New GTX 260

[Denithor]

--- Opening Log file [April 3 01:11:33 UTC]


# Windows GPU Console Edition #################################################
###############################################################################

Folding@Home Client Version 6.23

http://folding.stanford.edu

###############################################################################
###############################################################################

Launch directory: C:\Users\Owner\Documents\Folding@home-Win32-GPU_Vista-623[2]
Executable: C:\Users\Owner\Documents\Folding@home-Win32-GPU_Vista-623[2]\Folding@home-Win32-GPU2.exe
Arguments: -gpu 1 -forcegpu nvidia_g80

[01:11:37] - Ask before connecting: No
[01:11:37] - User name: Denithor (Team 59)
[01:11:38] - User ID: 5A3305D3166B97D9
[01:11:38] - Machine ID: 3
[01:11:39]
[01:11:39] Loaded queue successfully.
[01:11:39] - Preparing to get new work unit...
[01:11:40] + Attempting to get work packet
[01:11:41] - Connecting to assignment server
[01:11:41] - Successful: assigned to (171.64.65.106).
[01:11:42] + News From Folding@Home: GPU folding beta
[01:11:42] Loaded queue successfully.
[01:11:44] + Closed connections
[01:11:45]
[01:11:45] + Processing work unit
[01:11:45] Core required: FahCore_11.exe
[01:11:46] Core found.
[01:11:46] Working on queue slot 07 [April 3 01:11:46 UTC]
[01:11:46] + Working ...
[01:11:47]
[01:11:47] *------------------------------*
[01:11:47] Folding@Home GPU Core - Beta
[01:11:48] Version 1.19 (Mon Nov 3 09:34:13 PST 2008)
[01:11:48]
[01:11:48] Compiler : Microsoft (R) 32-bit C/C++ Optimizing Compiler Version 14.00.50727.762 for 80x86
[01:11:48] Build host: amoeba
[01:11:49] Board Type: Nvidia
[01:11:49] Core :
[01:11:49] Preparing to commence simulation
[01:11:49] - Looking at optimizations...
[01:11:49] - Created dyn
[01:11:50] - Files status OK
[01:11:50] - Expanded 66260 -> 348500 (decompressed 525.9 percent)
[01:11:50] Called DecompressByteArray: compressed_data_size=66260 data_size=348500, decompressed_data_size=348500 diff=0
[01:11:50] - Digital signature verified
[01:11:50]
[01:11:51] Project: 5777 (Run 10, Clone 120, Gen 2)
[01:11:51]
[01:11:51] Assembly optimizations on if available.
[01:11:51] Entering M.D.
[01:11:59] Working on Protein
[01:12:02] Run: exception thrown during GuardedRun
[01:12:06] Run: exception thrown in GuardedRun -- Gromacs cannot continue further.
[01:12:15] Going to send back what have done -- stepsTotalG=0
[01:12:19] Work fraction=0.0000 steps=0.
[01:12:30] logfile size=0 infoLength=0 edr=0 trr=23
[01:12:30] - Writing 635 bytes of core data to disk...
[01:12:30] Done: 123 -> 124 (compressed to 100.8 percent)
[01:12:30] ... Done.
[01:12:31]
[01:12:31] Folding@home Core Shutdown: UNSTABLE_MACHINE
[01:12:33] CoreStatus = 7A (122)
[01:12:33] Sending work to server
[01:12:33] Project: 5777 (Run 10, Clone 120, Gen 2)
[01:12:33] - Read packet limit of 540015616... Set to 524286976.


[01:12:33] + Attempting to send results [April 3 01:12:33 UTC]

Folding@Home Client Shutdown.

Got in a new GTX 260 yesterday, installed today in my main rig. 9800GTX+ dropped into the second slot. Previously had 9800GTX+/8800GTS (G92) in this box, powered by Corsair 750TX.

As soon as I try to start the GPU2 client (running on 9800GTX+) it gives the error above.

I've tried running this under both XP Pro SP3 and Vista Ultimate 64-bit. No difference at all - as soon as the client starts it immediately fails.

This card has run flawlessly for the last two months and now suddenly it craps out completely? Something's amiss - please help!!

BTW - running 185.20 on XP (no change) and have updated from 180.48 to 182.06 under Vista hoping for an improvement. No luck.

 

[Insidious]

When I run into a wall like this, I uninstall all my folding stuff, then I go to Nvidia.com and download the latest drivers, uninstall my present drivers and reboot. when I come up, I install the downloaded drivers and then follow this guide.

I have found that if I follow it like a cook book with fresh (latest) drivers, things just have a way of straightening out.

Good luck,
Sid

 

[Drsignguy]

Drop your clocks at or below default levels. Then start cranking up the shader clocks only. I run my GTX 260/216's @ 576/1512/999 and at this level, good PPD, runs fairly cool(er).

I got the same errors and this is what I had to do...It may run really well in games but with F@H, you just never know sometimes..:roll: Good luck sir!

 

[Markfw]

You can't run a 9800 GTX and a gtx 260 in the same box....


"GTX 260 yesterday, installed today in my main rig. 9800GTX+ dropped into the second slot"

Won't work.

 

[Denithor]

All three cards are at stock speeds (none are factory OC models). CPU/RAM at stock speed.

System is totally stable with 9800GTX+/8800GTS folding.
System is totally stable with GTX 260 folding.

But I cannot start the second client with the GTX 260 in the box, period. Here are the combinations I have tried:

GTX 260 1 Fine
9800GTX 2 Fail

GTX260 1 Fine
8800GTS 2 Fail

9800GTX 1 Fine
GTX 260 2 Fail

In all three cases GPU client 2 will not start correctly (runs fine on the GTX 260 or 9800GTX, whichever is in slot 1).

Now, for the last two months I've had this setup running perfectly:

9800GTX Fine
8800GTS Fine

I will try the uninstall drivers/cards/folding tomorrow.

In the meantime, any other suggestions?

 

[Drsignguy]

Originally posted by: Markfw900
You can't run a 9800 GTX and a gtx 260 in the same box....


"GTX 260 yesterday, installed today in my main rig. 9800GTX+ dropped into the second slot"

Won't work.

Ah, good catch there Mark. Thats right, Must be of the same series of Gpu for this to work properly.

 

[Denithor]

Originally posted by: Markfw900
You can't run a 9800 GTX and a gtx 260 in the same box....


"GTX 260 yesterday, installed today in my main rig. 9800GTX+ dropped into the second slot"

Won't work.

Why not? This blows...I wish I'd known before I dropped $165 on the 260...

Sigh.

Oh, well - grounds to buy a second 260?

 

[Drsignguy]

Originally posted by: Denithor


Oh, well - grounds to buy a second 260?

Yea, that's what I said...Now 7 cards later.....:Q

 

[Markfw]

Oh, and Sid says they must also have the same number of shaders....a 260 192 won't work with a 260 216 shader....

 

[Denithor]

Crap. I may have to set up a second machine for this GTX 260 instead, have all the parts (or can cannibalize from my HTPC) and really want to use the new card for gaming but didn't really intend to buy two of them right now.

Maybe if the M$ Live Cashback goes up at eBay again...

 

[Drsignguy]

Originally posted by: Markfw900
Oh, and Sid says they must also have the same number of shaders....a 260 192 won't work with a 260 216 shader....

No, it works... But, you should set both cards at the same shader clock speeds. I had a 55nm 216 with the 192 core in the same rig at 1 time but I had to set the clocks at the same speed.

 

[Philippart]

Originally posted by: Denithor

--- Opening Log file [April 3 01:11:33 UTC]


# Windows GPU Console Edition #################################################
###############################################################################

Folding@Home Client Version 6.23

http://folding.stanford.edu

###############################################################################
###############################################################################

Launch directory: C:\Users\Owner\Documents\Folding@home-Win32-GPU_Vista-623[2]
Executable: C:\Users\Owner\Documents\Folding@home-Win32-GPU_Vista-623[2]\Folding@home-Win32-GPU2.exe
Arguments: -gpu 1 -forcegpu nvidia_g80

[01:11:37] - Ask before connecting: No
[01:11:37] - User name: Denithor (Team 59)
[01:11:38] - User ID: 5A3305D3166B97D9
[01:11:38] - Machine ID: 3
[01:11:39]
[01:11:39] Loaded queue successfully.
[01:11:39] - Preparing to get new work unit...
[01:11:40] + Attempting to get work packet
[01:11:41] - Connecting to assignment server
[01:11:41] - Successful: assigned to (171.64.65.106).
[01:11:42] + News From Folding@Home: GPU folding beta
[01:11:42] Loaded queue successfully.
[01:11:44] + Closed connections
[01:11:45]
[01:11:45] + Processing work unit
[01:11:45] Core required: FahCore_11.exe
[01:11:46] Core found.
[01:11:46] Working on queue slot 07 [April 3 01:11:46 UTC]
[01:11:46] + Working ...
[01:11:47]
[01:11:47] *------------------------------*
[01:11:47] Folding@Home GPU Core - Beta
[01:11:48] Version 1.19 (Mon Nov 3 09:34:13 PST 2008)
[01:11:48]
[01:11:48] Compiler : Microsoft (R) 32-bit C/C++ Optimizing Compiler Version 14.00.50727.762 for 80x86
[01:11:48] Build host: amoeba
[01:11:49] Board Type: Nvidia
[01:11:49] Core :
[01:11:49] Preparing to commence simulation
[01:11:49] - Looking at optimizations...
[01:11:49] - Created dyn
[01:11:50] - Files status OK
[01:11:50] - Expanded 66260 -> 348500 (decompressed 525.9 percent)
[01:11:50] Called DecompressByteArray: compressed_data_size=66260 data_size=348500, decompressed_data_size=348500 diff=0
[01:11:50] - Digital signature verified
[01:11:50]
[01:11:51] Project: 5777 (Run 10, Clone 120, Gen 2)
[01:11:51]
[01:11:51] Assembly optimizations on if available.
[01:11:51] Entering M.D.
[01:11:59] Working on Protein
[01:12:02] Run: exception thrown during GuardedRun
[01:12:06] Run: exception thrown in GuardedRun -- Gromacs cannot continue further.
[01:12:15] Going to send back what have done -- stepsTotalG=0
[01:12:19] Work fraction=0.0000 steps=0.
[01:12:30] logfile size=0 infoLength=0 edr=0 trr=23
[01:12:30] - Writing 635 bytes of core data to disk...
[01:12:30] Done: 123 -> 124 (compressed to 100.8 percent)
[01:12:30] ... Done.
[01:12:31]
[01:12:31] Folding@home Core Shutdown: UNSTABLE_MACHINE
[01:12:33] CoreStatus = 7A (122)
[01:12:33] Sending work to server
[01:12:33] Project: 5777 (Run 10, Clone 120, Gen 2)
[01:12:33] - Read packet limit of 540015616... Set to 524286976.


[01:12:33] + Attempting to send results [April 3 01:12:33 UTC]

Folding@Home Client Shutdown.

could that be the reason?

 

[Denithor]

Originally posted by: Philippart

Originally posted by: Denithor
Arguments: -gpu 1 -forcegpu nvidia_g80

could that be the reason?

No, that's the first thought I had - took it out, restarted the client, instant error again.

Grrrr. :|

 

[Denithor]

Originally posted by: Drsignguy

Originally posted by: Markfw900
Oh, and Sid says they must also have the same number of shaders....a 260 192 won't work with a 260 216 shader....

No, it works... But, you should set both cards at the same shader clock speeds. I had a 55nm 216 with the 192 core in the same rig at 1 time but I had to set the clocks at the same speed.

I'm now considering stepping up my 9800GTX+ to one of the new GTX 275 cards - with 240 shaders. I just want to confirm before I do this that I will be able to run folding on the 260 & 275 in the same box. Comments?

 

[Insidious]

My comment would be to go to the folding @ home forums and read up on this.

You are less likely to be disappointed if you have good information going in. That particular forum (despite the distasteful atmosphere they foster) has a wealth of information from users who have done exactly what you are considering.

Not to mention there is a post or two in this thread that describe results some people have achieved. (not stellar when cards are mis-matched)

-Sid

 

[Denithor]

Originally posted by: Drsignguy

Originally posted by: Markfw900
Oh, and Sid says they must also have the same number of shaders....a 260 192 won't work with a 260 216 shader....

No, it works... But, you should set both cards at the same shader clock speeds. I had a 55nm 216 with the 192 core in the same rig at 1 time but I had to set the clocks at the same speed.

Can you confirm your 192 & 216 sp cards ran the expected ppd?

For a short time I ran a G92 8800GTS (128SP) and an 8800GS (96SP) together and saw both cards put out drastically lower ppd than either one running solo (like 30% for each, combined they were barely better than just running the GTS alone). The SP were clocked differently but the clients ran fine, just with reduced ppd.

I currently run a 9800GTX+ and the same 8800GTS in a box, both have the same 128SP but again they are clocked differently. The client runs fine and I get full ppd (actually at the top end of expectation for both cards).

I really want to know if mixing a GTX 275 (240SP) with a GTX 260 (216SP) will run at full ppd or result in the same drastic reduction that I saw with the mis-matched G92 cards.

This is all because I have the option to stepup my 9800GTX+ to a GTX 275 and I'm wondering if it would be worth doing. If it plays nice with my 260 I'll do it, if not I'll probably just hold off.

Originally posted by: Insidious
My comment would be to go to the folding @ home forums and read up on this.

You are less likely to be disappointed if you have good information going in. That particular forum (despite the distasteful atmosphere they foster) has a wealth of information from users who have done exactly what you are considering.

Not to mention there is a post or two in this thread that describe results some people have achieved. (not stellar when cards are mis-matched)

-Sid

I looked around over there and they only have a single post dealing with the GTX 275 so far. And nothing at all concerning folding with it combined with other cards. Grrr...

 

[Insidious]

Bummer on the no info.

That's an expensive experiment too. Maybe when it's been out a little longer. ??

I really would be cautious of mixing cards with differing numbers of shaders though. I know there is posts about that.

My #shaders match, but the frequency doesn't with matching cores (G92). But they change so much with new cards that I don't even know if that is relevant info.

sorry I can't help more

-Sid

 

[Gravity]

I've run the 260 216 and the 260 stock in the same boxes. I've mixed and matched with good results. I've run 9800's with 8600s consistently. Put your fastest card in the first slot and consider the 2nd slot production gravy, reglardless of what it is. It's best to match but like I said, I've mixed it up pretty good with decent results. The driver you are using was not authorized at one point and fucks up the science pretty good. It increases your PPD but stanford doesn't like it. Use whatever cuda driver is approved and maybe the results will be better.

YMMV,

Gravity

 

[Denithor]

Thanks for that input Gravity.

You're right, Insidious - a rather expensive experiment. I'd hate to buy a 275 and discover it doesn't play nice with my 260/216.

I've still got ~35 days on my 9800GTX+ for stepup so maybe I'll just wait a bit and see if anyone puts anything out saying they play nice (or not).

I could always step up the 9800 to a 275 and if they don't play nice - step up the 260 also...