Associate
Some of you may have seen my woes with regards to the Sapphire 7990 card I have been mining with for 24/7 for the last 3 or so months.
The card runs hot as you probably know and so I went down the water cooling route with a big heavy EK copper water block.
All started well with temps from mid 90's down to mid 50's. I am running the card 100% on both GPU's 24/7
However I was constantly getting GPU no1 at 4 degrees higher than the 2nd GPU and over the course of a week or so, the difference would increase to beyond 10 - 12 degrees.
Every time I stripped down the water block, re-applied the paste and connected all back up again I would go back to square 1 again and again slowly over the cause of a few days the GPU's would start to wander again.
Racked my brain over this and got very frustrated.
Eventually flex was questioned. Even though the copper block is extremely stiff and the PCB is firmly attached to it, I reckoned that due to extreme 24/7 temperatures I was placing on the card, it was flexing slightly and due to constant heat the PCB was flexing away from one of the GPU's.
So complete strip down.
1. Dis mantle the water block and give it a good proper clean.
2. Re-order new thermal pads and give these a new lease of life
3. Re order my water loop so it is pushing against gravity and not with. Also helps with bleeding the loop between GPU, CPU and 2 x radiators.
4. Add in a wooden block to support the card and stop flex.
The temps are within 1 degree of each other and it may be the flex that is the problem.
I will keep an eye on them over the next week and if it works, I will get a more pretty solution than a wooden block.
If you are experiencing the same problems, it could be flex with such a big and heavy card hanging off a PCI port by only 2 screws.
The card runs hot as you probably know and so I went down the water cooling route with a big heavy EK copper water block.
All started well with temps from mid 90's down to mid 50's. I am running the card 100% on both GPU's 24/7
However I was constantly getting GPU no1 at 4 degrees higher than the 2nd GPU and over the course of a week or so, the difference would increase to beyond 10 - 12 degrees.
Every time I stripped down the water block, re-applied the paste and connected all back up again I would go back to square 1 again and again slowly over the cause of a few days the GPU's would start to wander again.
Racked my brain over this and got very frustrated.
Eventually flex was questioned. Even though the copper block is extremely stiff and the PCB is firmly attached to it, I reckoned that due to extreme 24/7 temperatures I was placing on the card, it was flexing slightly and due to constant heat the PCB was flexing away from one of the GPU's.
So complete strip down.
1. Dis mantle the water block and give it a good proper clean.
2. Re-order new thermal pads and give these a new lease of life
3. Re order my water loop so it is pushing against gravity and not with. Also helps with bleeding the loop between GPU, CPU and 2 x radiators.
4. Add in a wooden block to support the card and stop flex.
The temps are within 1 degree of each other and it may be the flex that is the problem.
I will keep an eye on them over the next week and if it works, I will get a more pretty solution than a wooden block.
If you are experiencing the same problems, it could be flex with such a big and heavy card hanging off a PCI port by only 2 screws.
