This Thermalright XP-90 review took much longer than anticipated... but it's finally done, complete with new MP3 files and measurements of the Nexus 92 fan as well. If the XP-120 is the new king, then I suppose we could call the XP-90 the crown prince?

In the XP-120 review, you had the Zalman in the charts as a reference. Could you be persuaded to do the same here? Always nice to have a fixed reference. I would buy the XP anyhow, if not for the n3rd factor then because I just love the look of the pipes - they kinda remind me of performance headers

Thanks for the review!

rp -- have another look at the last page. Direct comparisons are difficult because of different fans / sizes... but all in all, when you consider the cooling performance at similar low noise levels (ie, airflow), the XP90 is clearly better than anything else we've tested thus far. Except for the XP120.

Will you do a review of SI-97, for the sck A? When? SI-97 vs SP-97...mmmmm

Have to ask Russ, who should have the sample by now. IMO, no contest. The SI-97 will stomp all previous socket A coolers.

In process.....just arrived yesterday.

First impression of the Si-97... boy it's pretty.

Nice to have the review to compare the XP90 to the other top quiet heatsinks. I am confused by one thing though. It isn't clear from looking at the review how the XP90 is superior to the Zalman (as stated in the text). The conclusion I would reach from the review is that the XP90 is roughly comparable to the Zalman in cooling power/noise at the one given comparison point.

Given the comment that the Zalman is comparable in noise to the Panaflo at 9v, a (rough) interpolation of the data for the XP90 suggests that the XP90 with a 9v Panaflo should be about .38 C/W (TDP).
(Just plotting the 3 points, draw a rough curve through them, take reading, being generous to the XP90). But at that noise level the Zalman scores slightly better at .35 C/W (TDP).

Certainly this analysis is very back of the envelope, but stare at it as much as I can, I haven't worked out in what way the XP90 is obviously superior to the Zalman. Is the advantage that you have room to go quieter with the XP90 that you don't have with the Zalman (assuming you can use the lower cooling performance). Or is this one of those "have to hear it to understand" kind of things. I am probably looking past something obvious.

Thnaks

scdr

You're right it is not that clear...

OK, so the Zalman at 5V is the same loudness as a Panaflo 80L at 9V. That's also about the same as a Nexus 92 at 12v. Slightly different sonic signature, but the SPL readings of all those fans is within 1 dBA of each other at 1m. (try listening to the MP3 files.)

So just compare the Z7000 @5v vs the XP-90 + Nexus92 @12V. The latter has a C/W of 0.33...

hmm... less of an advantage than I thought. Maybe that Z7000 needs to be tested on this system again??

si-97! *blank stare*
WELL i was really confused for a while there. for whatever reason, i'd somehow got it into my idiot head that xp-90 was a socket A cooler.
now i'm wondering why the xp-90 even exists. is the only reason to try and snatch up remaining market remaining from mobo compatibility problems of xp-120? it can't have been a 'low-cost alternative', given the similar mass i can't imagine they'd cost significantly less to manufacture (ok 1 less pipe..probably more than offset by development costs to machine new parts)

this si-97 looks pretty cool. they must have discovered the aluminium fins conducting base to top weren't really doing much at all (negligible compared to phase-change action). i suppose nobody could dare doubt the effectiveness of those thermalright heatpipes now.



i am happy to have the bias mentioned, even if a fanswap wasn't done in the end.

a very minor gripe but i think spcr reviews would be better looking if the pictures were more consistent. (i don't mean you should go change it of course, but to keep in mind in future). i mean there are some with black borders and some with fade-out edges, some stamped with a logo, others aren't... some of the pics are all dolled up with background greyscaled or washed out - these photoshop gimmicks are not what makes for good presentation, but rather simplicity and consistency like the rest of the site.

ps: it is a Very Good Thing (tm) to have the ambient recording available, thank you!

That last comparison in the review was too rushed; made some changes to try and put all the HS with fans at similar loudness. Mind you, because they're different fans AND different sizes, it's not really a good apples-to-apples, but.....

Another interesting review mike - one niggle: Why does the link from the front page of SPCR take you to page 3 of the review?

MikeC why did you edit out me pointing out the 0.33 instead of just posting another reply? your site = your right, i spose, but having my posts (unnecessarily) edited without any notification makes me uncomfortable.

a question about these airflow measurements, is there a resource anywhere on the site or the forums about how MikeC and/or the SPCR lab slaves measure fan cfms? if so please link me, i'm interested in the details of this process.

Rusty075: if the SI-97 testing is not yet finished, would you please include a Z7000 fanswap with nexus 92mm (or whatever was the quietest performer) in the comparison tables of the review? pretty, pretty please? i know we argued about this earlier, and you slagged off the idea. but it's a really easy mod, simpler than some things which have been tried in other reviews, and since this review will be a decider for what's the top socket A cooler i think it's important to try a bit harder. minor bribe: if it's in the review, i'll make a donation to SPCR



hm it's a tad better now. although in that last table it's awfully close (~2 degrees) between the Z7000 and the XP-90, and i can't help wondering how the low-flow numbers would compare with the nexus 92mm, allowing some more room for the zalman to go quieter (keeping in mind Z7000 reputation for faring very well under very low-flow)
you said elsewhere that this moves into the realm of modding and ceases to be a review. i say 'so what', this is silent pc review not quiet pc review and it's impossible to build a silent pc without a little bit of modding. it's no disservice to thermalright, and your readers aren't scared of a little bit of modding.

*braces himself for some backlash..*

This makes me wonder why they kept the orentiation of the fins instead of turning them 90 deg. like a tower design to make it easier to duct / use "passivly" ?

Rusty075, will you include in the review of SI97, the results with an SP97 and a low-airflow fan (like nexus)? For what i know, aluminium and low airflow are a better mix then cooper and low airflow...But i need to be sure, i just bought the SP97....
When will the review see the light of the day? I can't wait....

It is not necessary to remove the motherboard to put the XP's on a K8. With the board horizontal, remove the screws and bracket, and the back plate will drop a 1/4" or so. Thead one of the original (or longer 6-32) screws into one back plate hole then pull the plate up, you can then insert one of the thermaltake screws through the new retaining plate and screw it lightly into the remaining hole. Remove the original screw and line up the back plate, then tighten both new screw as needed.

wim: I'd be happy to. PM/Email me and I'll give you the address to send the Z7000 to.

DG: We'll probably reference the SP-97 review but there won't be any new testing done on it. (since I don't own one to test. ) You'll be able to compare the °C/W numbers well enough.

And there is nothing about aluminum that makes it superior to copper for heat transfer, in any airflow. It's advantages are lower weight, lower cost, and easier machining...that's it. That myth rears its ugly head every once in a while.

I'm awaiting the arrival of the Nexus 92mm that was used in the XP-90 review, before testing can begin. I'll keep you posted

Not really....IIRC, aluminium disipates heat much better than cooper, but absorbes it much slower than cooper; that's why the base of the XP90 it's made from cooper, and the fins are made from aluminium. And beacause it disipates heat better than cooper, aluminum i think will have better results with a low-airflow fan, than cooper.

I don't want to take this too far OT, but...

Nope. Completely, totally, nope.

Here's an excellent article on general heat transfer principles: Dave Smith's General Heat Transfer Guide @amdmb.com It's a very interesting read, and I highly recommend it to everyone.

The section that applies to this topic is on page 4.

Two points:

1) You say you have no idea who manufactures the Nexus 92. While the 80mm and 120mm Nexus look like Yate Loon products, the model numbering scheme on the 92mm Nexus is inline with that of Dynaeon. Unfortunately Dynaeon site is down as I write this so I cannot chech.

2) One thing I noted while trying the Nexus 92 on the SP97 was that unlike other fans, placing the Nexus down on the HS did not create much turbulance/resistance noise. Is this the case with the XP90 too?

you don't have a Z7000..?? ack. maybe you could get it sent over along with that nexus fan


OH NO! the Al vs. Cu argument again .. please spare me

By the looks of the XP-90, when you put a 92mm fan on it, there are gaps at the side of the fan, but an 80mm fan would be flush with the fins.

It seems counter-intuitive to me that they designed it with these gaps, and even more so the gaps present in the XP-120. Surely this will mean a fraction of the fan's airflow is wasted.

I don't particularly expect anyone to go to the bother, but I'd be interested to know if a bit of tape along the sides improved the performance a degree or two, particularly with the XP-120.

Another counter-intuitive feature to me, is the placement of the heatpipes at the very edge. I'd have thought distributing all 4 more evenly across the face of the fan would be better, something like: ---o----o------o----o--- Who knows? (Well, Thermalright I assume)

Not one that I have access to anymore. (we don't just keep these things around forever, gathering dust )

For the price of shipping a Z7000 internationally, you could just buy a Nexus 92mm, and do your own testing with your Z7000. Plus you'd get an excellent fan to add to your collection.

There's no magic to having one of us "official" SPCR people do the comparison, the results would be just as interesting if you did them. Afterall, this is a community of silent tinkerers, right?

Nice article as usual, guys.

One strange question though. Why is Edward Ng's name linked to an email whereas Mike Chin and Ralf Hutter's are not?

I think the pictures and most of the text is Ed's. Like the "smooth like butter" part...

Cheers,

Jan

Jan Kivar, my...thank you for noticing! I'm truly touched.

This Thermalright XP-90 review is great. And just in time as I want to use this heatpipe in my next build. However we need more data. Because of the heatpipe construction and operating theory, It is capable of removing its maximum ammount of heat when oriented on a horizontal motherboard. How many of us out here are building computers in a case with a horizontal motherboard? I would guess the vast majority of builds and upgrades are using a computer case with a vertically mounted motherboard. Hence, we need more data. A heatpipe of any given design will remove different ammounts of heat depending on what plane and rotation it is mounted. The Thermalright web site warns against mounting their heatpipes on a vertical motherboard with the loop in the pipe oriented down. We need some real world data on the XP-120, and XP-90 mounted on a vertical motherboard, and oriented with the loop of the heatpipe up, right, left, and if you dare just for the sake of science, down. I am guessing that if you create a data table with all these orientations and include the data from your horizontal motherboard test bed, things are going to get real interesting.

This should have been remarked on. A quick answer:

Ralf Hutter did some post-review testing of vertical vs. horizontal orientation of the XP-120 HS and found no difference in performance. See the XP120 review article discussion thread in this forum...

This sort of testing was done in the XP120 review, it was found to make no real difference. I would not worry about this at all. I bet even in the worst orientation the HS would perform within a few percent of the best.

A well designed heatpipe with a strong capillary action, such as scintered tube designs which is likely what thermalright are using, is as effective to within a few percent in any orientation. Some hetpipes are little more than an empty tube with a fluid, and these do indeed suck at transferring heat downwards but the heatpipes used in all good heatsinks are much better than this and don't care about orientation.

I think it would be good for a piece on the operation of heatpipes to go up or at least be included in the review of a hetpipe cooler, as it is logical to think that a heatpipe will not work in the wrong orientation however this is wrong for most heatpipes used today.

I respectfully disagree with Dave Smith's conclusion that aluminum finned CPU coolers can't/don't dissipate heat through convection faster than copper fins. Dave's excellent article dispels a lot of false beliefs but it also illustrates precisely why aluminum finned CPU heatsinks can in fact deliver very minor benefits over copper fins. Part of the answer is in Dave's comment here in regards to convection:

"Many variables affect the value of “h”. Among the most important are the fluid and surface temperatures, the fluid velocity, the fluid density, the fluid viscosity, the fluid specific heat, and the geometry of the surface. While standard engineering equations allow one to calculate convection coefficients with reasonable accuracy for simple shapes and flows, it is not possible to adequately determine the airflow through a heat sink analytically to predict performance with high accuracy. This falls under the realms of laboratory trials and a software tool called Computational Fluid Dynamics (CFD)."

In addition to the above variables, a PC and the CPU do not operate in a 100% steady-state mode because of the constant change in CPU thermal load, yet the cooling fan speed is relatively constant unless of course you are using a temp controlled fan. So in practice the results are slightly different than if you input a fixed thermal load and measure the steady state results when things "stabilize". The CPU dynamic thermal load is in constant flux as is case ambient air temps in many instances depending on case airflow volume and thermal load. Measuring the instantaneous CPU die temp with an embedded thermal couple and calibrated circuit would be perhaps more valid than the relatively slow response Mobo thermister readings, but in practice the end result, i.e. good cooling from a quality heatsink design, is really the objective.

Rather than argue the minor details of surface area, finish, air velocity, and all the other parameters that affect the convection coefficient, it's easier and just as valid to measure the dynamic real world results. In a controlled experiment with identical heatsinks other than one with copper fins and one with aluminum fins on a copper base, operating inside an actual PC, it is my professional opinion you will see a minor, though measurable increase in heat dissipation with the aluminum finned heatsink IF the fins are attached properly so the thermal resistance is as low as possible and the case airflow is sufficient to maintain a consistent temperature delta. For all practical purposes, one accurate scientific test is worth more than a thousand opinions, mine included. I don't have the means to perform this test but it would be interesting and perhaps enlightening or not ???

I would offer that it's cheaper and easier to produce a quality one-piece copper heatsink than a copper base/aluminum finned heatsink of equal quality so I doubt the Thermalright XP-90/120 design is intended to reduce costs, but instead is intended to improve dynamic cooling. I suspect Thermalright's proprietary engineering data would support the premise that the aluminum finned copper base heatsink w/heatpipes performs better than the solid copper heatsink w/heatpipes by a small margin in actual application.

I think the outstanding performance of the XP-90/120 speak for themselves and I'm not interested in a "debate" on the subject. I just wanted to offer my perspective on the "variables" that impact real world results as these variables are difficult to account for in theoretical engineering.

Hat's off to Dave Smith for eliminating a lot of the misconceptions and to SPCR for an excellent XP-90 performance review. Now you know why I chose the XP-90 for my latest PC build !

does this HS work with socket 775 (the new one)?