Very nice write up! :up:
Printable View
Very nice write up! :up:
One thing I'm going to do differently in my next CPU block test is use the same brand thermal probe for everything. In this test I used two different models and I'm seeing that there is some difference. My comparitive results are still good, but I wouldn't get too wrapped up in the air to water temperature, I think my deltas are actually too low for what's shown here. In the end we'd probably have a hard time comparing since we're using different thermometers, but as long as you're consistent between tests of a comparison the relative accuracy is still good.
And as far as radiator output...the higher the water/air delta the more heat the radiator can kick out. I heard on OCN the other day that HWlabs told some folks a 480GTX can consume 2200 watts, yet HWlabs conveniently left out what water delta that was. An MCR120 can to....might be a 50C delta, but I bet it can!...:D
Yup, you really have to slow the flow waaay low before any noticable temp differences. (Lots of stuff in that loop)
nice job! thx
Awesome Job man :D
was really helpful.
this was a very nice read... ty :)
Nice reading... :up:
btw,
Anyone tried upgrading the MCW60 with a base plate from an Apogee GT(X)?
I did.
Pins on the GT(X) base plate are some 1.5mm shorter so I figured this would be a problem as the empty space between the pins and the top would screw up the flow making less water going through the pin matrix. So I glued a thin (~1.5mm) rubber slab between the inlet and outlet. It should keep the water flowing through the pins.
And it seems MCW60 does benefit from the upgrade...
I can tell that atleast idle will no change much, if at all, but what's interesting is that according to Rivatuner it seems my stock 8800GTS stays at a comfy 38ºC - be it idle or load. Before the swap load bumped temps to 40ºC, now it's just 38ºC. Load generated with rthdribl.
Gonna have to try putting some real load other than the windowed rthdribl on the GPU and see if I get load temps atleast one degree higher than idle. :rolleyes:
Very nice review!
nice read.
I've heard the question and it would make sense to me the Apogee base would do a bit better because of the finer pins.
But yeah, it's hard to measure. Short of multiple measurements logging ambient, water, and core temp...it's hard to know what you're getting.
I'm working on some CPU waterblock testing now, and man it's alot of work.
I'll stop hijacking this thread and start another when I have some numbers with greater heatloads.
:)
Hey Martinm,
Congrats on getting the data out. I did testing almost identical to your setup (not yet published) and had almost the same results. Glad to see there are some like minded souls out there who enjoy testing.
Hats off to you,
Jay
I wonder if any comparison test on how 2 gpu block affects the cpu temp if gpu block mcw30 and fuzion are on the same loop?
Interesting review Martin... I have a GFX myself and was planning to mod it before installation - your review has got me thinking twice now.
Has anyone modded a FuZion GFX, and noticed any improvement in flow?
Would a single Swiftech MCP655 pump have enough push to run a D-Tek FuZion, MCW30 and a single D-Tek FuZion-GFX?
BTW nice work and a good read!
The biggest affect on CPU temp with GPU combo loops will be the added heat into the loop. The difference due to flow rate is minimal in comparison.
An example might be your flow rate was reduced from 1.5 to 1.0 gpm. If you were using the CPU fuzion, here is the published thermal performance difference:
http://lib.store.yahoo.net/lib/sidew...uziongraph.gifhttp://lib.store.yahoo.net/lib/sidew...uzionchart.gif
So 1.5gpm to 1.0 GPM The C/W difference is .003 @ 100 watts is .3C
Even if you dropped down from 1.5 to .8 GPM the C/W difference is .006 or .6C.
It's pretty hard to get below .8 GPM, so this is a good example to show you how some blocks do very well even at low flow rates like the Fuzion.
This doesn't account for some loss you get in your radiator, but as long as you have plenty of capacity there, that won't amount to a whole bunch either.
So combining GPU and CPU might raise your CPU temps by several degrees, but most of that is going to be due to the added heat which you'll get regardless of which block you choose.
Yeah, I agree, you'd have well over 1 GPM. Depending on the rad/tubing it could be around 1.3GPM which is plenty. It's pretty hard to bog down a DDC or MCP655 to the point where the Fuzion CPU block starts loosing much efficiency.
While more flow rate is always desireable, the difference between 1.0 GPM and higher is extremely small, so taking a hit on flow rate for better thermal capabilities is a worthwhile tradeoff.
Even radiator performance doesn't drop off much clear down to .5 GPM.
The PA120.3 for example may loose 5% from 1.5 down to 1.0, but it jumps to almost a 20% loss from 1.0 to .5 GPM.
http://www.thermochill.com/PATesting...lowrateGPM.jpg
So, just make sure you have plenty of radiator capacity and lower flow rates will function just fine.
Once again, you pwned the review boards :D
Great reivew!