Let's go Team !!

[sswilson]

Originally Posted by _dangtx_

you can always super stress it and if its cool just do it via bios? :)
this way you wont have to deal with sliders and programs screwing up.
keep a copy of the original just in case. oh yeah and put the fan on dynamic.
this way if driver decides to take a brain fart your card wont die.

I'll see if it happens again. I much prefer to avoid flashing the bios unless I absolutely have to. I like to run my folding OCs right on the edge of 100% stable, so it's nice to be able to drop it down a notch when/if a new WU pushes the OC beyond stable.

As far as Dynamic goes.... My preference would be to peg it at 100% for safety's sake.

edit: Looks like an unstable OC... I completed that WU, but got NANs on the next one. Dialed the OC down by 10 core clocks.

[boggtrotter59]

I don't think flashing is a good idea either... I periodically have to lower the "modest" overclocks on my fermi cards so the cards don't get EUE errors... seems to happen more often when it gets milder outside and my room gets hotter. My non-fermi cards running gpu2 seem to be immune to this...Don't remember ever having to lower shader clocks to get the card stable again...

[sswilson]

I think GPU3 Fermi WUs work the non-core architecture harder than previous clients. With this in mind, my theory is that those components are more sensitive because fan speeds only ramp up based on Core temp.

Currently running 1:01 / 1:02 on a 6805 and reporting 17.8K - 18K+.

[LCB001]

Man those are some sweet numbers, how do you like the Frozr II cooler compared to the Cyclone...

[sswilson]

I like the look better, and It seems to be quieter, but that's hard to tell as the frozr is inside a case.

edit: Dropped the OC again overnight. I guess I'll have to do some research on the issue.... could be described either as a bug, or a feature. :) What's wierd is that the FAH client is reacting... It gives me a Corestatus = 63 error, but then starts up on it's own at the lower clocks. In the past, if an OC failed it would just die or bluescreen. This seems to be defaulting back to stock (bios settings) and carrying on.

[04:56:50] Completed  13000000 out of 50000000 steps (26%).
[04:57:52] Completed  13500000 out of 50000000 steps (27%).
[04:58:54] Completed  14000000 out of 50000000 steps (28%).
[04:59:55] Completed  14500000 out of 50000000 steps (29%).
[05:01:07] CoreStatus = 63 (99)
[05:01:07] + Error starting Folding@home core.
[05:01:12] 
[05:01:12] + Processing work unit
[05:01:12] Core required: FahCore_15.exe
[05:01:12] Core found.
[05:01:12] Working on queue slot 09 [February 13 05:01:12 UTC]
[05:01:12] + Working ...
[05:01:12] 
[05:01:12] *------------------------------*
[05:01:12] Folding@Home GPU Core
[05:01:12] Version 2.15 (Tue Nov 16 09:05:18 PST 2010)
[05:01:12] 
[05:01:12] Build host: SimbiosNvdWin7
[05:01:12] Board Type: NVIDIA/CUDA
[05:01:12] Core      : x=15
[05:01:12]  Window's signal control handler registered.
[05:01:12] Preparing to commence simulation
[05:01:12] - Ensuring status. Please wait.
[05:01:22] - Looking at optimizations...
[05:01:22] - Working with standard loops on this execution.
[05:01:22] - Previous termination of core was improper.
[05:01:22] - Files status OK
[05:01:22] sizeof(CORE_PACKET_HDR) = 512 file=<>
[05:01:22] - Expanded 40499 -> 162639 (decompressed 401.5 percent)
[05:01:22] Called DecompressByteArray: compressed_data_size=40499 data_size=162639, decompressed_data_size=162639 diff=0
[05:01:22] - Digital signature verified
[05:01:22] 
[05:01:22] Project: 6805 (Run 1767, Clone 2, Gen 0)
[05:01:22] 
[05:01:22] Entering M.D.
[05:01:24] Will resume from checkpoint file work/wudata_09.ckp
[05:01:24] Tpr hash work/wudata_09.tpr:  1665624460 1727712844 3563030993 133258715 1075771764
[05:01:24] Working on ALZHEIMER'S DISEASE AMYLOID
[05:01:24] Client config found, loading data.
[05:01:24] Resuming from checkpoint
[05:01:24] fcCheckPointResume: retreived and current tpr file hash:
[05:01:24]    0   1665624460   1665624460
[05:01:24]    1   1727712844   1727712844
[05:01:24]    2   3563030993   3563030993
[05:01:24]    3    133258715    133258715
[05:01:24]    4   1075771764   1075771764
[05:01:24] fcCheckPointResume: file hashes same.
[05:01:24] fcCheckPointResume: state restored.
[05:01:24] fcCheckPointResume: name work/wudata_09.log Verified work/wudata_09.log
[05:01:24] fcCheckPointResume: name work/wudata_09.trr Verified work/wudata_09.trr
[05:01:24] fcCheckPointResume: name work/wudata_09.xtc Verified work/wudata_09.xtc
[05:01:24] fcCheckPointResume: name work/wudata_09.edr Verified work/wudata_09.edr
[05:01:24] fcCheckPointResume: state restored 2
[05:01:24] Resumed from checkpoint
[05:01:24] Setting checkpoint frequency: 500000
[05:01:24] Completed  14500001 out of 50000000 steps (29%).
[05:01:24] Starting GUI Server
[05:02:46] Completed  15000000 out of 50000000 steps (30%).
[05:03:57] Completed  15500000 out of 50000000 steps (31%).
[05:05:08] Completed  16000000 out of 50000000 steps (32%).
[05:06:19] Completed  16500000 out of 50000000 steps (33%).
[05:07:31] Completed  17000000 out of 50000000 steps (34%).

[chrisk]

Corestatus 63= (99) code means something is up with the core:
CoreStatus codes - FaHWiki

According to FAHwiki, it happens when you change from one version of a core to another. Strange.

I would try deleting the work, core, and queue files, and add the "-forcegpu nvidia_fermi " flag as well.

[_dangtx_]

guys, you know how i usually mention everest for checking temps and voltages right?
well now im mentioning aida64. was using aida32 before it got renamed to everest, now back to the roots baby!

whats nice is the 3 digits for voltages, so comparing both everest and aida64 at the same time you get something like

vcore 1.20 vs 1.193

which can and will help you with tweaking!

[lowfat]

I use AIDA64 to display temps, usages, etc on my Logitech G13. Only issue I have with it is that it locks up occasionally and I need to restart it.

[_dangtx_]

nice. does it lock up with a ton of cpu usage or not?
32 or 64 bit os?
i usually leave mine on, but in the long run it can eat away some cpu cycles.

[lowfat]

Well my CPU is almost always @ 100% due to it folding. But even when I drop down to -smp 7 and CPU usage is around 85% it still locks up. 64-bit Windows. I just installed the newest beta this morning, we will see if it resolves the lock ups or not.