So I recently (okay, it was about a month ago, but I've been too busy to post about it till now) added on my first watercooling system to my desktop. All well and good - learned a LOT about the whole watercooling scene, had a lot of fun experiences with leaks, and was happy getting the whole thing working in general. It's now up and running and I've yet to have any showstopping issues with the system. That said, I've noticed an odd problem with one of my GPU's in the loop not performing the same as the other two and was wondering if anyone has seen anything like this.
So lemme break down my build/loop before I get into the problem:
CPU: Core i7-3960X (O.C.'d to 3.9 GHz currently)
GPU: 3 x EVGA GTX 580's
CPU Waterblock: Swiftech APOGEE HD Ultimate Extreme Performance Water Block
GPU Waterblock: eVGA Hydro Copper 2 Waterblock for GTX 580
Radiators: 1 x Black Ice GTX Extreme 120 and 1 x Black Ice GTX Extreme 360
Fans: 8 x Xigmatek XAF-F1255's
Resevoir: EK-Spin Resevoir
Pump: MCP655
And that pretty much covers it. Now the watercooling system has been running great. Temps are far lower than they ever were on air cooling, especially the 3-way SLI graphics cards, as those used to run at about 93 C! But one of the three graphics cards is standing as a sort of "odd-man-out."
Generally, my temps look like this idle:
CPU - 50 C
GPU1 - 36 C
GPU2 - 42 C
GPU3 - 35 C
As you can tell there, GPU2 is running a little hotter than the other two cards. But no biggie, it's a minor difference. The problem comes into play when I start using the cards. To test the cooling powers of the build, the first thing I started running was the Unigine DX11 Benchmark. It's a stressful benchmark for the GPUs, pushing them all to a constant 99% usage. GPU1 and GPU3 perform great! They go from 36 C idle to about 42 C on full load and stick there for hours. GPU2 on the other hand? Well he's kind of *****y and he IMMEDIATELY jumps from his 42 C idle temp to about a 50-55 C range. From that point on, GPU2's temperature continually rises and rises and rises. After about 30 minutes of benchmarking, GPU1 and 3 were at about 42 C while 2 had climbed to 78 C. I'm sure if I kept running the benchmark, GPU2 would eventually overheat and crash the system.
Now clearly, this is a problem, but it doesn't have too big of an effect on day-to-day usage. Generally, gaming never usually pushes all three cards to 99% usage which is the only case where I see such rampant out of control temperatures from the second card. Normal usage of the cards really only ever sees the GPU's hitting about 50% load, where GPU1 and 3 have the same temp cap and GPU2 hits somewhere in the high 50s.
Now, the thing is I'm not really sure why only one of the cards isn't being cooled like the rest. Could it be a problem with the waterblock's installation? Thermal compound not applied correctly? All three were installed identically, but I suppose I could have done something odd while installing the waterblock. Or could the waterblock be damaged? How would I check this? What steps should I take to diagnose the issue?
I've been apprehensive about breaking down the entire system to remove the waterblock, so I haven't tried reapplying it yet. I'm sure that's probably going to be the first step in troubleshooting, but draining the loop and breaking down the setup hasn't been feasible with how busy I've been recently. Just curious as to what more experienced watercoolers might know about my problem, before I go through the trouble of breaking it all down.
So lemme break down my build/loop before I get into the problem:
CPU: Core i7-3960X (O.C.'d to 3.9 GHz currently)
GPU: 3 x EVGA GTX 580's
CPU Waterblock: Swiftech APOGEE HD Ultimate Extreme Performance Water Block
GPU Waterblock: eVGA Hydro Copper 2 Waterblock for GTX 580
Radiators: 1 x Black Ice GTX Extreme 120 and 1 x Black Ice GTX Extreme 360
Fans: 8 x Xigmatek XAF-F1255's
Resevoir: EK-Spin Resevoir
Pump: MCP655
And that pretty much covers it. Now the watercooling system has been running great. Temps are far lower than they ever were on air cooling, especially the 3-way SLI graphics cards, as those used to run at about 93 C! But one of the three graphics cards is standing as a sort of "odd-man-out."
Generally, my temps look like this idle:
CPU - 50 C
GPU1 - 36 C
GPU2 - 42 C
GPU3 - 35 C
As you can tell there, GPU2 is running a little hotter than the other two cards. But no biggie, it's a minor difference. The problem comes into play when I start using the cards. To test the cooling powers of the build, the first thing I started running was the Unigine DX11 Benchmark. It's a stressful benchmark for the GPUs, pushing them all to a constant 99% usage. GPU1 and GPU3 perform great! They go from 36 C idle to about 42 C on full load and stick there for hours. GPU2 on the other hand? Well he's kind of *****y and he IMMEDIATELY jumps from his 42 C idle temp to about a 50-55 C range. From that point on, GPU2's temperature continually rises and rises and rises. After about 30 minutes of benchmarking, GPU1 and 3 were at about 42 C while 2 had climbed to 78 C. I'm sure if I kept running the benchmark, GPU2 would eventually overheat and crash the system.
Now clearly, this is a problem, but it doesn't have too big of an effect on day-to-day usage. Generally, gaming never usually pushes all three cards to 99% usage which is the only case where I see such rampant out of control temperatures from the second card. Normal usage of the cards really only ever sees the GPU's hitting about 50% load, where GPU1 and 3 have the same temp cap and GPU2 hits somewhere in the high 50s.
Now, the thing is I'm not really sure why only one of the cards isn't being cooled like the rest. Could it be a problem with the waterblock's installation? Thermal compound not applied correctly? All three were installed identically, but I suppose I could have done something odd while installing the waterblock. Or could the waterblock be damaged? How would I check this? What steps should I take to diagnose the issue?
I've been apprehensive about breaking down the entire system to remove the waterblock, so I haven't tried reapplying it yet. I'm sure that's probably going to be the first step in troubleshooting, but draining the loop and breaking down the setup hasn't been feasible with how busy I've been recently. Just curious as to what more experienced watercoolers might know about my problem, before I go through the trouble of breaking it all down.
