Topic: 6950 mining problems (Read 2016 times)
I think I've got it sorted now. I flicked the BIOS switch to the original BIOS and it's much more stable with the shader unlock BIOS for some reason. On the stock BIOS it won't even be stable at 850MHz with stock volts, or with 1.200V. It simply doesn't want to overclock at all on the stock BIOS without giving problems. At the moment I've had it running for a few hours with the shader unlock BIOS at 930MHz @ 1.200V. I then shut the computer down and started it up again just to be sure, with no issues. I don't really want to push any more voltage through it since I'm questioning the long term longevity of the card if I push any more voltage through it. It's doing about 410MH/s at 930MHz so I'm pretty happy with that.
Try stock clock and voltage..... See if you still get crashing (Screen goes blank and Windows display driver recovered)
If that is the case I would try running it as a 6950 overclocked without the shaders unlocked.
If that is the case I would try running it as a 6950 overclocked without the shaders unlocked.
Sometimes being how often? If you've got them unlocked still, drop them to stock and run them for a couple of days. If you get no display driver crashing, it's the overclock. Slowly dial the speed back up until you start hitting driver crashes again.
Sometimes the driver will crash within 2 seconds of mining, other times it will go for a few hours.
Dude, if windows driver is crashing it is NOT stable.
Thank you!
Thank you!
I seem to have got it stable now, except for the fact that the screen sometimes goes blank and I get a message in the system tray saying "display driver has crashed and successfully recovered". Is that due to an unstable overclock?
Sometimes being how often? If you've got them unlocked still, drop them to stock and run them for a couple of days. If you get no display driver crashing, it's the overclock. Slowly dial the speed back up until you start hitting driver crashes again.
An unstable overclock is the most common issue , your unlocked shaders if flaky can do that too
I seem to have got it stable now, except for the fact that the screen sometimes goes blank and I get a message in the system tray saying "display driver has crashed and successfully recovered". Is that due to an unstable overclock?
I think what a lot of people are saying is that if those locked shaders were locked because the chip failed validation, even more voltage won't necessarily make them stable.
Bingo
Turn them back off and then try OCing the card. I have some Sapphire cards that just plain unlock with the BIOS switch and they are quite stable but the two I tried to flash never did get stable on the unlocked shaders. I switched them back and just OC'd them as normal 6950's. I was a lot more pleased with their performance as great 6950's than as crippled 6970's.
I think what a lot of people are saying is that if those locked shaders were locked because the chip failed validation, even more voltage won't necessarily make them stable.
use cgminer, it will tell you if you're having hardware errors off the hop... very easy to use... 
I'll have a look into it. I just discovered the power control settings in Afterburner, so I set it to +20%. Hopefully that will help.
use cgminer, it will tell you if you're having hardware errors off the hop... very easy to use...
I tried increasing the power control settings in CCC but it keeps going back to 0% whenever I try to save the settings. What do I do to prevent that from happening?
Of course it's doing something, it's pumping more voltage into the chips and raising the temps.
To raise the temp the chip operates at, raise the frequency and/or raise the voltage.
That's one reason people underclock and undervolt their mining hardware.
What it can't do is make the new unlocked shaders 100 per cent stable. Borked is borked.
To raise the temp the chip operates at, raise the frequency and/or raise the voltage.
That's one reason people underclock and undervolt their mining hardware.
What it can't do is make the new unlocked shaders 100 per cent stable. Borked is borked.
When I change the voltage in Afterburner from 1.1V to 1.2V I notice a big difference in temperature, even if clock speeds are the same, so the voltage adjustment must be doing something.
Especially with an unlocked 6950 as there is no guarantee those unlocked shaders are good... doesn't matter much in Furmark because it just pops an artifact, but with mining it starts tossing errors into calculations and just stops working.
You can't change voltage in AB with unlocked bios. Try to underclock memory and start overclocking from 800Mhz with 10Mhz step. Also, practice has shown, that even hour in Furmark - does not guarantee stable mining.
I'm using GUIMiner v2012-02-19 with a reference 6950 with a shader unlock BIOS mod. The card is perfectly stable with Furmark at 900MHz core for a few hours @ 1.2V. The problem is when I'm mining if I run it at any higher than the stock 800MHz clock speed it will mine for a few minutes, even up to 30 minutes and then it will just stop mining. GUIMiner will say that it's still mining, but looking at Afterburner it says the GPU is idling, and Deepbit will also say that it's not mining. I've tried increasing the voltage to 1.225V, but then it still gives the same issues. If I run it at stock speeds and volts then it's fine all day long. The GPU temperature doesn't exceed 70°C.
Sometimes the screen will flash and I'll get a message in the system tray stating that the AMD GPU driver has crashed and successfully recovered, other times I don't get that message. I'm using Catalyst 12.6.
EDIT: Now I just got a BSOD with 870MHz @ 1.2V.
Sometimes the screen will flash and I'll get a message in the system tray stating that the AMD GPU driver has crashed and successfully recovered, other times I don't get that message. I'm using Catalyst 12.6.
EDIT: Now I just got a BSOD with 870MHz @ 1.2V.
Jump to: