Long story short(er):
A couple weeks back, I setup an i7 rig to replace my aging FX64 system. I temporarily installed a Zalman CNPS9900 HSF while I waited for all of my water cooling parts to come in. I think the Zalman warped my mobo.
Currently, I have an i7 920 OC'd to 3.6 under a DD MC-TDX WB (no backing plate was included), and an EK 295 GTX block on an EVGA 295GTX, a Swiftech MCR32-QP-K rad, three Scythe "S-FLEX™" S-FDB 120mm Quiet FAN - SFF21G fans, and an Eheim 1250 pump. All one loop. My reservoir coolant temps idles at 29 (my ambient is controlled at 22 C) and maxes at 32 C under load.
GPU temps are great; neither core ever goes over 55. However, my CPU temps *immediately* jump to the high 70's when the CPU is loaded. At least, that's the temps I'm getting now... I was hitting temps in the high 80's until three days ago. After reading a bunch of threads here (and especially Gabe from Swiftech's comments about it), I began to suspect that my mobo was warped, causing TIM joint problems. So I took the DD WB off the CPU. Imagine my surprise when I saw two areas of TIM that hadn't even been *touched* by the WB. So I replaced the TIM with some old AS5, and remounted the WB. All my temps have dropped ten degrees, but the CPU temps are still hitting and peaking in the high 70's immediately upon loading the CPU.
This has me thinking that my problem still lies in the TIM joint. If I was limited by total thermal cooling capacity, the temps would gradually rise over time as the coolant reached it's max temp for the cooling capacity available. Since it's peaking and stabilizing immediately, that makes me think the problem lies in the transference of heat from the IHS to the WB. As I said earlier, my coolant temp (reservoir) idles at 29 and maxes at 32 under load. To me, that says I still have plenty of cooling capacity that's not being used. Additionally, changing my fan speeds has no effect on CPU temps, either. Whether they are barely spinning, or maxed out, the CPU temp stays exactly the same, loaded or idle.
I had two choices to fix this: 1. Order the backing plate for the DD block to try and take out the MB warp and give me a better TIM joint, or 2. Take the expensive route and get a Swiftech Apogee GTZ WB that includes the backing plate. So of course I chose option 2. :) The GTZ should be here either tomorrow or Tuesday. I'm hoping that the backing plate will help to stiffen the mobo enough to where I can get a decent TIM joint and drop the temps down where they should be.
Anyway... purpose of this rather long winded post:
1. I'd love to see what temps you guys are getting on an i7 on water. There's a few posts around showing i7 temps, but it seems most of the temp posts are from other CPU's.
2. What other ways would you guys recommend of getting the flex out of my mobo so I can get a better TIM joint?
3. I'm wondering about using, for lack of a better term, a "heat spacer". A piece of copper cut to the shape and size of the CPU, lapped and polished. This would be placed between the WB and the IHS. Since I don't have access to the stuff I'd need to make this myself, I was considering using a TEC block (unpowered, of course). Would a TEC block efficiently transfer heat from the IHS to the WB if the TEC wasn't powered?
Thanks in advance for help/suggestions/critiques, etc. :)