My Thermalright Ultra-120 eXtreme Review

[Navig]

Phase one of my Thermalright Ultra-120 eXtreme review is completed. It involves a lot of fans and a lot of configurations:

Yate Loon (YL), Delta 120x25 M (D25), Panaflo M1 (P), Sanyo Denki 120x38 (SD), and a Delta 120x38mm VHE (D38).

Push, Pull, Push-Pull, 7v, 12v.


Link is here.



Here are my conclusions:



Conclusion: bigger fans do give better temps, but expect diminishing returns with the super big fans.





Conclusion:
Best fan configuration: Push-Pull
Second place: Pull
Last place: Push




Phase Two: Lapped Performance
Phase Three: Comparison to the Tuniq Tower and the Big Typhoon.

[Ender17]

I think those graphs should be redone with a starting value of 0.
The current way exaggerates the differences - the first one implies the D38 temp is less than half of the YL temp, when it's actually only about 7%.

[Andrew LB]

^^^ Ya. It would seem the OP of this thread may be in the field of advertising. haha! j/k

[crazyea]

Wouldn't it have been easier and/or make more sense to mount the extra fan with an elastic or twist-tie than butcher the heatsink like that?

[lad]

What's with all the vitriol tonight?

[Navig]

Hmm, such insightful comments.

If you're unhappy with my graphs, the numbers are all there. Have at it and post back with your "improvements".

The whole point of the graphs is to emphasize the difference! The actual values are on the bars themself, so yes, any idiot can look and see we're only talking about a few degrees difference. If that isn't enough, I even mention it in the text. Biggest case scenario between worst configuration and best configuration was 9c.

And uhm, I'm at a lost as to what exactly I might be advertising for? Just because you throw down a "j/k" doesn't make it useful post. Thanks for the bump? I guess?

Finally, I'm sorry I'm being criticized for hard mounting my fans? I changed the fan configuration what some 20 times during the testing, I wasn't inclined to having to redo straps each time. Not only that, there really isn't convenient method for strapping a fan. If you call that butchering, you should take a closer look to what I've done to the motherboard.

[crazyea]

At first I couldn't tell if the temps were in F or C... The load temps seem a bit on the high side, especially considering the test set up is out in the open. I do appreciate the time you put into it. Thanks.

[Navig]

Trust me, I've no shortage of vitriol meself!

The temps are in Celsius, I am thundering 1.535v thru it. Someone on my ocforums thread mentioned that coretemp had troubles with allandales, so I'll have to peek in on that. I might re-generate the graphs with abit uguru core temps, at a glance, tho they seemed to parallel the coretemp temps appropriately.

[Andrew LB]

And uhm, I'm at a lost as to what exactly I might be advertising for? Just because you throw down a "j/k" doesn't make it useful post. Thanks for the bump? I guess?

Relax man. I'm just poking a little fun because the graphs (while accurate and I do thank you for the time you took to make this review) remind me of how graphics card manufacturers tend to use methods like that to show during paper launches how much faster their product by using little things to make the differences look larger than they are.

I hope you understand that point. If not.... just regard this post as another bump.

[Speederlander]

Something doesn't seem right....

With this CPU (E4300 at 3.285ghz (9x365) vcore 1.535v.), you should not have load temps in the 70's with the "ultimate heatsink".

It's a dual core, not a quad and the volts are up but not ridiculous. Plus you said you were measuring at 3 minutes in, not even time to stabilize.

I mean, with my enzo and delta on my overclocked QX6700, my prime load temps don't go over 69C or so.

Maybe your heatsink was hardstoping on the mobo CPU frame? Or am I missing something? If those are the load temps we can expect, then this heatsink is a disappointment...which I doubt.

I am thundering 1.535v thru it.

Temps are way too high for 1.535v and that chip with this heatsink IMO. I would expect you to be in the mid 50's or better to be honest.

Note this review:
http://www.anandtech.com/casecooling...spx?i=2943&p=5

[Navig]

Yup, thanks for the comments. A few other folks have noted that Core temp inflates the values for Allendales.

If you look at my uguru temps, they are much closer to what you'd expect.

I do still think my conclusions are valid because they are all based on comparisons from the same baseline--if I replaced the absolute values on the graphs with "deltas" such as +1.2c or what have you, they would still all look the same.

Now I do believe my baseline is valid because I clocked this chip up from its box, for the expressed purpose of doing this review, and all the temps (I followed in uguru) followed voltage increases appropriately--when I was reaching stable load temps in the high 50s, is when I deemed I had a system hot enough to separate these top tier heatsinks. This was verified by several different AS5 applications, and even already several different heatsinks. I opted to report coretemp temps because I figured people wouldn't like using proprietary bios software to show temps, harkening back to the ol' IC7 verus P4C800 temp debate.

Anyway, I'll go back in and crunch the numbers for uguru, and see if they track similarly.

[Navig]

Alright, I added a little addendum at the thread.

[cozwin]

still a good benchy so gj.

[Ender17]

The whole point of the graphs is to emphasize the difference!

Wrong. Graphs are supposed to be a representation of the data.

[davexl]

Wrong. Graphs are supposed to be a representation of the data.

He labelled the y axis with a scale, so what's the problem?

The man is giving us the product of some hard work and should be thanked.

OP - did you ever test push-push vs push-pull?

[Eagleclaw]

edit: davexl, opps read it wrong, I didn't see the push push my bad, interesting idea though

OP: Thanks for the work you did on this....

He labelled the y axis with a scale, so what's the problem?

The man is giving us the product of some hard work and should be thanked.

OP - did you ever test push-push vs push-pull?

[P§I]

Thanks for the hard work, seems some people have to nit pick over stuff. Been wanting to get this HS and the more reviews and test i see the better.
Thanks

[Navig]

Wrong. Graphs are supposed to be a representation of the data.

I should have re-phrased that. MY graphs are meant to emphasize the difference. All visual representations introduce a spin on the presentation of the raw numbers--right down to the selection of colors (just ask any graphic artist). Anyone who says otherwise is probably trying to sell you something. Again, here was my logic in the selection of the y-axis. If you actually graph out the entire scaled y-axis (which, by the way I've noticed you have STILL NOT DONE), yes you can't tell the difference at all among the different bar heights. Well, then what is the point of including such a graph at all--its just wasting bandwith.

If you're not interested in sub 5c differences, don't bother with my review.

I have further news for you, you probably shouldn't bother with the rest of the review. Lapping the heatsinks is only going to yield a few degrees of benefit. Even the separation from the top tier heatsinks is not going to be more than 5-10c. I'd question the vailidity of any review that finds >10c differences among them.

If you want >10c differences, well then you can compare the heatsink to the stock Intel sink. But why bother? Rest assured, this heatsink is superior to the stock Intel sink.

But this is xtremesys. the last time I've even touched a standard cpu heatsink was to mount it to my northbridge.

[CPLB]

Nice job Navig, your reviews are always very complete and professionally done. Looking forward to the comparisons.

[Navig]

Update: Phase 2 ~ Lapped vs Unlapped

Full thread here.

Summary tables:

Load temps, Coretemp






Load temps, uguru




Summary: Stock vs Lapped

Through 3 different configurations, the Lapped heatsink beat the Unlapped heatsink by 1.7c (Coretemp) or 0.7c (uguru). I’d call this a consistent but small difference


Okay, finally on to my comparison (necessary steps as the other heatsinks are all lapped).

[Okda]

hope in ur comparison we will see the T120 and Ultra-X

[joshd]

Something doesn't seem right....

With this CPU (E4300 at 3.285ghz (9x365) vcore 1.535v.), you should not have load temps in the 70's with the "ultimate heatsink".

It's a dual core, not a quad and the volts are up but not ridiculous. Plus you said you were measuring at 3 minutes in, not even time to stabilize.

I mean, with my enzo and delta on my overclocked QX6700, my prime load temps don't go over 69C or so.

Maybe your heatsink was hardstoping on the mobo CPU frame? Or am I missing something? If those are the load temps we can expect, then this heatsink is a disappointment...which I doubt.


Temps are way too high for 1.535v and that chip with this heatsink IMO. I would expect you to be in the mid 50's or better to be honest.

Note this review:
http://www.anandtech.com/casecooling...spx?i=2943&p=5

It is all to do with coretemp. 0.95 for some reason, uses a 100*C tjunction, whereas every other c2d uses 85*C. The CPU reports the difference between the actual temp and the tjuntion temp (the temp at which it shuts off), and temp is given by temp = tjunction - difference. Therefore, if the 100*C tjunction is indeed 15*C too high, then the temps coretemp reads will also be 15*C too high.

I am confused as to what to belive at the moment, the newest Speedfan gives temps 15*C lower than CT 0.95 (as it seems to use an 85*C tjunc.)

My E4300, clocked at 3.2GHz, 1.425v apparently loads at 60*C on a tuniq tower, which I am somewhat skeptical about. 45*C seems alot more reasonable.

At the moment, I am going with coretemp to be on the safe side, but am awaiting updates to coretemp to see what happens.

[BobHarris]

Thanks for the review

Do you think it will fit on the new P35 boards with the huge heatpipes on the northbridge?

Is it possible to mount the plastic backing without taking out the motherboard?

[RealTelstar]

Please try a push-push fan setup.

[Okda]

push-push is not a good idea at all