A growing problem with GT/GTX 6xx series cards.


log in

Advanced search

Message boards : Number crunching : A growing problem with GT/GTX 6xx series cards.

Previous · 1 · 2 · 3 · Next
Author Message
Profile Wiggo
Avatar
Send message
Joined: 24 Jan 00
Posts: 7965
Credit: 98,332,301
RAC: 25,387
Australia
Message 1372842 - Posted: 28 May 2013, 22:01:16 UTC - in response to Message 1372841.
Last modified: 28 May 2013, 22:01:29 UTC

seems to be fixed :D

It does look like it is.

Cheers.

Profile SliverProject donor
Avatar
Send message
Joined: 18 May 11
Posts: 281
Credit: 7,192,891
RAC: 111
United States
Message 1372844 - Posted: 28 May 2013, 22:09:46 UTC

Nothing appears to be wrong with mine. It's unfortunate I didn't buy two cards when I built this computer. Little did I know they my GTX 460 would be discontinued before I could get the fundage. That's life I guess.
____________

Profile Wiggo
Avatar
Send message
Joined: 24 Jan 00
Posts: 7965
Credit: 98,332,301
RAC: 25,387
Australia
Message 1372846 - Posted: 28 May 2013, 22:14:19 UTC - in response to Message 1372844.

Nothing appears to be wrong with mine. It's unfortunate I didn't buy two cards when I built this computer. Little did I know they my GTX 460 would be discontinued before I could get the fundage. That's life I guess.

That's because you have a GTX 460 and not a GT/GTX 6xx as the thread title states. ;-)

Cheers.

Profile SliverProject donor
Avatar
Send message
Joined: 18 May 11
Posts: 281
Credit: 7,192,891
RAC: 111
United States
Message 1372848 - Posted: 28 May 2013, 22:15:26 UTC - in response to Message 1372846.

That's because you have a GTX 460 and not a GT/GTX 6xx as the thread title states. ;-)


Oh boy. I need to get my eyes checked. Sorry guys :/

____________

Josef W. SegurProject donor
Volunteer developer
Volunteer tester
Send message
Joined: 30 Oct 99
Posts: 4335
Credit: 1,113,795
RAC: 779
United States
Message 1372899 - Posted: 29 May 2013, 1:59:31 UTC - in response to Message 1372848.

That's because you have a GTX 460 and not a GT/GTX 6xx as the thread title states. ;-)


Oh boy. I need to get my eyes checked. Sorry guys :/

That's OK. Some of the 6xx cards are actually rebranded 4xx or 5xx Fermi cards, others are actually Kepler. Any confusion over what 6xx means can be blamed on NVIDIA.
Joe

spitfire_mk_2
Avatar
Send message
Joined: 14 Apr 00
Posts: 462
Credit: 13,116,725
RAC: 4,918
United States
Message 1373077 - Posted: 29 May 2013, 15:43:51 UTC - in response to Message 1372844.

Nothing appears to be wrong with mine. It's unfortunate I didn't buy two cards when I built this computer. Little did I know they my GTX 460 would be discontinued before I could get the fundage. That's life I guess.

I saw some good deals on ebay. One auction was for two gtx460 for 140 or 150 dollars, I think it was buy now price, so no dickering.
____________

Profile Wiggo
Avatar
Send message
Joined: 24 Jan 00
Posts: 7965
Credit: 98,332,301
RAC: 25,387
Australia
Message 1375037 - Posted: 1 Jun 2013, 23:19:22 UTC - in response to Message 1373077.

This problem has certainly been fixed with the release of SETI V7. :-D

Now if we can just get those people with out of control GTX 560Ti's to fix their rigs.

But another problem is starting to rear an ugly head and that is a lot of Radeons are now either just throwing errors or invalids. :-(

Cheers.

Tom*
Send message
Joined: 12 Aug 11
Posts: 114
Credit: 4,815,461
RAC: 82
United States
Message 1375100 - Posted: 2 Jun 2013, 3:13:31 UTC

Yes there does seem to be an extraordinary amount of ATI running SDK 2.5

so many in fact that several times Ive seen two sdk 2.5's validate against one

another.

That cannot be good science

Profile MikeProject donor
Volunteer tester
Avatar
Send message
Joined: 17 Feb 01
Posts: 24923
Credit: 34,409,003
RAC: 8,444
Germany
Message 1375163 - Posted: 2 Jun 2013, 7:16:14 UTC

I mentioned it already at beta.
I hope Eric can fix this.

____________

Grant (SSSF)
Send message
Joined: 19 Aug 99
Posts: 5918
Credit: 61,708,409
RAC: 19,121
Australia
Message 1375175 - Posted: 2 Jun 2013, 7:29:01 UTC - in response to Message 1375163.


From the v7 release thread in the News forum-

I found this hostid=6343230.

This machine D/L 8 MB AP files just for to trash them immediately.
Wasted bandwidth.

This machine have also a too old driver installed?

Why send SAH then WUs to such hosts...
Maybe the admins could adjust something for to prevent this?


From Eric
The change to prevent this went in yesterday. Let me know if its still happening.

____________
Grant
Darwin NT.

Profile Oz
Avatar
Send message
Joined: 6 Jun 99
Posts: 164
Credit: 113,843,221
RAC: 38,890
United States
Message 1375292 - Posted: 2 Jun 2013, 11:45:34 UTC

I recently purchased a gtx660 and I am sorry I did. Neither msi afterburner, evga precision or nvidia desktop offer any usable control of the card. I managed to get it running stably with x41zc 5.0, but with the rollout of sah v7 it now is erroring out wu's. When I try to use gpuz to check it [in a failure state] gpuz uses 50% proceesor [one core] and fails to load, a reboot is the only way to fix it. An attempt to load 301.42 driver yields a "no nvidia hardware found" [!?!] message. ANY suggestions leading to a resolution would be greatly appreciated.
[currently installed on host 4695891]
____________


Profile Wiggo
Avatar
Send message
Joined: 24 Jan 00
Posts: 7965
Credit: 98,332,301
RAC: 25,387
Australia
Message 1375307 - Posted: 2 Jun 2013, 12:10:54 UTC - in response to Message 1375292.

I recently purchased a gtx660 and I am sorry I did. Neither msi afterburner, evga precision or nvidia desktop offer any usable control of the card. I managed to get it running stably with x41zc 5.0, but with the rollout of sah v7 it now is erroring out wu's. When I try to use gpuz to check it [in a failure state] gpuz uses 50% proceesor [one core] and fails to load, a reboot is the only way to fix it. An attempt to load 301.42 driver yields a "no nvidia hardware found" [!?!] message. ANY suggestions leading to a resolution would be greatly appreciated.
[currently installed on host 4695891]

I run 2 of these on my old Q6600 using 314.07 driver plus I have good control over them with EVGA Precision X (though I only use it to run the fan faster).

Cheers.

Profile Oz
Avatar
Send message
Joined: 6 Jun 99
Posts: 164
Credit: 113,843,221
RAC: 38,890
United States
Message 1375313 - Posted: 2 Jun 2013, 12:28:24 UTC - in response to Message 1375307.

Thanks for the reply Wiggo. I have precision X, it lets me set the voltage, any "power" setting from 85 to 110% gives the same clockrate [1175] for a while, then 1050 then 300. Fan speed override only works with precision X open [running]. I guess I was spoiled by afterburner which applies specific voltage, clock and memory clock rates, then can be closed and returns used system memory and processor. The latest version doesn't have any effect on the card at all - clicking apply yields no change. Maybe I need to rma the card.
____________


ClaggyProject donor
Volunteer tester
Send message
Joined: 5 Jul 99
Posts: 4215
Credit: 34,478,815
RAC: 13,959
United Kingdom
Message 1375315 - Posted: 2 Jun 2013, 12:31:39 UTC - in response to Message 1375175.
Last modified: 2 Jun 2013, 12:38:14 UTC


From the v7 release thread in the News forum-

I found this hostid=6343230.

This machine D/L 8 MB AP files just for to trash them immediately.
Wasted bandwidth.

This machine have also a too old driver installed?

Why send SAH then WUs to such hosts...
Maybe the admins could adjust something for to prevent this?


From Eric
The change to prevent this went in yesterday. Let me know if its still happening.

The change as far as I know was to stop v7 work going to hosts with too old a APP runtime, this host has never had AMD v7 work because it is running Boinc 6.12.34, this Boinc doesn't do OpenCL detection so can't supply APP runtime version:

Application details for host 6343230

The AMD OpenCL AP problem is maybe a corrupt dll, exit code is: - exit code -1073741515 (0xc0000135), or maybe a total lack of AMD OpenCL support, drivers around that time had the OpenCL driver as an optional download,
the AMD OpenCL AP work it's receiving is using the ati_opencl_100 plan class, this uses the assumption that a GPU with driver x will also have OpenCL support, looking at the Cat 10.9 driver page there's no mention of an OpenCL driver being included,
Cat 11.1 is required before an OpenCL driver is mentioned as being included, since CAL version 1.4.900 is also shared with Cat 10.12, the minimum driver for this app version should be at least 11.2 (CAL 1.4.1016)

Edit: Posted Duplicate to the News thread, since that is where the post i'm replying to came from, and Eric is more likely to Read it there.

Claggy

Profile Wiggo
Avatar
Send message
Joined: 24 Jan 00
Posts: 7965
Credit: 98,332,301
RAC: 25,387
Australia
Message 1375330 - Posted: 2 Jun 2013, 12:58:03 UTC - in response to Message 1375313.

Thanks for the reply Wiggo. I have precision X, it lets me set the voltage, any "power" setting from 85 to 110% gives the same clockrate [1175] for a while, then 1050 then 300. Fan speed override only works with precision X open [running]. I guess I was spoiled by afterburner which applies specific voltage, clock and memory clock rates, then can be closed and returns used system memory and processor. The latest version doesn't have any effect on the card at all - clicking apply yields no change. Maybe I need to rma the card.

I'm just running mine at their default settings, 1100MHz GPU, 6010MHz memory with 1.175v, only running the fan speed locked at 70% instead of auto.

Also if you have Precision X ticked for windows start up it works in the back ground without the console.

Cheers.

Profile arkaynProject donor
Volunteer tester
Avatar
Send message
Joined: 14 May 99
Posts: 3728
Credit: 48,768,260
RAC: 1,737
United States
Message 1375552 - Posted: 2 Jun 2013, 19:45:48 UTC - in response to Message 1375292.

I recently purchased a gtx660 and I am sorry I did. Neither msi afterburner, evga precision or nvidia desktop offer any usable control of the card. I managed to get it running stably with x41zc 5.0, but with the rollout of sah v7 it now is erroring out wu's. When I try to use gpuz to check it [in a failure state] gpuz uses 50% proceesor [one core] and fails to load, a reboot is the only way to fix it. An attempt to load 301.42 driver yields a "no nvidia hardware found" [!?!] message. ANY suggestions leading to a resolution would be greatly appreciated.
[currently installed on host 4695891]


If I remember correctly, you will need a newer driver than that for the installer to recognize the GTX660.

I am running 320.00 on both of my machines and it works just fine.
____________

Terror Australis
Volunteer tester
Send message
Joined: 14 Feb 04
Posts: 1759
Credit: 206,464,641
RAC: 10,852
Australia
Message 1375565 - Posted: 2 Jun 2013, 20:04:21 UTC - in response to Message 1375330.
Last modified: 2 Jun 2013, 20:05:33 UTC

Thanks for the reply Wiggo. I have precision X, it lets me set the voltage, any "power" setting from 85 to 110% gives the same clockrate [1175] for a while, then 1050 then 300. Fan speed override only works with precision X open [running]. I guess I was spoiled by afterburner which applies specific voltage, clock and memory clock rates, then can be closed and returns used system memory and processor. The latest version doesn't have any effect on the card at all - clicking apply yields no change. Maybe I need to rma the card.

I'm just running mine at their default settings, 1100MHz GPU, 6010MHz memory with 1.175v, only running the fan speed locked at 70% instead of auto.

Also if you have Precision X ticked for windows start up it works in the back ground without the console.

Cheers.

I found Precision X had issues, including failure to control the 550Ti's properly, high CPU usage (5 to 10%), caused the whole computer to slow down and some other weird effects. I rolled back to an earlier version and the cards and the computer generally were a lot happier.

T.A.

Profile Oz
Avatar
Send message
Joined: 6 Jun 99
Posts: 164
Credit: 113,843,221
RAC: 38,890
United States
Message 1375723 - Posted: 3 Jun 2013, 3:20:14 UTC - in response to Message 1375565.

Thanx T.A., I'll try that.

@Wiggo, I'm experiencing what T.A. refers to - I have no control of the cards clock[s] or fan. It runs wherever it wants, I can control the fan, but only if precision is running in the foreground, using ~10% cpu and a wad of ram. As soon as it's closed, the fan reverts to auto - the core clock and memory clock cannot be adjusted at all by any program. I should have spent the extra $100 and bought a 560ti.
Sorry for ranting, but I get pretty p.o.ed when I spend a good chunk of change on something that doesn't work as it should or worse, tries to "help" by doing what it thinks you want instead of allowing user control.
____________


bill
Send message
Joined: 16 Jun 99
Posts: 861
Credit: 24,148,117
RAC: 1,872
United States
Message 1375731 - Posted: 3 Jun 2013, 3:40:02 UTC - in response to Message 1375723.
Last modified: 3 Jun 2013, 3:44:24 UTC

I'm confused.
You're posting about problems with a nvidia 550
but the machine you referenced to and everything
else I looked at said the gpu was a 660. If it really
is a 550 could the pc be confused sbout it's gpu
and be causing your problems?

Edit: never mind, I AM confused. I must of been thinking
of another thread. BTW, what version of Precision were
you using so I can keep an eye out for it?

tbretProject donor
Volunteer tester
Avatar
Send message
Joined: 28 May 99
Posts: 2897
Credit: 218,381,557
RAC: 26,949
United States
Message 1375745 - Posted: 3 Jun 2013, 4:23:42 UTC - in response to Message 1375292.
Last modified: 3 Jun 2013, 4:26:28 UTC

I recently purchased a gtx660 and I am sorry I did. Neither msi afterburner, evga precision or nvidia desktop offer any usable control of the card. I managed to get it running stably with x41zc 5.0, but with the rollout of sah v7 it now is erroring out wu's. When I try to use gpuz to check it [in a failure state] gpuz uses 50% proceesor [one core] and fails to load, a reboot is the only way to fix it. An attempt to load 301.42 driver yields a "no nvidia hardware found" [!?!] message. ANY suggestions leading to a resolution would be greatly appreciated.
[currently installed on host 4695891]


I admit that I do not have a 660. I have 660Tis and 670s.

Yes, you are correct, you cannot control the card the way you used to be able-to, but that shouldn't keep you from enjoying the card.

The *first* thing I would do is get rid of all of your programs that control the card. What I mean is uninstall all of them and wipe the directories out. Like you, I've used Afterburner, NVIDIA Inspector, and Precision, most of the time I've had more than one installed.

On more than one of my computers, but with different combinations of programs, drivers, and hardware I've had problems that I fixed by doing what I just suggested you do (uninstall it ALL) and then reload the drivers with a "clean install" and I even go so far as to name a new directory for the fresh driver install and wipe the old directories.

THEN install Precision X.

I was having your issue. I had these nice shiny new 670s and they were running fine then down-clocking. There was an early driver with a "monitor sleep problem" but if you run current drivers (I do, there's no problem with them) then that problem has been fixed. I didn't have *that* problem, yet my cards would be running fine and I'd check the next day and they had down-clocked.

I was really unhappy.

After spending hours and hours trying to find the culprit, I found some reference to Afterburner in a Windows error log. I uninstalled Afterburner (just shutting it down didn't work), reloaded the drivers with a clean install to a new directory (the idea being to make the registry reference a different subdirectory and avoid whatever had been written to the old one) and the down-clocking problem went away.

I liked and did not want to give-up my Afterburner. That didn't matter.

Now that the newest version of Precision X is out, I like it as well or better than I ever liked Afterburner. It does take some getting used-to.

You will never "dictate" to the Kepler like you did with the Fermi.

But here's where I run-out of any knowledge. The 660, not the Ti version, is a different chip, the GK106. I'm not sure that what little I know translates to the GK106. I hope if it doesn't translate that you aren't frustrated with me because I'm just trying to be helpful.

Jason posted a lot about dealing with a Kepler when the first 600s were released. Let me see if I can find something for you...

I'm so glad I remembered: Jason's Profile with Instructions

Just try it. Get rid of everything but the latest version of Precision X and look carefully at what Jason wrote.

I really hope it helps.

Previous · 1 · 2 · 3 · Next

Message boards : Number crunching : A growing problem with GT/GTX 6xx series cards.

Copyright © 2014 University of California