Topic: CGminer - GPU0: invalid nonce - HW error (Read 31073 times)

Not all updates, just security ones. The intent is to prevent the server from being compromised, however ubuntu "security only" updates aren't just security updates.

WHY is it configured to automatically take updates?  You aren't running windows.

I think you have identified the source of your frustrations.

Turn off automatic updated.  If it ain't broken, don't break it.

https://bitcointalksearch.org/topic/amd-134-and-phatk-kernel-on-poclbm-possible-workarround-221041
https://bitcointalksearch.org/topic/bfgminercgminer-catalyst-134-and-58xx-fix-227918

How do I do that?


Benchmark mode results in invalid nonce.

That not a linux driver error only, that's happen also on windows.
Look like amd made change in their driver and amd_bytealign function didn't work as expected.

If you want to use 13.4 driver, just modify the opencl kernel by removing the BFI_INT optimization which cause the problem on 13.4

Try running Cgminer in benchmark mode, eliminate network from variables.

--benchmark         Run cgminer in benchmark mode - produces no shares
--debug|-D          Enable debug output

So it's  " cgminer  --benchmark   --debug  "

I'm certain that none of those are the cause - it's entirely a driver problem. Trouble is I can't get the right driver to work with my newer kernel. It's a linux specific problem I suppose. I'm really wanting it to work on this rig because I use it for oh so many things, but linux GPU support is just horrid.

Reduce intensity.  I have Gigabite 7970.  Intensity over 14 gives the same errors you have.  I started with intensity 8, and then increase 1 at a time until the HW error.

I can run for days without HW error at intensity 13.

Cgminer 3.1.1 
Catalyst driver 13.1

I was having the same problem on my Sapphire 7950 Vapor X - I could only run it at 12 intensity and 375 kH/s, anything over that gave endless HW errors.  I was on the 13.1 driver.  I rolled my driver back to 12.4 with AMD APP SDK of 2.7 and set the following GPU memory command in my Windows command line:

setx GPU_MAX_ALLOC_PERCENT 100

I believe it was this last step that solved my problem.  Now I can run cgminer 3.1.1 at 19 intensity getting 610kH/s (for LTC).  Hope this helps!

I'm being constantly frustrated with this issue, and I'm still not sure the best way to solve it.

I have my ubuntu server running headless with cgminer, and it is configured to automatically take security updates. However it keeps updating kernel versions that break my existing and working radeon drivers. When it does so, I am forced to update the driver. I was at a state where everything was working fine with APP 2.6, however this morning I had a power outage and when the server came back up, cgminer segfaulted.

It seemed that this was due to a kernel incompatibility, which required a driver update. I tried a minor update, which wouldn't install due to some kind of DKMS issue. Only driver that installs is catalyst 13.4. So I updated the driver, seg fault goes away. But now I get the dreaded invalid nonce error (like I've had in the past.) Tried reinstalling APP 2.6, and I get a seg fault. So, I update to app 2.8; no more seg fault, but now I get nonce error.

I like automatic security updates because this server IS exposed to the public internet (its only exposed port protected by FWKNOPD though,) but these compatibility breaking kernel updates have finally done me in, so I'm turning it off (It would be nice if it security updated everything *but* the kernel, and notified me if a very serious vulnerability was found and is in need of immediate attention - otherwise none of these updates have ever been truly needed.)

Having said all of that, is there any known working driver/app sdk combo that works with the 3.7.0-7-generic kernel and bitcoin mining? By the way, this all happens with bfgminer too, old versions and new.

Alternatively, is there a way I can get DKMS installation to work when I install older amd drivers on newer kernels?

Could be due to change in driver ... and more recent version of cgminer use some optimisation unsupported/regression by recent amd driver :

see : https://bitcointalksearch.org/topic/amd-134-and-phatk-kernel-on-poclbm-possible-workarround-221041

I too was having the same issues nonce issues with cgminer past version 2.11.2 on my 6970. CGminer 2.11.2 and AMD Catalyst 13.3 beta was working fine on Windows Vista x64. When I upgraded to cgminer 2.11.3 or newer, I kept getting nonce errors.

I tried Catalyst 12.8 and it didn't work for me, I kept getting the nonce errors. I rolled back all the way back to Catalyst 12.4 and I can use the latest and greatest CGminer 3.11.1

It appears that versions 12.4 - 12.9 use AMD APP SDK 2.7 as per this webpage http://developer.amd.com/tools-and-sdks/heterogeneous-computing/amd-accelerated-parallel-processing-app-sdk/system-requirements-driver-compatibility/

Here's a table from that webpage for future reference:

12.8 seems to have done the trick for radeon 6950 invalid nonce errors. Now I can use cgminer, which is where I wanted to be in the first place.

Thanks, Zanatos666!

clocked to 935, 375 MH/s, temp 57. Not fully tuned yet. Phoenix claimed 415 MH/s at 925 clock and 56 degrees. unstable when clocked to 950.

From CGMiner FAQ

Q: Which AMD driver is the best?
A: Unfortunately AMD has a history of having quite a few releases with issues
when it comes to mining, either in terms of breaking mining, increasing CPU
usage or very low hashrates. Only experimentation can tell you for sure, but
some good releases were 11.6, 11.12, 12.4 and 12.8. Note that older cards may
not work with the newer drivers.

Q: Which AMD SDK is the best for cgminer?
A: At the moment, versions 2.4 and 2.5 work the best for R5xxx and R6xxx GPUS.
SDK 2.6 or 2.7 works best for R7xxx. SDK 2.8 is known to have many problems.
If you are need to use the 2.6+ SDK or R7xxx or later, the phatk kernel will
perform poorly, while the diablo or my custom modified poclbm kernel are
optimised for it.

Sam

Try adding "-d 1 --remove-disabled" to your command line.

Here's the info from the CGMiner Top Post

--device|-d   Select device to use, (Use repeat -d for multiple devices, default: all)
--remove-disabled   Remove disabled devices entirely, as if they didn't exist

Sam

Ive read this a couple of times and it seems to be the same cards being the problem.  Solutions I have seen are:

1) Upgrade CG to at least 3.0.1
2) Roll back your drivers to 12.8

If you are using it for gaming, you might have to decide whether or not you are going to mine if you want the updated drivers for gaming, otherwise just stick to the 12.8 ones that will keep CG happy and still allow you to game.

Having the same problem with a 6950 calling cgminer directly from DOS box in Win7Pro.

Try using BTCGuild's version of GUIMiner and configure it to use anything other than cgminer, e.g. Phoenix with phatk2 kernel.

also see https://bitcointalksearch.org/topic/radeon-hd-5450-errors-139406 - may be a bug in cgminer
specifically see rs77063 reply of 2013-02-04 00:12:18 re Phoenix works, cgminer doesn't

Note: I use cgminer on most of my other miners and I love it. I use GUIMiner (BTCGuild's) on one machine that has a 5970 with one GPU burned out. Can't get cgminer to ignore GPU1.

I started getting the following errors on my wife's Windows 7 x64 gaming system. Additionally Slush's pool reports no activity from that miner in about the same amount of time. It's got an ATI/AMD HD 6950 in it I believe. It may have started after a driver upgrade, but I've since tried a couple newer (including the latest beta) driver versions. I keep the drivers updated on my gaming system & HD 7950 (currently running Catalyst 13.3, the latest beta) up to date & haven't had any issues. There aren't any obvious errors or graphics glitches showing up when she does any gaming, so I'm a little suspicious of it being an actual problem with the video card.

Any thoughts? My only remaining thought is uninstall the video drivers & try to figure out what the old one she had been running, then revert to it.