"Normal" user testing: MCR-220 Stack

[Zehnsucht]

Ultra Kazes Done, next up Yate Loon

Hardware:

D-Tek Fuzion v2 no nozzles + E6600 1.5V 3.6 GHz
MCW-60 + HD4870
DDC 3.2 + Petras Top
EK 1/2” Fittings, 7/16” Tygon tubing.
EK RES 150
Petras Yate Loon M / Scythe Ultra Kaze 2000

Programs used for heating is FurMark and OCCP running in the background. Power draw from the socket is 373W, which translated using the curve of HX520 @ 240V approximates 320W actual system power consumption.

Testing methodology
I do not have any fancy temperature probes or other gear, but I'll KISS.
The testing will be divided into two parts; the actual testing of radiator performance on three different voltages (5, 7 and 12V), and secondly I'll investigate how RPM changes from open air to single and stacked at these three voltages (if any, I'm writing this 14/3, while I expect to produce results 21/3).
The Delta temperatures in the graph is the average CPU temperature minus air intake temperature reading during test. I've calibrated my temperature sensors with the Real Temperature calibration test, and none of the cores seems to be stuck and they also show the same temperature throughout the temperature range.
Pics of radiators:
http://img.techpowerup.org/090317/IMG_0010818.jpg

During single radiator testing the fans will be in push. In "stack" configuration, I'm using them as intended by Swiftech, fan in middle and water going in parallel between radiators.

Results
The temperature stabilized at around 40-45 minutes of testing, and the temps are substracted with the air in temperature.

(more heat load, more interesting results.. Ohh exciting :) )




Conclusions

Ultra Kazes are way overkill for an OC:d E6600.

[Hondacity]

sleep long...

good stuff btw...welcome to the testing world :D

[Sadasius]

What's 'stacked' defined for? Radiators or fans in push/pull or other?

***EDIT*** Never mind....re-read methodology.

[Hondacity]

What's 'stacked' defined for? Radiators or fans in push/pull or other?

look at the picture:D

[Vapor]

Yate Loon SM at 5V beating UK2000 at 12V? Golden :up:

Also considering the SMs run at like...720RPM at 5V (from my memory....), and they're showing a noticeable improvement, that says that maybe the Nanoxia fans used in HES's testing just weren't up to snuff or the radiators he used were no good for stacking (considering how much of a downgrade occurred).

[Zehnsucht]

Yate Loon SM at 5V beating UK2000 at 12V? Golden :up:

Also considering the SMs run at like...720RPM at 5V (from my memory....), and they're showing a noticeable improvement, that says that maybe the Nanoxia fans used in HES's testing just weren't up to snuff (considering how much of a downgrade occurred).

Hmm you're right. I checked the data, and yes, stacked with YL@5V beats single UK @ 12V. That is really strange. I'll redo the testing on that one to see if I come to the same results.

One thing I should mention is that the UK are brand new, while the YL has been in use for almost a year now. I think you came to some conclusion that fans improved after a little while?

[Vapor]

A 2000RPM UK might run at like 1900RPM when it's new....a YL at 5V is wayyy slower than that regardless :p:

A longer load may prove to equilibrate a bit more :shrug:

EDIT: the more I look at it, there's something else at work here. Going from 5V to 12V (basically doubling cooling power for each setup) is only showing a few degrees of improvement. Adding a stacked rad shouldn't out-do THAT. :confused:

[faster3200]

Hmm you're right. I checked the data, and yes, stacked with YL@5V beats single UK @ 12V. That is really strange. I'll redo the testing on that one to see if I come to the same results.

One thing I should mention is that the UK are brand new, while the YL has been in use for almost a year now. I think you came to some conclusion that fans improved after a little while?

Well, single vs. single the UK wins so I think your results make sense. It just shows that stacked owns single and that HES's tests used too weak fans.

[Zehnsucht]

EDIT: the more I look at it, there's something else at work here. Going from 5V to 12V (basically doubling cooling power for each setup) is only showing a few degrees of improvement. Adding a stacked rad shouldn't out-do THAT. :confused:

Yeah, but remember I'm measuring CPU temperature, and not water temperature. Other factors such as mount play a role here. I'm deleting the graph for now, and will return with more results tomorrow.

[Vapor]

Yeah, I know you're doing CPU temp....but you still have ambients to deal with. Going from 5V to 12V should roughly bring your water temps twice as close to ambient. For reference, adding a 2nd radiator alongside the first should have the same effect. Stacking a radiator should be noticeably less effective than that. :-/

Water temps scale linearly CPU temps in this case :)

Will be interesting to see what's causing the CPU temps to plummet with the stack :p: Investigational testing is pretty fun, IMO. :D

[NaeKuh]

Ye Investigational testing is pretty fun, IMO. :D

only if your setup is tweek friendly. :rofl:

[Zehnsucht]

Seriously, I'm getting some conflicting results here. I think I'll buy a water temperature sensor to see what's really going on.
And yes, I'm still not impressed by the Ultra Kazes.
What programs do you recommend for stressing the CPU? I'm using Orthos now, but are there better alternatives?

[miwhite]

Seriously, I'm getting some conflicting results here. I think I'll buy a water temperature sensor to see what's really going on.
And yes, I'm still not impressed by the Ultra Kazes.
What programs do you recommend for stressing the CPU? I'm using Orthos now, but are there better alternatives?

Intel burn test will take your CPU to its knees within seconds.
http://www.xtremesystems.org/forums/showthread.php?t=197835

[Vapor]

I prefer OCCT over IBT. More of a static load which is better for temp logging and this kind of testing. :)

[Zehnsucht]

Great, I have OCCT. It has a nice graph-logging feature as well. Thanks for the input.

[Taylor01]

This is great with all this real world testing. I am still on the fence here whether to go stacked mcr320's or just the feser 360. I want low speed fans and I don't have room for push&pull setup. So thanks I am really waiting on some hard data here!

[skinnee]

Hows it coming? If you have any questions or what not, feel free to hit me up. :up:

[Waterlogged]

Zehn, I think the hub size on the UK is the reason for your conflicting data. As a pull fan they'll work great, as a push fan without a shroud, they're gonna suffer hard.

[Zehnsucht]

Yesterdays results were all crap, since the I power cycled during post and bios was reset, so no overclock. This is also the reason why the first graph I posted showed huge gains with stacked configuration. I'll make a point and have CPU-z running when I take the screenshots (I'm logging RPM as well, but SpeedFan doesn't seem to be able to output the results in a text format).

I should have some initial results going this evening since went up a lot earlier this morning just to do some testing for you guys ;) Nah not really, I'm as interested as you are how this works out.

WL: Yes, I think you are correct.

[HESmelaugh]

Thanks for testing Zehnsucht. I'm really looking forward to seeing your results. And good job finding the issues with the first test and redoing them. :up:

[Zehnsucht]

Well, there you go, Ultra Kazes are done. And done. And done...
Results are consistent now, and I'm confident on my testing routine.
I've made this three times now, and get same results; minor improvement on 5V, and even less at 12V. I don't think that an OC:d E6600 is enough to put a strain on this radiator config with such high performance fans.
Sound levels? At 5V in stack they are more silent than on the single. 12V is unbearable in either configuration.
When I'm done with the YL, I'll add my HD4870 in the loop and crank up the voltage on the CPU to spice it up a little :)

[NaeKuh]

I don't think that an OC:d E6600 is enough to put a strain on this radiator config with such high performance fans.

at most your little puppy puts out around 100-150W.

Pifft, an MCR120 would be enough on that little processor.

[Zehnsucht]

at most your little puppy puts out around 100-150W.

Pifft, an MCR120 would be enough on that little processor.

Correct. Not everyone can afford the latest and greatest :(
But I think the temperatures will be different with a 4870 in the loop.

[Vapor]

Increase the voltage to like 1.65V and retest :p:

[HESmelaugh]

Interesting.
I agree with your conclusion that the ulta's are overkill here. But what's an important find in my opinion is that your stacked setup is not worse than the solo radiator setup. That confirms to some extent what I could only hint at in my testing. :up:

[Zehnsucht]

Increase the voltage to like 1.65V and retest :p:

That's not a bad idea. What voltage you reckon is the safest maximum I can do?

[Taylor01]

Yeah as I see it. it doesn't look like the stacked are much of a improvement over the single rad. What maybe 2-3c?

[Vapor]

That's not a bad idea. What voltage you reckon is the safest maximum I can do?E6600? 1.7V+ :) :p:

[Zehnsucht]

Yeah as I see it. it doesn't look like the stacked are much of a improvement over the single rad. What maybe 2-3c?

I think it depends on the heat load actually. As seen on my temperatures I don't have much gain going from 5V to 12V in a single rad configuration. That's why I'm adding the GPU tomorrow. That's a 100-150W more.
I'm doing an all nighter to get the YL done, so that I can do the added heat tomorrow.
I realize that it's kind of pointless to do the YL as we have established that the heat load is not enough, but I still want to do it for the completeness sake of this first initial testing.

[Sadasius]

E6600? 1.7V+ :) :p:

:nuts:

[Vapor]

I've been doing OCCT at 3.6GHz and 1.67V recently...on a scorcher of a B1 Conroe. Staying below TJMax though.

I figure 1.7+ is okay for people whose chips aren't high-leakage :p:

[gillbot]

I've been doing OCCT at 3.6GHz and 1.67V recently...on a scorcher of a B1 Conroe. Staying below TJMax though.

I figure 1.7+ is okay for people whose chips aren't high-leakage :p:

My old E6550 took 1.65v no problem for months so you should be fine in that range on a E6600.

[Hondacity]

my e6750 died at 1.65v

my e7200 died at 1.7v

good luck :D

[Sadasius]

Exactly .... It's a shot in the dark. It can tank or it can live, but just to load a loop? Has to be a better way.

[Waterlogged]

Zehn, testing looks good so far. :up: Can't wait to see how well it works with a higher heat load. :yepp: