D-Tek Fuzion GFX Vs. Swiftech MCW-60 Showdown - 8800GTX
Ok, so this is my first complete comparison of two blocks, so if you see something I should change in my process, let me know.
INTENT
Compare the following two blocks for both pressure drop and thermal performance on an Evga 8800gtx. BTW I paid for both of these blocks out of my own pocket, niether company supplied me with a full block.
http://www.overclock.net/gallery/dat...sidebyside.jpg
Swiftech MCW60 with G80 adapter kit - $47 at Petra's
D-Tek Fuzion GFX - $50 at Petra's
INITIAL IMPRESSION
The swiftech MCW-60 is a solid performer and sort of leaves you with the impression you purchased a good solid block at a reasonable price point. It does come with plenty of mounting hardware, however it does not come with a backplate and it also does not come with metal barbs. I found the long plastic barbs it comes with make tubing removal difficult and I ended up turning the barbs trying to remove tubing. But overall, the base is nice and sturdy and the pins are solid and big enough for easy cleaning with a brush. The water block design is fairly simple in that water comes in one side (Not directly centered) and crosses the block through the diamond pin matrix to the outlet.
The D-tek fuzion GFX is a nicer looking block and more complex. The metal barbs that come with the block are the good high flow type and the o-ring is recessed into the barbs themselves...very nice. The rest of the block consists of three parts....the copper base with round pins, the middle center nozzle block, and the top acrylic top. This redirects the water from the side inlet to more of a direct jet so the intake water impacts the center base and runs out toward the perimeter through the pins. This ensures the coldest water gets good direct coverage in the center of the base and is very similar to the CPU block water flow. The only construction element to watch is the smaller copper pins that could be damaged by cleaning if you're not very careful.
http://www.overclock.net/gallery/dat...mcw60-open.jpg
Here is a closeup of the bases side by side, both of these blocks have been used previously:
http://www.overclock.net/gallery/dat...0-baseOnly.jpg
And here is the bottom of both blocks, the GFX base that is stepped down and in contact is slightly larger than the MCW-60.
GFX = 30 x 30mm
MCW-60 = 27 x 27mm
The G80 GPU IHS is much larger than both of these, but I believe they cover the actual die under the IHS sufficiently, but I do like the fact that the GFX base is larger.
http://www.overclock.net/gallery/dat...ntactAreas.jpg
And this is what each block looks like installed on my card. In my setup I have some old Thermalright HR-03 plus memory sinks I've artic adhesived on there some time ago(Yes I've had many many coolers on this 88..:D). I do have to say the GFX is very nice looking and takes first in the coolness category..:D
Note: I used the same hardware and backplate to mount both blocks.
http://www.overclock.net/gallery/dat...talled-GFX.jpg
http://www.overclock.net/gallery/dat...lled-MCW60.jpg
PRESSURE DROP TESTING & FLOW RATE
I already knew the GFX was a fairly restrictive block, and I had a previous curve I ran on the MCW-60. But I figured I run a curve on the GFX and compare it to the published curve. I did go ahead as I've been doing recently and subtracted out the pressure drop of my testing equipment. What I found was something close, but slightly less restrictive than the D-Tek published curve, and this is how several blocks compare in flow rate.
http://www.overclock.net/gallery/dat...X-PDcurves.jpg
The GFX is a very restrictive block, and the MCW-60 is a very free flowing block. This just means if you are running multiple block loops, you need to ensure you have a strong enough pump if going with the GFX. This can typically be done with an pump equal to or stronger than the D5 or DDC series with top, I would only suggest that you should ensure you still maintain at least .9 to 1 GPM.
Here are some quick flow rate examples using each block:
GPU only Loop
-MCW60, D5Vario, 7' Tubing, T-line, MCR320 = 2.42 GPM
-Fuzion GFX, D5Vario, 7' Tubing, T-line, MCR320 = 1.56 GPM
CPU, NB, GPU loop
-MCW60, D-Tek Fuzion, MCW-30, D5Vario, 7' Tubing, T-line, MCR320 = 1.67 GPM
-Fuzion GFX, D-Tek Fuzion, MCW-30, D5Vario, 7' Tubing, T-line, MCR320 = 1.30 GPM
THERMAL TESTING
I'm still waiting for some final cure time on the artic silver 5 I used on the GFX, but I think it's settling down now. For my testing I used ATI tools to read GPU temps. I then used two digital thermometers with temperature probes to read the ambient air temperature into the radiator (MCR320 with Yate Loons), and one to measure the coolant temperature in my T-line. Note the coolant temperature is "AFTER" the blocks, so the actual water temperature entering the blocks would be very slightly lower.
Measuring all three allowed me to gather enough data to get an average delta for "Air to Water", and "Water to Core". With all of these I was then able to adjust for my variation in ambient temperatures to compare them against. This was my collected raw data:
http://www.overclock.net/gallery/dat...VsD-TekGFX.jpg
And this is the comparison chart using 21C as the target ambient temperature entering the radiator.
http://www.overclock.net/gallery/dat...ekGFXchart.jpg
CONCLUSION
In the end I was a bit suprised to see the GFX win in the thermal category, but it did by a couple of degrees under load. The MCW-60 is clearly the pressure drop winner, but in my case I have a GPU only loop so the GFX is the better choice.
In cases where someone has a CPU and GPU (SLI) setup, they would have to weigh out what is more important to them. The reduced flow rate of chosing the GFX blocks would reduce flow rate enough that CPU temps may be higher by as much as .5C depending on the block.
You would just need to decide if .5C higher CPU temps are worth 1-2C lower GPU temps.
PRODUCT SUPPORT
One other note I would like to make is regarding D-tek and their product support. During my testing process I had inadvertantly broken a copper pin in the GFX by cleaning with a brass brush. At the time I didn't think I was being too forcefull, so I contacted D-tek about the broken pin situation I had. They are sending me out a brand new base free of charge, and were extremely helpful in the entire process. I have to give D-Tek a :up: for product support, it's very refreshing to see that type of support.
BOTTOM LINE FOR MY SYSTEM
The MCW-60 has had a happy few months in my system and has done a nice job, but the D-Tek Fuzion GFX is doing a better job cooling, so it's taking over the G80 realestate...:D
Hope this is useful, I havn't done alot of thermal block testing yet, so go easy on me:)
If I continue to see gains by curing of this AS5, I'll repost a new table and chart later in the week. The MCW-60 was well cured, but the D-Tek only had a few days of cure time. I know for best results I would duplicate this effort over many mounts and remounts, but this was what I have currently.
