Card 3 of 4 on GPU Loop (295X2) 10C+ Difference

[Ocmersh]

I am running 2x 295X2 in quadfire, with koolance waterblocks. MSI AB and OpenHardwareMonitor both read that the 3rd GPU(out of the 4 GPUs on 2 cards) is running a full 10C+ hotter while under load. The rest are within a degree or two of one another (hovering around 60-63C under load for the other three). The problem child can reach 73 or 74C causing it to throttle. They are roughly the same temps at idle (Around 40C, I have ULPS disabled). Any ideas on how find the issue? I am using koolance waterblocks on the two cards, I have already tried repasting them. While in idle all 4 temps on all 4 cards are very close. Whilst underload, the temperaturature curve mimics the other 3 but is 10-12 degrees Celsius hotter. Could a bad temperature probecause this? Should I attempt to repaste again?  Thank you. (The teal/green line on the bottom is the CPU temp)

[sms1202]

Probably try running them on stock cooler and identify the issue. Maybe you could send the card for a much needed fix back to the manufacturer

[Ocmersh]

I hadn't thought of the stock cooler idea. I made the mistake I putting on the block soon after I recieved the second one. Although the manufacturer wont do much if it is operating as advertised, wont be much to repair in that case.

[AnonymousGuy]

With my old 680's I had one that would be like 10C hotter than the other.  Think it was silicon related.  Would also check to make sure you had good thermal paste clamping action by diasssembling and inspecting.

[thekeemo]

It shouldn't throttle until 95c...

[Necrodead]

It shouldn't throttle until 95c...

Yeah 75 is still great.

[Footy]

I had the same once and found I had different OC settings on my cards.

[Ocmersh]

Yeah 75 is still great.

They set the throttle temp on the 295x2 at 75C because on the pump that sat on the gpu with the closed loop that came with it. Being im using aftermarket blocks this is now irrellevant, but its in the gpus BIOS

[Ocmersh]

I had the same once and found I had different OC settings on my cards.

 

Did you determine this using stock software or did you have to use aftermarket software to determine it?

[Ocmersh]

It shouldn't throttle until 95c...

 

I don't how to change the temp max thats on this cards BIOS, thats what AMD set it at because of the pump.

[thekeemo]

I don't how to change the temp max thats on this cards BIOS, thats what AMD set it at because of the pump.

http://www.tomshardware.com/reviews/radeon-r9-290x-hawaii-review,3650-5.html

[Necrodead]

They set the throttle temp on the 295x2 at 75C because on the pump that sat on the gpu with the closed loop that came with it. Being im using aftermarket blocks this is now irrellevant, but its in the gpus BIOS

OIC, didn't know as my 295x2 is also underwater and only gets to 60C.

[Ocmersh]

http://www.tomshardware.com/reviews/radeon-r9-290x-hawaii-review,3650-5.html

 

Ya i've been to that screen before. They highest it allows for me is 75C

[Ocmersh]

OIC, didn't know as my 295x2 is also underwater and only gets to 60C.

 

Then I am obviously doing something terribly wrong here.... ive got 2x 560 rads a 400 a 240 and 2x 140 rads. 2x 295x2's in the loop with coolance blocks and the cpu on a phobya block.  Noctua fans and ap182 for the radiators.. Except for the 2x 560's, I just have crap bin 140's on there.... The 60C's im getting on the other card is respectable is the 74C thats really bugging me. Hence this thread. Ill try repasting again and see where it gets me, switching the block might be a good idea too.

[Necrodead]

Then I am obviously doing something terribly wrong here.... ive got 2x 560 rads a 400 a 240 and 2x 140 rads. 2x 295x2's in the loop with coolance blocks and the cpu on a phobya block.  Noctua fans and ap182 for the radiators.. Except for the 2x 560's, I just have crap bin 140's on there.... The 60C's im getting on the other card is respectable is the 74C thats really bugging me. Hence this thread. Ill try repasting again and see where it gets me, switching the block might be a good idea too.

Woah thats some rad space. It must be something in the block/paste as I wouldn't expect anything less than amazing with that much radiator. Did this problem just occur and it was fine before, or is this a new card with a new block? Maybe even gunk in the cooling channels, I would take the block apart and clean it, could be particles from the rads or something.

[Ocmersh]

Woah thats some rad space. It must be something in the block/paste as I wouldn't expect anything less than amazing with that much radiator. Did this problem just occur and it was fine before, or is this a new card with a new block? Maybe even gunk in the cooling channels, I would take the block apart and clean it, could be particles from the rads or something.

 

To be honest this is my first time trying a custom loop. I simply assembled everything and filled. Frustrated with the lack of performance however. A lot of rads to push through but I have 2 phobya 260 pumps in serial to push them. I could just disassemble everything and reassemble. Use that time to wash out the radiators and such. I didn't wash them out initially, or the blocks. Just hot water?

[Necrodead]

To be honest this is my first time trying a custom loop. I simply assembled everything and filled. Frustrated with the lack of performance however. A lot of rads to push through but I have 2 phobya 260 pumps in serial to push them. I could just disassemble everything and reassemble. Use that time to wash out the radiators and such. I didn't wash them out initially, or the blocks. Just hot water?

Rads don't add much restriction, you should be fine on flow. If you didn't clean out your stuff and you have that many rads you may have junk in your block obstructing the flow. For the rads you can wash them out with warm water or warm water and vinegar. For your blocks you may want to take the problem one apart and look at the flow channels for debris. You may be able to dislodge with water but for 100% you'll have to dissassemble.

[Dango]

R9 295X2 will throttle at 75C, its different from R9 290/X.

 

If your block is serial config, then 10C different is normal for dual core GPUs. That's why I always use parallel for my GPUs. 

[brandishwar]

Flush the block. I had to do that with a Koolance VID-NX680 after noticing a major temperature difference between two cards in SLI.

[Ocmersh]

Flush the block. I had to do that with a Koolance VID-NX680 after noticing a major temperature difference between two cards in SLI.

 

Was it a physical obstruction? I'll give it a go then, I have to take it apart anyway so it will be a good opportunity.

[brandishwar]

Was it a physical obstruction? I'll give it a go then, I have to take it apart anyway so it will be a good opportunity.

I don't think it was an obstruction as I changed the configuration from serial to parallel without any significant change in outcome. But once I flushed the blocks the temperatures were about what one would expect between two graphics cards in SLI. If the temperatures didn't change, I would've returned both blocks to Koolance (bought both direct from them) for inspection or replacements.