Spending $445 on this card warrants a dedicated H20 system. Hopefully this will ameliorate the artifacts when overclocking, however; it would be prudent regardless. First I tested the system outside of the case, to ensure no leaks were present.
I've yet to experience a leak using Danger Den hardware. The only time this happened is when I'd placed tension on the tubing from the TDX, CPU-water block outlet entering the bay reservoir inlet. Rushing, I failed to check the lines, after hearing the all too familiar sound of air entering the system. Just a few drops escaped, but enough to cause the system to shut down, and in my stupidity, I tried restarting it twice Prior to realizing a few drops had fell onto the card, and into the PCIe slot. At 75W there's a significant amount of current there. Even more troubling, was the fact this was the very first time the card had been powered up. After removing the card and allowing the system to dry over night everything seems to have turned out OK. Nonetheless, it was a nerve racking experience, and I've definitely internalized the rule of immediately discovering the source when air enters your system. When installing a new system it's important to measure and connect the hoses without clamps, then place the entire system into the case prior to adding water so you'll know how the hoses are going to flex and twist. Then you can attach the zip-ties or other fastening devices securing tubing to all components (radiator, reservoir, pump, etc.). Take note in the photo above, the outlet from the GPU water block is fed into the reservoir and that tubing is angled upward. The idea was to eliminate any unnecessary tension on the block, ergo the video card.
Now let's take a look inside the MAZE4 GPU water block
The MAZE4 is based on a channel design which doesn't center the inlet stream onto a specific impingement zone centered above the GPU core. At first I was a bit perplexed by this. While it may seem like an illogical design it's important to understand there are vast architectural differences between GPU's and CPU's. Without delving into the subject beyond a prima facie explanation, heat dispersion on cores such as the R420 or NV40 simply aren't given the same priority as more complex CPU water block designs. The reasons behind the often elementary approach to GPU block design based on several principles. Foremost would be cost, if there is such a thing as an average water cooling Enthusiast, their expectation for a GPU block design are not as demanding as for CPU design. Since GPU blocks are often ancillary devices, they are less costly and complex.
This philosophy, however; is changing rapidly and among some European H20 makers, might seem archaic, as even their older GPU water blocks are the most intricately designs I've seen.
Danger Den, however; hasn't become the premiere H20 manufacturer by remaining complacent. In fact if there is an outdated model it might be the block featured here. Nonetheless, as you’re about to see it's still quite an effective device. Danger Den's recent
6800 GPU block exemplifies a much more thorough approach to video card cooling, as the design not only cools the NV40 core, but the card's memory as well.
Another reason one finds less complex impingement zones in GPU design, would be flow related. As 90% of all multi block designs are in linear, fed from a single pump, there's a substantial pressure drop by the time water exits the CPU block and enters the GPU block. Another factor dictating the design of the MAZE4 would be temperature. As the water entering the GPU will be carrying with it heat transferred from the CPU, its temperature will be raised significantly above ambient, if a temp that low can be attained in the first place.
Looking at the large channels above, one may see the logic behind this block to be well served. The intent is to allow the least amount of restriction on flow and the best circulation of the water through the channels such that it travels through the block as rapidly as possible removing heat with it. It's apparent the water exiting the CPU block is lower in temperature then the GPU itself, otherwise multi-block systems would obviously fail. I wouldn't advise against multiple blocks in-line necessarily, however; there are methods to avoid feeding heated water into the GPU block. Employing a powerful pump, and dual radiator, one can split the outlet from the radiator, so there's a separate feed to each block, and likewise in return utilize a Y-adapter to merge the heated water from both blocks into the reservoir.
In this review I placed the block in an isolated system. The photo below shows the final installation.
Onto the performance tests ->
