Thought I'd update everyone on performance of the nateman_doo chipset block. I can only share flow rate impact right now....not temp data. But more on that in a minute.
I'll spoil the surprise,and start off by saying that I had a nice pick-up in flow rate when I swapped out the Mips chipset block for the nateman_doo chipset block. Sorry. Late spoiler alert. But, I thought this would be a good time to summarize all the changes I've made to the loop that have impacted flow rate....and this block was one of them. So....now you're forced with the decision of whether to read all the gobbledy gook in the middle of this post....or skip straight to the end to see the latest impact. Your call.....
So....a couple of months ago, I was getting about 0.95 gpm flow rate. While this is definitely not poor, it was a little surprising to me to see this as the max, since I had 4 pumps running. I had been expecting a little higher flow rate. Yeah...it's having to be pushed through a lot of blocks, as well as 7 radiators, .....but I mean....it's 4 pumps. Come on. Over the course of the last couple of months, I have made many changes to blocks, as well as the loop, and have recorded the impact of each. Time for a review:a)
placed the two AMS radiators in parallel flow. This added about 0.05 gpm to the flow rate, bringing it to about 1.00 gpm. Not huge....but every little bit is helpful.b)
added a valve right above the OU reservoir, so that the recirculation loop could be shut off. When this was added, and the valve was closed....it added 0.10 gpm to the flow rate, increasing it to 1.10 gpm. A nice little pickup. This clearly showed that the 7 radiator path of the loop was less restrictive than the waterblock portion of the path. When the coolant was forced to make a serial loop instead of allowing it to choose the path of least resistance, this forced more coolant through the waterblock portion. All testing since this change has been done with the valve closed.c)
I removed the Mips memory waterblocks from the loop layout. The blocks are still sitting on the RAM sticks...but not included in the loop layout. This increased the flow rate by 0.10 gpm, bringing it to 1.20 gpm. Getting better.d)
I swapped the Mips VRM waterblock, for the nateman_doo VRM waterblock. This increased the flow rate by 0.05 gpm, bringing it to 1.25 gpm.e)
And last but not least, I have now swapped the Mips chipset block for the nateman_doo chipset block. This increased the flow rate by another 0.10 gpm, bringing my new max flow rate to 1.35 gpm.
I'm sure some of you can probably get 1.35 gpm with just one or two pumps. But I think mine is slowed down somewhat from all the blocks, radiators and fittings. Even without the memory blocks, there are 8 other waterblocks in the loop right now, along with 7 radiators, 2 flow meters, etc. If I had to take a guess at the next most restrictive part of my loop, I would have to guess the 2x aquagrafx GTX 690 waterblocks. These blocks are huge and although I have not seen any specific flow rate tests done with them....some previous aquagrafx benchmarks for other cards showed they were very restrictive for video card blocks. I can only imagine that these are also. But....my temps are good, so I don't feel the need to search for other ways to improve flow rate even beyond 1.35 gpm. Actually, I'll probably put back on the Mips memory waterblocks because cooling is more than adequate at 1.25 gpm, which is what the max flow rate should drop back down to.
And since the pumps will be running at reduced speeds for the majority of the time anyway....this should be fine.
Now...here is one more shot of everything running again in it's current state.....before I have to tear it down again.
Areca card needs to be sent in for servicing. Could not recover from the firmware upgrade. Won't load a BIOS now. And since I've made some pretty substantial modifications to the card...I'm not going to hold my breath that they will cover it under warranty. But...we'll see. I'd actually love to see their faces when they open the package and find my custom water cooled solution and copper heatsinks all over their card.
I can still work on some wiring/stickers/etc while card is away....but this will slow me down yet again.