Advanced search

Message boards : Graphics cards (GPUs) : Unexpected error?

Author Message
Profile Logan Carr
Send message
Joined: 12 Aug 15
Posts: 207
Credit: 30,647,175
RAC: 32,853
Level
Val
Scientific publications
wat
Message 47234 - Posted: 16 May 2017 | 13:30:27 UTC

Hi all,

Here's an error I have received after running this work unit overnight.

http://www.gpugrid.net/result.php?resultid=16301823

I do not know what has caused this... Is this an issue with anyone else?
____________
Cruncher/Learner in progress.

Jim1348
Send message
Joined: 28 Jul 12
Posts: 460
Credit: 1,130,761,180
RAC: 18,722
Level
Met
Scientific publications
watwatwatwatwatwatwatwatwat
Message 47235 - Posted: 16 May 2017 | 17:00:12 UTC - in response to Message 47234.

It happens all the time if your card is overclocked.

JoergF
Avatar
Send message
Joined: 20 Apr 15
Posts: 206
Credit: 297,383,461
RAC: 1,110,115
Level
Asn
Scientific publications
watwat
Message 47236 - Posted: 16 May 2017 | 17:21:30 UTC

From the error message it seems that the card runs at pretty normal parameters... as this one is an EVGA ACX 2.0 OC and therefore overclocked ex factory. So I don't see anything special in this regard.
____________
I would love to see HCF1 protein folding and interaction simulations to help my little boy... someday.

Jim1348
Send message
Joined: 28 Jul 12
Posts: 460
Credit: 1,130,761,180
RAC: 18,722
Level
Met
Scientific publications
watwatwatwatwatwatwatwatwat
Message 47237 - Posted: 16 May 2017 | 17:28:41 UTC - in response to Message 47236.

Factory overclocked is overclocked. It may work for games, but not for the harder GPUGrid work units.

Profile Logan Carr
Send message
Joined: 12 Aug 15
Posts: 207
Credit: 30,647,175
RAC: 32,853
Level
Val
Scientific publications
wat
Message 47238 - Posted: 16 May 2017 | 18:02:04 UTC

Any way I can downclock for the project? And increase back for my games?
____________
Cruncher/Learner in progress.

Jim1348
Send message
Joined: 28 Jul 12
Posts: 460
Credit: 1,130,761,180
RAC: 18,722
Level
Met
Scientific publications
watwatwatwatwatwatwatwatwat
Message 47239 - Posted: 16 May 2017 | 19:09:32 UTC - in response to Message 47238.

I use Nvidia Inspector 1.9.7.3, though the 1.9.7.6 versions had problems.
Here is the latest, but I haven't tried it.
http://www.majorgeeks.com/files/details/nvidia_inspector.html

A fairly safe bet is MSIAfterburner, which works on both AMD and Nvidia and a lot of people use it.

I think either of them can set the clock permanently so that it survives a reboot, or you can remove the clock when you want to, but I have not used them for a while.

JoergF
Avatar
Send message
Joined: 20 Apr 15
Posts: 206
Credit: 297,383,461
RAC: 1,110,115
Level
Asn
Scientific publications
watwat
Message 47240 - Posted: 16 May 2017 | 19:38:15 UTC - in response to Message 47237.
Last modified: 16 May 2017 | 19:46:48 UTC

Factory overclocked is overclocked. It may work for games, but not for the harder GPUGrid work units.


I have never had any error on my factory overclocked Palit Super Jetstream gtx1070. The conclusion {OC => calculation error} does not sound logical to me anyway. Half of the GPUGRID crunchers would have permanent failures in that case.
____________
I would love to see HCF1 protein folding and interaction simulations to help my little boy... someday.

Profile Logan Carr
Send message
Joined: 12 Aug 15
Posts: 207
Credit: 30,647,175
RAC: 32,853
Level
Val
Scientific publications
wat
Message 47241 - Posted: 16 May 2017 | 20:06:13 UTC - in response to Message 47240.

Factory overclocked is overclocked. It may work for games, but not for the harder GPUGrid work units.


I have never had any error on my factory overclocked Palit Super Jetstream gtx1070. The conclusion {OC => calculation error} does not sound logical to me anyway. Half of the GPUGRID crunchers would have permanent failures in that case.



I think the issue that mine has is that it is "SSC" which I believe means superclocked more than just sc .

ok once i get home i will try nvidia inspector
____________
Cruncher/Learner in progress.

Jim1348
Send message
Joined: 28 Jul 12
Posts: 460
Credit: 1,130,761,180
RAC: 18,722
Level
Met
Scientific publications
watwatwatwatwatwatwatwatwat
Message 47243 - Posted: 16 May 2017 | 20:35:18 UTC - in response to Message 47240.

Half of the GPUGRID crunchers would have permanent failures in that case.

Almost.
http://www.gpugrid.net/forum_thread.php?id=3299#28672

Jacob Klein
Send message
Joined: 11 Oct 08
Posts: 1067
Credit: 1,146,403,839
RAC: 1,089,717
Level
Met
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 47246 - Posted: 16 May 2017 | 21:17:52 UTC
Last modified: 16 May 2017 | 21:22:52 UTC

Guys, factory overclocks aren't always stable to begin with. Then, on top of that, when new drivers come out, with optimizations that make portions of the GPU hardware work harder (to perform better) than prior drivers, that too can make an overclock become unstable.

I recommend using GPU-Z, and looking at the "Graphics Card" tab, and look at "GPU Clock", and compare against the URLs below, looking at "Base Core Clock (MHz)", to see how overclocked you are:
https://en.wikipedia.org/wiki/GeForce_10_series
https://en.wikipedia.org/wiki/GeForce_900_series
https://en.wikipedia.org/wiki/List_of_Nvidia_graphics_processing_units

I recommend MSI Afterburner for controlling the GPU Core Clock offset.

I recommend the following test to determine if your GPU clock is stable:
- Download the latest PrimeGrid Genefer OpenCL app
(One version is: http://www.primegrid.com/download/primegrid_genefer_3_3_1_3.13_windows_intelx86__OCLcudaGFNWR.exe)
- Open a Command Prompt window
- Run with the following arguments, including the double quotes:
primegrid_genefer_3_3_1_3.13_windows_intelx86__OCLcudaGFNWR.exe -q "43322502^131072+1" -d 0
- If the application completes successfully with a .log file that has "is composite. (RES=" in it, then you are short-term-stable.
- If you want to run on a different GPU, change the -d parameter
- Another good test is to swap out that number in double-quotes, with this number: "43370168^131072+1"
- You could create a .bat file, to run it overnight, to ensure better stability. Details here:
http://www.gpugrid.net/forum_thread.php?id=4310&nowrap=true#43617

As an example of how I've had to adjust my factory-overclocked GPUs for complete stability at maximum stable clock, my GPUs are:
eVGA GTX 980 Ti FTW: -70 MHz
Dell GTX 980 Ti: +105 MHz
eVGA GTX 660 Ti 3GB FTW: -52 MHz
MSI GTX 660 Ti TwinFrozr III OC 3GB: +40 MHz
eVGA GTX 970 4GB FTW: -10 MHz

eVGA is NOTORIOUS for factory-overclocking too high, from my examples.

Fun!

Profile Logan Carr
Send message
Joined: 12 Aug 15
Posts: 207
Credit: 30,647,175
RAC: 32,853
Level
Val
Scientific publications
wat
Message 47252 - Posted: 17 May 2017 | 0:44:33 UTC - in response to Message 47246.

Guys, factory overclocks aren't always stable to begin with. Then, on top of that, when new drivers come out, with optimizations that make portions of the GPU hardware work harder (to perform better) than prior drivers, that too can make an overclock become unstable.

I recommend using GPU-Z, and looking at the "Graphics Card" tab, and look at "GPU Clock", and compare against the URLs below, looking at "Base Core Clock (MHz)", to see how overclocked you are:
https://en.wikipedia.org/wiki/GeForce_10_series
https://en.wikipedia.org/wiki/GeForce_900_series
https://en.wikipedia.org/wiki/List_of_Nvidia_graphics_processing_units

I recommend MSI Afterburner for controlling the GPU Core Clock offset.

I recommend the following test to determine if your GPU clock is stable:
- Download the latest PrimeGrid Genefer OpenCL app
(One version is: http://www.primegrid.com/download/primegrid_genefer_3_3_1_3.13_windows_intelx86__OCLcudaGFNWR.exe)
- Open a Command Prompt window
- Run with the following arguments, including the double quotes:
primegrid_genefer_3_3_1_3.13_windows_intelx86__OCLcudaGFNWR.exe -q "43322502^131072+1" -d 0
- If the application completes successfully with a .log file that has "is composite. (RES=" in it, then you are short-term-stable.
- If you want to run on a different GPU, change the -d parameter
- Another good test is to swap out that number in double-quotes, with this number: "43370168^131072+1"
- You could create a .bat file, to run it overnight, to ensure better stability. Details here:
http://www.gpugrid.net/forum_thread.php?id=4310&nowrap=true#43617

As an example of how I've had to adjust my factory-overclocked GPUs for complete stability at maximum stable clock, my GPUs are:
eVGA GTX 980 Ti FTW: -70 MHz
Dell GTX 980 Ti: +105 MHz
eVGA GTX 660 Ti 3GB FTW: -52 MHz
MSI GTX 660 Ti TwinFrozr III OC 3GB: +40 MHz
eVGA GTX 970 4GB FTW: -10 MHz

eVGA is NOTORIOUS for factory-overclocking too high, from my examples.

Fun!


Okay, I have reduced the core clock to the basic core clock as stated on Wikipedia for the 960, since it was default 1279mhz and the boost was 1342 mhz. Gpu-z now states it is 1127 mhz with a boost of 1190. I will test it out until I get an error, and if I get an error, I'll report back. Thank you all.

So overall I subtracted -152 on the core clock. Please let me know if I perhaps made it too weak?

____________
Cruncher/Learner in progress.

Profile Logan Carr
Send message
Joined: 12 Aug 15
Posts: 207
Credit: 30,647,175
RAC: 32,853
Level
Val
Scientific publications
wat
Message 47253 - Posted: 17 May 2017 | 0:46:42 UTC - in response to Message 47252.

Also my power usage on the 960 has always been around 70% under load, isn't it supposed to be at 100%? my 660 was
____________
Cruncher/Learner in progress.

Jacob Klein
Send message
Joined: 11 Oct 08
Posts: 1067
Credit: 1,146,403,839
RAC: 1,089,717
Level
Met
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 47254 - Posted: 17 May 2017 | 1:57:45 UTC
Last modified: 17 May 2017 | 1:59:32 UTC

Power Usage on the GPU can be anything, I'd ignore it. Though, in MSI Afterburner, I set the Power Limit % to be as high as it'll go.

And, regarding what a safe overclock is... the only way to know, is test test test. As people previously stated, seeing "unstable" in a GPUGrid result is almost always a sure sign you're overclocked too much.

I use the PrimeGrid test to try to find a maximum stable overclock. For each GPU, I find its maximum stable overclock, then set to 10 MHz below the value I found. :)

Post to thread

Message boards : Graphics cards (GPUs) : Unexpected error?