I just wanted to say that I was probably one of the first guinea pigs for the new style XSPC 7970 block a while back (I got it right when it was released)... I noted a lot of problems with it and e-mailed XSPC and never heard back from them. I got it to work, but VRM temps are quite bad and getting the block to lay correctly (or as close to correctly as possible) on the GPU die is a chore.
It's been a long time since I e-mailed them about the problems with the block, but it might be interesting for other's to read... here it is:
I have a few concerns to raise with you regarding the new style of the 7970 water-block. Please take the time to read this thoroughly as I will try to explain in a clear and concise manor.
1) The square that is intended to make contact with the GPU die makes poor contact. This could be in part due to the fact that ONE of the FOUR crucial mounting points around the GPU core was omitted from the design for some reason... These 4 points seem crucial to applying even tension across the core... Only having 3 of the 4 does not seem to apply appropriate pressure. Also, it may seem that the square that is intended to make contact with the GPU die is not thick enough as it requires lots of thermal paste to make up the gap... Then again, maybe it all comes back to the fact that the one point of mounting is missing, this could be all it needs to make better contact with the GPU die... I don't quite understand the logistics of this, perhaps it was a design over-sight?
2) This next one is even more problematic and the largest area of concern... I'm sure you've heard feedback regarding the cooling of the VRM on the cards... This is a serious area of concern. There are multiple points to this problem and I will do my best to explain them thoroughly.
The TIM given to cover the VRM ICs (blue) is causing problems (perhaps it's slightly conductive?). If the blue strip of TIM over the ICs is too close to the ferrite cores next to it (the parts that use the grey TIM), as soon as you load up a 3D application the video card fails resulting in monitor not receiving signal and computer locking up entirely within seconds - sometimes immediately. If I pull the TIM over a little so that it does not touch the ferrite cores, it will not entirely lock up; however, it will cause the display driver to crash and cause the video card to throttle back to 500MHz core, 150MHz memory (obviously a major problem). I thought to myself, maybe the problem lies within how the TIM is lying on the blank parts of PCB with solder points (on Reference 7970's there is 2 VRM ICs missing in the row, opposite of the PCI slot), so I figured I would cut out the piece of TIM so that it covers the ICs ONLY and not the bare part of PCB with solder points on it... This prevents the card's drivers from crashing immediately; however, it still continues to throttle down about 150MHz on the core and 200MHz on the memory clocks from what it should be running at. It should also be mentioned that the VRM temperatures at this point are absolutely horrible, spiking to around 115 degrees Celcius after only a few minutes of testing...
Now here is the interesting part... I tested removing the blue TIM over the VRM ICs entirely (however, leaving the grey TIM for the Ferrite Cores)... and the card DOES NOT CRASH OR THROTTLE and the temperatures, while still on the hot side are NOT as hot as with the TIM. It appears having NO COOLING on the VRM ICs is better than having the black piece of metal with TIM touching them. The card is functional without the blue TIM on the VRM ICs. My big problem with this is, I do not feel safe running them without cooling (although it technically being cooler than with). Plus, honestly, I purchased this with the intent to overclock and with it's current performance and obvious issues, that is not a foreseeable possibility.
Is the blue TIM conductive, even slightly? Is this causing some kind of short across the solder points when it's attached?
I'm sure you've received criticism over the fact that the VRM is not actively cooled (which is a big reason to have a "full coverage" water block in the first place is it not?). If the design actually worked as intended I would have no problems with it, but there are some serious issues and it's causing my card to not operate properly (or at all). I have now taken this block off the 7970 to re-seat everything and make adjustments many more times than I should have and it's causing me much grief. I wanted to bring up these potential design flaws to you early in the product's life and also see what could possibly be done for people like myself who already bought the product and are left with something that renders their video cards useless? As sad as it sounds, problem #1 regarding the GPU was actually the lesser of the two problems as the card will function OK if it weren't for the VRM issue. Also, the VRM temperatures will limit overclocking substantially even if the locking up/shorting problem was not occurring. Part of the reason to purchase a full coverage water-block is to be able to do some overclocking on your gear. I can tell by GPU temps (when I get a decent enough amount of thermal paste on the GPU to make contact) are decent, though would probably be better with better contact in the first place (though how do you not utilize all 4 mounting points around the GPU core?!) Despite that, the bigger problem area is the VRM. Like I said, I do not even mind that the VRM is cooled passively if the design worked, but it causes lock ups...
Honestly, if I knew this block would be such problems I wouldn't have purchased it... However, I use a lot of XSPC gear and wanted my 7970 block to match my other gear, especially the new Raystorm (I have a Raystorm CPU block, RASA in my other build, one of the dual bay reservoirs, your compression fittings, and an RX360).
Please read this carefully and please advise. I eagerly look forward to your response on this matter. If any additional information is needed from me, I would be more than happy to explain further.
Thanks for taking the time,
(I had realized later, I should have just referred the "ferrite cores" as chokes... oh well, not like they responded anyway.)
Edited by TamaDrumz76 - 7/10/13 at 3:06am