tl;dr
2500K at stock clock/volts has a load temp of about 28C over ambient. Loop is a Cuplex Kryos Delrin, Swiftech MCP35X, PA120.3 w/3 Yate-Loon D12SM-12s. Rad is external. Loop has been bled and run for about 2 weeks, waterblock has been remounted, tried more/less mounting pressure w/no change, TIM contact patch looks fine. Is this normal or a problem?
Long version
I went back to watercooling after many years on air - my last water-cooled rig was an AMD Opteron with GTX 6800s in SLI.
I have a 2500K on an Asus P8Z68-V PRO. It's under a Cuplex Kryos Delrin. Pump is a Swiftech MCP35X, radiator is a PA120.3 with 3x Yate-Loon D12SM-12's (1650RPM) pulling through the rad. At the moment the rad is outside the case. Case is a Corsair 600T - I temporarily sat it on a radbox over the top grill, but I'm planning on getting a better case for an internal mount and cooling the GPU later. For now it's just the CPU, and the PA120.3 is almost certainly overpowered for a CPU-only loop - yet the temps seem rather high. In fact, they're only a few degrees lower than they were under the TRUE that preceded it.
At stock speeds and voltages, the CPU cores idle at 30/38/38/30 (cores 1 and 4 often fluctuate between 30 and 38 in less than a second with no graduation in between, which is weird - the temps on my last loop were really stable and climbed slowly even changing from idle to full load). Ambient is 28, so I guess the average core temp is 6 degrees over ambient, which isn't bad. Measuring with CoreTemp 0.99.8. RealTemp 3.70 doesn't disagree by more than a degree or two.
Under load, they go up to 51/61/61/51. That's an average of 56 degrees, which is 28 degrees over ambient, which seems very hot for stock volts and temps with this much radiator power.
I thought it might be the mount as the Cuplex has a really crappy mounting system, but a remount didn't do anything. TIM contact patch looked good. Also, since there's no way to gauge the "correct" pressure with its mounting system, I noticed the CPU socket area was noticeably bowed on the back of the motherboard; I backed off a little on the mounting pressure but that did nothing either. I have no way to measure the water temp, but I notice after an hour or two of BF3 that the radiator and the air coming out of it are noticeably warm, so there must be good contact between the waterblock and the CPU - right?
I shook all the air I could out of the loop during bleeding and leak-testing. The CPU block was harder to bleed, but I think I got it all. The loop has been up and running for two weeks now, so you'd think any small air pockets would have worked their way out by now.
Any ideas? I'm considering changing CPU blocks to something like the Apogee HD, with it's better mounting system, but I don't want to overlook anything. Or do these chips just run that hot? I don't have any experience with them - last CPU was a Wolfdale that ran way cooler.
Bookmarks