help needed, XFX ATI 7950 DD issue

[Hoobs1981]

help needed, XFX ATI 7950 DD issue *SOLVED

Hey,

My ATI card is probably dying but i wanted to run it past you guys in case there is anything i can do.

The stock speeds are GPU 800mhz, Mem clock 1250mhz and this is on an 800w PSU

At this setting after a while my pc locks up with vertical lines, usually when playing a game or using Youtube.

Within CCC i increased the power control setting to 10%, then 20%, both gave same results.

I tried a fairly large leap and took down the clocks from 800/1250 to 500/750!
since I did this i have not noticed the crash, only when i put it back to stock, i'm not very good with this sort of thing but obviously running the card at half specs is not ideal.

is there anything i can do, is the card not getting enough power at stock setting or is it time to change cards?

cheers in advance

 

[Hoobs1981]

Okay will do, am at work at the moment so will do it tonight, as far as drivers go i have done a complete clean install of the latest catalyst drivers, and monitored temp (not in gpu z) doesn't seem to go above 40c as its on a watercool loop....but I'll get the screen shot of gpuz tonight

 

[Hoobs1981]

Yeah i wasn't even gonna try them, the card isn't new anyway so its a case of fix or shell out on another card unfortunately.

 

[Hoobs1981]

Thanks for all your responses.

My PSU is a Corsiar 800GS, i previously had an OCZ 600 which it did the same thing.

Ill post up some gpuz screens when i get home...issue seems to be random, for example at stock settings i can load up guildwars 2 and leave it is the character select screen and it will crash, sometimes 5-10 mins other times not for around 30 mins. Same with Youtube.

Everytime i have to reboot the pc then when i load back into a game the timing can be random

Due to this issue i currently dont have anything OC'd on my pc

 

[Hoobs1981]

okay so this GPUz and ccc screenshot was taken whilst in the gw2 character select screen which usually causes a crash at stock settings.

As you can see the GPU and Mem clock have been reduced so the crash does not occur

http://imageshack.com/a/img35/4941/fimh.jpg

 

[Hoobs1981]

i also noticed that sometimes the MVDDC will drop for a second, not sure if this has any bearing

http://imageshack.com/a/img89/5792/xln.gif

 

[Hoobs1981]

okay so i clocked back up the GPU clock to 800mhz and have now left the memory clock setting at 750mhz and it is still stable, which is leading me to believe
it's the memory clock that is causing the issue.

Now would this mean the memory is failing, i guess there is nothing i can do about that, or is it something that can be dealt with, with maybe tweaking or something i have not really thought about like Thermal compound misplacement?

when installing my waterblock i used arctic silver on the GPU chip and thermal pads on the vram and vrms.

 

[Hoobs1981]

looking at the max settings in GPUz the VRM temp 2 shows a lot higher than temp 1

http://imageshack.com/a/img203/1681/ud4.gif

although that does seem to just be a short spike then down to 40 again a second later

 

[Hoobs1981]

Just a few ?'s

How long have you had the card?

I assume it came with the stock cooler - Did it?

Did you run it with the stock cooler at all ? - Was it fine?

Which block did you use?

How long before it started going wrong?


It does sound like memory may be failing but before jumping to conclusions .....

Okay so forget what i said about the mem clock, it failed again with the gpu core back up to 800.


i've had the card for a couple of years

It came with the stock cooler, when using it with the stock cooler it still had the issue but not as much, certainly not enough to warrant me underclocking...however the temps are showing it running cooler with the water cooling.

The waterblock i am using is this one:
http://www.aquatuning.co.uk/product...aquagrATIx-for-HD-7950--Tahiti-Pro--G1-4.html

trouble is i cant put a finger on when the problems increased as i had changed somethings i.e, installed windows 8, added that waterblock.

but like i say it occasionally happened before that.

 

[Hoobs1981]

so now ive taken the gpu core down and left the mem core up to see if it crashes again

 

[Hoobs1981]

Your using the same block I am and provided you added a small amount of thermal paste onto each memory chip it should mean that the memory is being cooled correctly.

what i did was add paste to the gpu and then added thermal pads that came with the waterblock to the other chips.

do you think i should take the pads out and use themal compound instead?

 

[Hoobs1981]

so far since reducing the core clock and keeping the memory as stock it's been stable

 

[Hoobs1981]

Hey thanks for your reply,

Yeah will try that, you're probably right, the issue may never go away but it's worth a try, especially as even the stock cooler was poorly pasted / padded when i took it off, so you never know

 

[Hoobs1981]

Hey just to let you guys know i have fixed the problem.

I remembered i did put thermal paste on the ram and put pads on the vrm so i have not taken the water block off to re apply.

What i had noticed is that after extensive testing i noticed the card crashed more when it was a low temp, say 30c, it stopped crashing more when it got a little warmer.

So i increased the voltage through Afterburner and since then its been stable, no crashes no matter what i do.

Now whether this is due to the card having issues running cold, just a voltage issue or even a soldered crack that needed to heat up, either way its fixed for now.

Thanks to all you guys for your replies and suggestions