Well, here it is.
The GTZ squeaked out the EK Supreme by 0.04 C. The D-Tek Fuzion by .25c and my beloved AC Cuplex DI by 1.64 C.
By the way, if your cooling a 45nm quad, and have the Fuzion V2 with quad core nozzle, take the nozzle out. I saw over a 1.0c improvement without it. Not sure how it will act with a Kentsfield, but it hurt the Yorkfield.
Flow seemed equal to the Fuzion V2, and noticably morethan the EK. How much, I'm not sure. I don't having testing capibility yet for flow rate or presure drop.
All the technical specs and internal pics can be found on Swiftechs site.
Test bed: EVGA 790i Ultra.
CPU: Q9450 OC'd-3.500 Voltage under load- 1.28 (provided by Swiftech for testing comparison with their tests) Nice little chip too. I'm going to punish it before I send it back
One Swiftech 355 with XSPC top.
Temp monitoring hardware and software. Crystal Fontz 635 with SCAB. Wintest_635 logging software. Dallas/Matrix 1 wire probes. Accurate to 0.2C within testing range. One probe on each rad fan measuring air in temp. One probe for block fluid temp in. One probe for block fluid temp out.
Core Temp 0.99.1 Tj.- Max adjusted to 95c.
All tests run after 50-60 minutes warm up with Prime 95 using 8k in place FFT's only. Temps recorded for +/- 30 minutes. All temps averaged for Rad Air In, Fluid Temp in and Core temps.
Multiple mounts tested, and best mount for each block taken for comparison. Temps varied per mounts +/- 0.35c. I did however notice a interesting decrease in core temps depending on how long the TIM set up, and amount used.
Here's the slides.
The first is the raw data averages rounded off to .0000 c. Here you can see a comparison of all the temps and supporting data.
Here a graph showing the temps rounded off to 0.00 C.
Note on disassembly::
Everything inside the GTZ has a particular orientation to get the block to perform. Make sure you mark each and every piece of the internals, and re assemble exactly as it was taken apart. There's a specific internal bowing mech. that bows the block when assembled at the factory initially. You'll see what I'm saying when you open the block. If you swap orientation of the internal pieces, it can bork the base and ruin it as far as optimal performance. So the top, internal ring and base plate must go back together exactly as it was from the factory.
To be honest, up until this point I didn't care for Swiftech CPU blocks. Have never owned one or cared to use one. To me there were other blocks better out there, I liked them better, so that's what I used,
However, this block changes my opinion. Given the thermal performance equal to the EK, the thing I like about it best is it's mounting mechanism. It uses a special crafted backing plate that centers itself with small protrusions that stick up thru the MB mounting holes.
It attaches itself with sticky tape, which along with the 4 spars protects the underside of the MB from damage. The best part is how the block sets over the CPU for pre mounting. The bottom of the 4 studs secured to the hold down plate, engages the holes in the backing plate prior to disturbing the TIM on the CPU IHS.
It suspends the block ever so slightly over the CPU before tightening.
Based on my experience, a good mount ends like it starts. The GTZ mounting plate assures that if used properly. It's the best mounting mech. I've ever used or seen come stock on a block. Once tightened all the way down, the bottom of the studs have stops that let the user know they've reached proper mounting tention.
This along with the stiff springs really make for a strong mount. Once the stops are engaged, the top plate virtually turns into a hard mount with the springs nearly compressed all the way. This keeps the block from rocking in all but the most extreme cases. That will come in handy for those tight NB bends
I also like the barb spacing which allows for a variety of fittings, including Bitspower and Danger Den 1/2-3/4 inch compression fittings. But, there in comes the one thing I noticed about the block that I think may cause problems for some. The recess around the GTZ G 1/4 fitting area are somewhat narrow, but deep.
The Swiftech barbs supplied with the block were the older type G 1/4 fittings without recessed areas for the O rings. These worked fine when tightening. The O ring stuck up enough to allow it to seat with feedback when snuggling down.
BP 1/2 fittings on left, Swiftech on right.
However, the BP 1/2 fitttings have a flare or collar around the nut area, and a recessed O ring. When I installed those they sealed fine. But it was one of those things where the fitting just hit the collar and stopped, without any feedback that the O ring was compressing for a good seal like I'm used to seeing. I double checked for leaks and there wern't any. And after repeated mounting and moving the tubing around, I didn't experience any.
But it was one of those things that caught my attention, because it was the first time I've seen a fitting seat like that, so I mention it. Might be nothing, but again something to pay attention to when connecting the tubing.
With thermal performance equal to the EK with less pressure drop, and a better mounting mechanism than the Fuzion Pro mount (which I removed from the Fuzion while testing) Swiftech has a real winner on it’s hands.
I’ll give it my 5 dead horses award as the highest honor that I can bestow on any block.
I'll update with some more pics after I've rested up a bit. I've poured over 60,000 lines of data over the last 7 days. I've enjoyed this opportunity and learned a great deal from it. It's tried my patience at times, but that's on me.
Congrats to Gabe and Stephen for designing and producing a top contender in the WC CPU block world. It's a great block and it will stay on my main rig for quite sometime to come.
Thanks,
andyc