Beta Orionis – Part XIV

Build Log:

It’s still on the heart lung machine. The replacement AX860 I received is exhibiting the same problem as before under load. Initially it was fine for a couple days, then it started cutting out on me. This actually makes be believe the mainboard may be the concern, not the power supply. But then the question is why the problem would be exhibited with an AX860, which is a platinum rated 860W power supply, but not the GS800, which is a bronze-rated 800W power supply.

Something just doesn’t add up on that.

So a replacement mainboard is on the way from NewEgg. I’m not replacing it with another Gigabyte mainboard — I want to be able to overclock this, and the weird power cycling this mainboard does when you toggle a couple settings has me worried. I’m going instead with the Asrock 990FX Extreme6. It’s basically a blue version of the Fatal1ty Killer board, but without the E2200 LAN chip and a 1-year warranty instead of a 3-year warranty. So if the board dies anytime in 2015, I can RMA it. After that, might was well look at upgrading.

If I can confirm the board is the problem, I’ll RMA it and then just set aside the replacement for something else.

Moving on…

New graphics card blocks

In the previous section I mentioned that I was utterly disappointed with the temperature performance on the Koolance blocks I had on my graphics cards. It seems that Koolance’s VID-NX680 just wasn’t up to the task. I mentioned that I opted to replace it with the Watercool Heatkiller GPU-X3 since I was able to find them for a great price from FrozenCPU — who actually sent me 5 of the backplates instead of just 2. And no I’m not kidding on that last one:


That’s 5 backplates on top of the two water blocks — probably a little difficult to see underneath. The order called for two blocks with a backplate for each, and they sent me 5 backplates. With several people checking these orders, how did this escape notice? And when the person who initially packed it saw they had packed 5 backplates for just two water blocks, that immediately should’ve prompted scrutiny to ensure they were getting things right.

The extra plates went back Monday afternoon via 2-day USPS Priority Mail — a week after I received them.

So the big question, of course, is how well they perform. And truthfully, the difference is like night and day. These are the temperatures running Heaven Benchmark 4.0 on its benchmark option — max column is the important one.


There’s still a slight differential in temperature, but this is about what you want to see: at most only a few degrees deviation. With the Koolance blocks I was seeing 10C difference between the cards, with one card still reaching 60C. So this is quite a stark contrast. The temperatures for Valley Benchmark were about the same: 3C difference with the hotter card reaching 45C. Recall that this is also with three radiators, all of them ST30s, with two 240mm and one 360mm radiator.

Certainly a major improvement. And I should add these temperatures are with the cards running in parallel.

Now this doesn’t mean that Koolance’s block is bad. It just means you can’t use it for any serious load. Perhaps the blocks I received were faulty, or perhaps it’s just Koolance’s design. Either way, I’m glad I went with the Watercool blocks. Plus the backplates give a little better piece of mind with having that weight on the card. Plus the nickel and copper go well with the fittings and copper tubing.

So now it’s just about waiting for the mainboard to arrive from NewEgg — hopefully in time for the weekend as that’d be the perfect time to take this whole thing down. I will need to pull new tubing — again! — for the lines going to and from the CPU block, and I’ll probably re-evaluate the fitting setup I have coming off the top radiator at that time as well, likely replacing it with straight tubing or some other kind of bend.

Update: After playing Bioshock Infinite for about an hour and a half, these were the observed temperatures — again, max column is the important one. Compare this to the previous temperatures with the Koolance blocks. Again only about a 3C difference between the cards, and these Watercool blocks are keeping them nice and cool. I should’ve gone with these first and saved myself the trouble.