figure out how best to run the GPUs. It's a shame, too - it only happened having the priority set to 'realtime' instead of 'high'. So, I've learned me lesson there.

One GPU steps at ~11min, the other at ~9min. I can live with that. I'm completely forgoing the CPU console, as my P4 (even OC) just turns too slow. I'd rather just leave the CPU console out of it so the CPU itself can attend to the 2 GPUs, which only put the sys at about 70% load.

TPU is now up to position 344

Different work units will give different times between steps. Yesterday my x1950Pro was taking ~20min per step. It switched work units overnight and now it is taking ~16min per step.

 

[imperialreign]

just use the Xfire console and be done with it

didn't think there was one

 

[newtekie1]

The 6.00 Beta 1 GPU Console supports crossfire.

 

[[I.R.A]_FBi]

hd support yet?