ZEROtherm Nirvana CPU Cooler

Want to talk about one of the articles in SPCR? Here's the forum for you.
RoGuE
Posts: 540
Joined: Mon Jul 06, 2009 9:11 am
Location: Massachusetts, USA
Contact:

Post by RoGuE » Fri Oct 30, 2009 10:29 pm

I couldn't agree more.

dev
Posts: 43
Joined: Sun May 10, 2009 12:47 am
Location: Here

Post by dev » Sat Oct 31, 2009 12:44 am

Lets see if I can make myself clearer: extrapolation doesn't equal real life testing. Extrapolation is a speculation based on previous data and a pseudo-linear behavior.

Thats the first and worst mistake when testing heatsinks: assuming a given dB value for different fans will result in the same output and air pressure. Well it doesn't. I have no doubt that the Nirvana fan is worse than the Nexus you are using at the same dB value.

The whole point here is why did you guys fell the need to publish the last chart when the data is not real and certainly not a result of testing.

I saw someone bringing up synthetic platforms. Thats another huge mistake because it assumes all CPUs are the same and they generate heat uniformly across the IHS.

In the end if you want to be relevant for a handful of people than extrapolation might be just fine.

reddyuday
Patron of SPCR
Posts: 169
Joined: Sat Jul 25, 2009 12:47 pm
Location: Birmingham, UK

Post by reddyuday » Sat Oct 31, 2009 4:03 am

MikeC wrote: Me neither. The only thing that might explain it is that our results conflict with those other results, which somehow threatens them? I dunno.
No, I won't try to explain their ranting. But in the midst of all that ranting, there is perhaps one useful point that has been missed, viz., that the figures given for Nirvana were using the stock fan but they were compared with figures for other coolers using a high-quality reference fan. The table on the last page is very misleading because it puts Nirvana far below all the other coolers just because its stock fan is far worse than the reference fan. The table doesn't say all that much about Nirvana, the cooler. For a user, the comparison cited by Burebista is actually more useful because it is comparing different coolers with the same or similar fans, and it becomes clear that Nirvana, the cooler, is not all that bad after all.

You ask why would one bother to buy a cooler with a bad fan and swap it? The users might have their own reasons for doing so. The cooler might have a better fit with their case or motherboard, or they might have gotten a good deal on it, or whatever. Swapping fans on coolers and PSUs is a pretty common thing among the SPCR community. We are tinkerers who like to mix and match the best components to achieve a good combination.

I think this problem is easily solvable. If you can create your comparison tables normalized to constant CFM figures instead of constant dBA figures, then you can factor out the effect of the fans in the comparison. I have been doing this myself in my own tables because I couldn't get the data with the reference fan for all the coolers I was considering.

Uday

burebista
Posts: 402
Joined: Fri Sep 02, 2005 12:05 am
Location: Romania

Post by burebista » Sat Oct 31, 2009 4:28 am

reddyuday wrote:For a user, the comparison cited by Burebista is actually more useful because it is comparing different coolers with the same or similar fans
Same fan, same RPM, same CPU, same frequency, same Vcore, same ambient, same TIM and 23 different coolers.
reddyuday wrote:and it becomes clear that Nirvana, the cooler, is not all that bad after all.
Thank you, that's exactly my point based on testing in same conditions not on extrapolating.

MikeC
Site Admin
Posts: 12285
Joined: Sun Aug 11, 2002 3:26 pm
Location: Vancouver, BC, Canada
Contact:

Post by MikeC » Sat Oct 31, 2009 8:00 am

dev wrote:Lets see if I can make myself clearer: extrapolation doesn't equal real life testing. Extrapolation is a speculation based on previous data and a pseudo-linear behavior.
I was ready to throw you out of here when I read this line... but reddyuday's post made me go back to the introductory text on that last page. Now I could see what the confusion was about -- the language Larry used could be misinterpreted to mean that we guessed at what the Nirvana would do with the reference fan. That is not what was done, and the language has been corrected. But that doesn't forgive dev and others here for not reading my posts carefully -- if you had really read my last post, you would not have been confused about what we extrapolated.

Here's the core adjusted text from http://www.silentpcreview.com/article988-page6.html which you should go back and re-read before commenting here again:
Since the Nirvana fan is not easily replaceable, it cannot be easily compared to other 120mm fan heatsinks where our reference quiet fan was used. So we compared it with its own stock fan set to the same SPL levels as the reference fan. This is a patently unfair comparison, but it shows precisely how bad the stock Nirvana package is for quiet cooling.

The Nirvana has the look of a top-performer, but its fan is a loser. When we compare cooling results with its stock fan set to noise levels equivalent to those of our reference fan, the Nirvana trails the Thermolab Baram by only 3°C at 16 dBA. However when airflow is reduced further, its performance tanks, with the difference widening to double-digits. The main culprit is that stock custom fan.
The Nirvana's own fan was used in the comparison table on the last page, while all the other heatsinks used the quiet reference fan. You can charge that this is unfair, but there was no significant extrapolation; ie, we did not guess what the Nirvana would do with the reference fan -- rather, we compared the Nirvana's performance with its own fan set to the same SPL as the reference fan on the other heatsinks.

Was this unfair? Sure. It shows just how lousy Nirvana's package is -- due mostly to the fan and its mounting design.

Should we have modded the Nirvana and used the reference fan for that comparison? This is one of the points debated here all along. But you will see that in all previous coolers reviews, we simply did not make any significant mods. It's been policy for years. This would be a significant mod.
reddyuday wrote:You ask why would one bother to buy a cooler with a bad fan and swap it? The users might have their own reasons for doing so. The cooler might have a better fit with their case or motherboard, or they might have gotten a good deal on it, or whatever. Swapping fans on coolers and PSUs is a pretty common thing among the SPCR community. We are tinkerers who like to mix and match the best components to achieve a good combination.
Sure, and we've posted articles in the past about doing fan swaps on Zalman heatsinks, for example. But not as part of the review, and at a time when Zalman heatsinks were some of the best coolers around. While HS modding was mandatory if you wanted a silent computer >7 years ago when SPCR first came on line, today, the market is full of high performance coolers that are extremely quiet straight out of the box. Just look at that comparison list of coolers on the last page of the article. Straight out of the box, they offer better performance at low noise, many at prices lower than the Nirvana. If you don't think that's a competitive disadvantage we should report on.... :roll:
reddyuday wrote:If you can create your comparison tables normalized to constant CFM figures
Not so easy. Accurate CFM is extremely difficult to come by -- and in the case of the stock fan on the Nirvana, just about impossible to measure.

reddyuday
Patron of SPCR
Posts: 169
Joined: Sat Jul 25, 2009 12:47 pm
Location: Birmingham, UK

Post by reddyuday » Sat Oct 31, 2009 8:33 am

burebista wrote: Thank you, that's exactly my point based on testing in same conditions not on extrapolating.
I think the extrapolation issue is a red herring. If you understand what was extrapolated and how, you will see that it is a perfectly reasonable thing to do. The main difference is that your friends replaced the fan whereas SPCR didn't. While I admit that their comparison is therefore more valuable, I don't buy that SPCR did anything improper in their review. As Mike has explained, Nirvana's fan is permanently attached to the cooler and is not designed to be replaced. So, SPCR is perfectly justified in testing it as is.

RoGuE
Posts: 540
Joined: Mon Jul 06, 2009 9:11 am
Location: Massachusetts, USA
Contact:

Post by RoGuE » Sat Oct 31, 2009 10:15 am

I'm not even sure why people were complaining so much about this review...The fact is, it's a shitty stock solution, and there are many better ways to cool your cpu quietly. Makes me wonder if anyone in here has any ties to ZEROtherm.... :roll:

JL5
Patron of SPCR
Posts: 40
Joined: Tue Oct 07, 2008 12:02 am
Location: Denmark

Post by JL5 » Sat Oct 31, 2009 12:14 pm

Thank you for the review. It was useful as always.

Parappaman
Posts: 419
Joined: Mon Aug 31, 2009 12:30 am
Location: Italy

Post by Parappaman » Sat Oct 31, 2009 1:01 pm

Well, I have to agree with the ones complaining about the final "extrapolated results", though I really can't understand why some are being so rude expressing their opinion.

It has sub-par performance if compared to some of the other coolers previously tested, but I really think that the stock fan is holding this cooler's performance back, big time.

dev
Posts: 43
Joined: Sun May 10, 2009 12:47 am
Location: Here

Post by dev » Sun Nov 01, 2009 3:09 am

It took a while to get my thoughts across but in the end looks like half of the work is done.

While I fully agree that the stock fan is sub-par the good thing is that its changeable. Its not like this is a Zalman 9xxx. And I can tell you because I've tested this: Nirvana plus a Gentle Typhoon is a very decent and quiet solution.

Sometimes when evaluating products you have to tone down the criticism and try to see the good parts. Its really not fair to put down a product just because its fan is not up to your standards because in the end the fan if only a part of the equation and in this case is replaceable.

burebista
Posts: 402
Joined: Fri Sep 02, 2005 12:05 am
Location: Romania

Post by burebista » Sun Nov 01, 2009 3:44 am

Oh well, after two pages of fancy theories and graphs I find out that it's about Nirvana own fan extrapolation and not about reference fan extrapolation. Good. In fact it's not good, it's worse because I can understand that replacing the stock fan it's a NASA engineer job and it's against SPCR policy but I don't understand why was not used a fan controller or at least SpeedFan to adjust Nirvana's own fan at desired dBA levels instead educated guess some values.

This review is a perfect example of how not to do a review. If that table on page six doesn't exist it was a perfect review. We all know then that performance wise Nirvana is a very good cooler (remember is on par with Noctua and 2°C worst as Prolimatech with reference fan) with an awful stock fan. End of story for everybody.
Now for SPCR crowd next step should be testing with reference fan. It cannot be done because is too hard to change and against policy. Agree with that.
But let's see at least how Nirvana's fan performs at same dBA levels against reference fan. Ops, we don't have a fan controller nor SpeedFan so let's do a educated guess and throw up Nirvana from charts.

So in the end from a good heatsink with an awful stock fan we have an off the charts heatsink with Poor performance:size ratio.

One last thing though. I leave alone my friend tests, but maybe you have a sample with a broken fan or something. I just take a look at HWC and they were impressed
If this was just another pretty face we would not be so kind right now but with its impressively low noise envelope which is backed up by good performance totally won us over. Considering it performed extremely well while not breaking the bank like some other products, we believe it packs just the right mix of all the elements we are looking for in a modern heatsink.
XBit too
But I would like to specifically point out a remarkable performance of ZEROtherm Nirvana NV120 Premium – cooler with only four 6-mm heatpipes that managed to demonstrate the same cooling efficiency as Thermalright Ultra-120 eXtreme that has remained undefeated for quite some time now with very few exceptions.
And RoGuE thanks for your concern, I don't have anything to do with ZEROtherm, in fact I even don't like Nirvana, I speak here about reviews principles.

Nov1/09 -- Banned by Admin until he learns some manners.

MikeC
Site Admin
Posts: 12285
Joined: Sun Aug 11, 2002 3:26 pm
Location: Vancouver, BC, Canada
Contact:

Post by MikeC » Sun Nov 01, 2009 7:37 am

dev wrote:It took a while to get my thoughts across but in the end looks like half of the work is done.
This is no apology for all the ranting about extrapolation.

cordis
Posts: 1082
Joined: Thu Jan 15, 2009 10:56 pm
Location: San Jose

well....

Post by cordis » Mon Nov 02, 2009 3:05 am

First off, thanks for the review, quality work as always, succinct and to the point.

Now this is an interesting thread. Wading through some of the more truculent entries, is looks like there's a difference of philosophy with these testing methodologies. The points made here about consistency are certainly well taken, I'd certainly agree that using the same cpu for all coolers gives you consistent results across the years of testing that have been done. I suppose on the other side if you rerun all your cpu coolers on your new cpu every time, that would give you a certain level of consistency, although it sounds like a lot of work. Using the same cpu gives you consistency without the re-testing headache.

But there is a downside to using the same cpu for each test, it does lead to some worry about what would happen if the test system ever fails catastrophically. I suppose the right way to upgrade to a newer processor in that event would involve finding a cpu with roughly the same power usage, perhaps tweaking the frequency/voltage setup to make the power match as closely as possible, and then retesting a few of the older coolers to make sure the data matched. Of course, that might be tricky to do if the default test system dies suddenly and unexpectedly, has there been any thought to doing such testing before a system failure? Offhand, I'm not sure exactly where a thread about testing methodology might be. And please, don't think I'm disputing the testing methodology here, it seems sound and solid and very consistent, but having had systems of my own fail on me at various times, I do worry about the continuity of measurement when faced with a benchmarking system failure.

MikeC
Site Admin
Posts: 12285
Joined: Sun Aug 11, 2002 3:26 pm
Location: Vancouver, BC, Canada
Contact:

Post by MikeC » Mon Nov 02, 2009 8:00 am

cordis --

SPCR's Unique Heatsink Testing Methodology covers the topic more fully than the quick blurbs in the reviews themselves. It's also linked in every HS review. My guess is that a large % of HS review readers have not read this article; it is definitely worth a read.
I suppose on the other side if you rerun all your cpu coolers on your new cpu every time, that would give you a certain level of consistency, although it sounds like a lot of work. Using the same cpu gives you consistency without the re-testing headache.
This is extremely difficult to do. Not only the time/effort issue, but simple availability of older heatsinks. We hang on to the best coolers, generally, in case a close or unique comparison against newcomers is needed, but try to dispose of all others -- the storage requirements demanded by review samples are enormous.

The issue of test platform failure is a very real one that we've faced a couple times in the past. We touch on this in our intro summary of the test methodology article --
After our two main test beds for heatsink testing developed intermittant defects, we've rebuilt our test bed for the fourth (or is it the fifth) time so we can continue to bring you quality heatsink reviews. At the same time, we've brought our article on testing heatsinks up to date — a long overdue update that brings it out of the era of Athlons and P-IIIs into the present day where Socket 775 now rules the roost. We've also tossed in some gems about VRMs and testing CPU power, as well as a quick re-test of some old favorites to kick things off.
In fact, the current HS test system uses a different motherboard than the Asus P5LD2-VM described in the methodology article. It was changed to a new board at the beginning of 2009, and we reported on this on p4 of our Thermalright BigTyphoon 14 Pro review:
Test Platform change:
Due to the amount of wear and tear to which we subject this test platform (which we also use for hard drive testing) it was decided to replace the motherboard with a newer, more current model that could be easily replaced should it be damaged. Asus kindly provided us with two P5Q-EM's — one will be kept as a spare in case anything happens to the original. Testing showed that the CPU temperature reported by SpeedFan on the new motherboard was exactly what it produced on our original platform. The hard drive was also replaced with a solid state drive — the system now has no moving parts aside from the products we evaluate.
So that's the answer: a backup board for insurance. The 2nd spare board simply sits on the shelf most of the time, getting pulled into service for the odd function when it's the only appropriate board for whatever we're trying to do. The CPU is so unlikely to fail that we don't even consider the possibility.

MikeC
Site Admin
Posts: 12285
Joined: Sun Aug 11, 2002 3:26 pm
Location: Vancouver, BC, Canada
Contact:

Post by MikeC » Mon Nov 02, 2009 9:30 am

Whether a 95W TDP CPU can push current heatsinks enough to provide relevant data about their performance with hotter CPUs is not an unreasonable question, even though the way in which the issue was raised by burebista was unfortunate.

I've spent some time over the weekend building a new CPU power measurement tool, and running various experiments on different CPUs and boards, and heatsink/fans. This is at an early stage, and will continue, then all the info/data collected in an article in the near future.

Some early notes of interest:

A) A brand new AMD II X4 965 3.4GHz w/125W TDP was installed in an Asus M3A78-T mobo running Vista. This CPU is extremely useful because it is multiplier unlocked, allowing the CPU to be ramped up and down in speed/heat at will w/o stressing bus-speed dependent components on the mobo. With all stock setting and unconditionally good cooling for the CPU, VRMs and Chipset, the power draw at the 2x12V socket (which is where the board gets all the power for the CPU) was measured after 15 minutes at full load:

w/ prime95x4 -- 82W
w/ CPUBurn (on k7) x4 -- 107~108W

1. Obviously, for AMD processors, CPUBurn remains the stress utility of choice. Not so with Intel chips, which always run a bit hotter w/ Prime95.

2. The 107~108W measured load is way below the CPU's 125W TDP. These stress utilities heat up the CPU more than any real-use apps in real computers that people use -- including the latest 3D games. One big difference is that our stress test load is continuous, while even in the most intense games, there's some up and down, which prevents the CPU from getting quite as hot.

3. It is important to note that the power being measured is at the VRM input, where 12V is taken into and converted to the 1.0~1.5V required by the CPU. There are losses here; the CPU does not see all of that 107-108W. I've had help from an Intel power engineer in the past to measure the VRM efficiency of a few motherboards. The boards that were measured had peak efficiencies of 90% for the best (a high end workstation class Intel board) and about 81% for a mid-class consumer Intel mATX board. All of the VRMs exhibited variations in efficiency at different power loads. See the sample graphic below.

Image

This Asus board is rated to handle 140W TDP CPUs. I've requested an efficiency curve of the VRMs on this board from my tech rep at Asus, but I'm not holding my breath; past experience tells me that unless I'm face-to-face with a power engineer from Asus, this info will not be forthcoming.

The main point is this: Even the 108W measured here is NOT what this 125W TDP CPU generates; it's less than that. If the VRM efficiency is 90%, then the CPU power is 97W. If it is 80%, the CPU power is 86W.

4. With the same new power measurement tool, the power at the 2x12V socket for our Pentium D platform measures 88W -- 10W higher than with the previous tool (which we have to ignore because that tool is broken and so cannot be used to measure any of these mobos now).

If the mobo for our Pentium D has 90%/80% VRM efficiency, the CPU power is 79W/70W -- which is not far from the 97W/86W of the 125W TDP CPU.

B) Overclockers face a more difficult challenge. Deliberately overvolted to 1.45V and OC'd to 3.7GHz, the measured power went up to ~138W under the same conditions as above. Chances are, the VRM efficiency is lower at this point, tho, so the CPU power again is less than you'd expect.

Enuf for now, I'm starting to ramble...
Last edited by MikeC on Mon Nov 02, 2009 9:52 am, edited 2 times in total.

cordis
Posts: 1082
Joined: Thu Jan 15, 2009 10:56 pm
Location: San Jose

wow!

Post by cordis » Mon Nov 02, 2009 9:39 am

OK, now that's a plan, sounds like you have it all well in hand, I'm completely reassured. With a backup plan like that I expect to see the great reviews just keep on coming. Thanks for the info!

nyu3
Posts: 144
Joined: Wed Feb 18, 2004 6:07 pm
Location: Canada

Post by nyu3 » Tue Nov 03, 2009 4:36 pm

Even if you ignore the extrapolations, it is clear that the Nirvana isn't a top candidate for quiet cooling:
1) tight fin spacing
2) inability to replace stock fan (which is poor) without modding
3) poor performance vs noise ratio compared to Scythe Mugen 2 and Xigmatek HDT (they cost around the same)

dev
Posts: 43
Joined: Sun May 10, 2009 12:47 am
Location: Here

Post by dev » Wed Nov 04, 2009 8:15 am

No offence guys but until one of you tests or buys this heatsink your theoretical observations are just that.

1. The fins are not that tight
2. The fan can be removed very easily
3. I don't think this review did an apples to apples comparison so you only have subjective data.

Around Europe the Nirvana is around 35 euros which is very good value.

MikeC
Site Admin
Posts: 12285
Joined: Sun Aug 11, 2002 3:26 pm
Location: Vancouver, BC, Canada
Contact:

Post by MikeC » Wed Nov 04, 2009 12:10 pm

You know the Nirvana and Zen are virtually identical in almost every way except for the extra thin bit of metal on the ends of the fins which creates the " niche" for the stock fan in the Nirvana.

The spacing of the fins, the denser pattern in the middle, the base, the heatpipes, the way it mounts -- all these are identical. The Zen is a bit taller (less than 1cm) and in that extra height, there are 4 more fins than on the Nirvana. Also, the Nirvana is nickel-plated, which has no impact on cooling though it might wear better if you live near the sea (corrosion, oxidation).

So how did the Zen do with our reference quiet fan? see here: http://www.silentpcreview.com/article849-page6.html Quite well at >9V but increasingly worse at slower speed. It's likely that the Nirvana would perform similarly.

Which is the better buy? At newegg in the US, the Zen is $8 cheaper, and there's no issue whatsoever about fan swapping.

-------------
Further notes on fan swapping on the Nirvana --

The Nirvana's stock fan is easy enough to remove, but our first examination did not show any way to mount a standard 120mm fan; hence the test with just the stock fan. When burebista showed pics from dev's review, Larry and I both scratched our heads wondering how they managed to get that 120mm fan on there. The space in which the stock fan fits is about 1-2mm too narrow (tight) for a standard box 120mm fan.

However, after trying some different things this morn, I discovered that if you force it really hard, the fins bend a bit and a stock 120mm fan can indeed be jammed very tightly into that space. This is not intuitive, and I think there is some risk of damage to the heatsink. Still, the box fan can be mounted in there, and this is news to us.

Now if this been explained by dev (or burebista, if he knew) early in this discussion before all the crap started flying, we probably would have gone back to try it and retest.... but it's history now, we're busy with other reviews. The insistent, rude dissing of our review and extrapolation (which burebista never understood until his last post, then failed to apologize about, and which dev still fails to understand) makes this a difficult discussion not to dismiss as typical web BS.

new2spcr
Friend of SPCR
Posts: 230
Joined: Wed Apr 22, 2009 9:14 am
Location: Sweden

Post by new2spcr » Wed Nov 04, 2009 11:23 pm

MikeC wrote: but it's history now, we're busy with other reviews.
Thanks Mike. ^^

frenchie
Friend of SPCR
Posts: 1346
Joined: Tue Jul 08, 2008 4:53 am
Location: CT

Post by frenchie » Wed Nov 04, 2009 11:30 pm

Good job MikeC !!
Thanks for the review. Very usefull as usual !

Too bad this cooler doesn't perform well, I like it's look :)

RoGuE
Posts: 540
Joined: Mon Jul 06, 2009 9:11 am
Location: Massachusetts, USA
Contact:

Post by RoGuE » Thu Nov 05, 2009 12:07 am

MikeC wrote: but it's history now, we're busy with other reviews.
Thanks for taking the time to address some of the concerns laid out in this thread (as rude as some of them may have been...)

I think at this point now, you have done everything in your power to explain yourself and your review. At this point if anyone is still disagreeing or questioning it, then they are simply too ignorant to understand or interpret it all.

I almost wish this thread were locked..some people are still running in circles.

dev
Posts: 43
Joined: Sun May 10, 2009 12:47 am
Location: Here

Post by dev » Thu Nov 05, 2009 4:39 am

I enjoy the groupies as much as anyone out there but we were just trying to separate fact from fiction. The first 5 pages of the review are excellent and I have nothing bad to report aside from a bit of misleading about the replacing of the stock fan.

The last page troubles me. Why would someone want to publish that chart without proper testing is beyond me. We get your point on page 5: the stock fan is pretty weak undervolted but there's no way on this Earth this heatsink is 6 degrees worse than a FZ120. You said it Mike, its a nickel plated version of the FZ or if you are rigorous the FZ is a Nirvana without the nickel plating.

Below is a link to a little test, Nirvana and Megahalems powered by the same fan, Noctua NF-P12:

http://forum.crazypc.ro/showpost.php?p= ... tcount=336

As you can see its a few degrees behind but certainly not above 10.

To understand why I don't like extrapolation or comparing heatisnks with different fan setups you need to take a look at what mess is this method causing at frostytech, tweaktown to name just a few.

I know Mike and the guys mean well but sometimes they slip. Its no problem if they correct the mistake and improve later. I suggest you to pay a visit to hardwarecanucks and take a look at their latest heatsink test. It took a thread like this one to make them realize that they were doing it wrong.

Have a good one guys.

frenchie
Friend of SPCR
Posts: 1346
Joined: Tue Jul 08, 2008 4:53 am
Location: CT

Post by frenchie » Thu Nov 05, 2009 5:00 am

Dev, interesting link... I guess...
Can you explain what it means ?
What is the rpm of the fan ?
What is the load of the system ?
How and where is the temperature measured ?
What is the noise of the system ? (it is SPCR after all)

Since MikeC spent so much time explaining and justifying his results, it would be respectful to do the same ;)

jhhoffma
Posts: 2131
Joined: Mon Apr 25, 2005 10:00 am
Location: Grand Rapids, MI

Post by jhhoffma » Thu Nov 05, 2009 9:54 am

dev,

I'm just not sure how you're not understanding this. It doesn't matter how well the Nirvana performs with a different fan. It's not meant to be replaced!! In its packaged state the fan is a significant factor in the performance of the entire HSF combo. You want a separation between Nirvana the Heatsink and Nirvana the Heatsink WITH the included FAN. There is no distinction as it's hard-mounted!!

It's like complaining that a review between a Corvette and a Viper favors the Viper because the reviewer didn't put a better engine in the Corvette. Doesn't make much sense, does it?

Extrapolation is only used because of the confines of the review. And to that effect, all reviews are conducted in the same manner: unless the HSF allows a fan to be replaced without modification to the structure, it's not done. PERIOD. If you don't like comparing heatsinks with different fans, then you should be failing the Nirvana right away.

No one is saying the heatsink itself is not effective, just that the fan and it's mount is crap. Get over it. No exception was made for Zalman when they were the poo early on in SPCRs history and no exception was made for Zerotherm here.

By the way...SHAME ON YOU MIKEC FOR GOING THE EXTRA MILE!! :lol:

What should have been done is for burebista and yourself to say "Hey Mike, liked the review. However, we did some additional testing and had the following results. They do conflict with some of your extrapolations, however. Could you better explain how those were done?"

Probably would have saved about 20 posts of bullcrap... :roll:

RicaNeaga
Posts: 7
Joined: Wed Sep 03, 2008 10:31 am
Location: Romania

ZEROtherm Nirvana CPU Cooler

Post by RicaNeaga » Sat Nov 07, 2009 4:10 pm

Well, this is my first post although I'm an old member and a true fan of SPCR (I really love the tests regarding power supplies, cases and fans). Hell, it's about time! (starcraft 2 quote) :lol:

The reason behind this is that some of my compatriots are trying to make a point, but their manner is somewhat rude, and I would like to apologize on their behalf. So...


1. I asked the author of the review that was quoted by burebista how he managed to swap the fan on his Nirvana. He said that he had to do some minor improvements to the heatsink itself. But we both agreed that FOR THE USUAL USER, SWAPPING THE FAN ON A NIRVANA IS NOT THAT EASY (as a walk in the park). :wink:

I, for instance, when I saw for the first time a Nirvana cooler, it didn't cross my mind that the fan could be swapped... I really thought it was a Zalman look-alike.

The debate here regarding this point in my opinion is as futile as the extrapolation of the capabilities of the stock fan, since both are quite clear: the usual user won't swap the fan and the stock fan that Nirvana has is poor.


2. SPCR's methodology of testing coolers is in my opinion (also) outdated. And now I speak from a silent freack's point of view. I really want to see how the coolers behave on a modern CPU, that is hotter... and gets hotter as it's moderately overclocked (since some heatsinks CAN cool down even a quad-core with a low-rpm fan or fanless, than why not overclock it? I'm also into performance as much as I'm into silence).

And from what I've read in other tests, their results and their classifications doesn't concur with SPCR's. For example Cooler Master Z600 - a mediocre candidate on a SPCR test platform, can be a very powerful contender on an overclocked Q6600 or on an overclocked i920.

You don't agree Mike? Then maybe you should include in your...
various experiments on different CPUs and boards, and heatsink/fans
a HDT cooler (I bet you a chinese beer that on a small IHS, a HDT with many heatpipes will have a very different behaviour than on a large IHS processor like i7), Z600, a Ninja, and different frequencies for the processor (an i920 would be great). And maybe the results of this will make you reconsider the test platform (and since Intel 32 nm hexa-core is just around the corner, maybe that would be a great addition to it).

Please don't try to explain theoreticly I'm wrong, just test them and see for yourself in practice. I thank you in advance.


Also, I think that what you are going to test about the efficiency of the VRMs is a VERY interesting subject. It's also interesting if you consider video cards, because many people will try to replace the stock cooling with an aftermarket one to silence them. But most of them aren't aware about the danger of not taking into account the cooling of the vrm area, that may be very tricky on some implementations.


And since this is my first intervention, I'm not finished with the requests. 8) Please forgive burebista for his tone, he means well. He just wants to see better tests from SPCR, as we all do. And like me, he just tried to explain himself, and maybe sometimes his tone rose as much as his expectations from SPCR have risen over the years... :)

reddyuday
Patron of SPCR
Posts: 169
Joined: Sat Jul 25, 2009 12:47 pm
Location: Birmingham, UK

Re: ZEROtherm Nirvana CPU Cooler

Post by reddyuday » Sun Nov 08, 2009 2:06 am

RicaNeaga wrote:2. SPCR's methodology of testing coolers is in my opinion (also) outdated. And now I speak from a silent freack's point of view. I really want to see how the coolers behave on a modern CPU, that is hotter... and gets hotter as it's moderately overclocked (since some heatsinks CAN cool down even a quad-core with a low-rpm fan or fanless, than why not overclock it? I'm also into performance as much as I'm into silence).

And from what I've read in other tests, their results and their classifications doesn't concur with SPCR's. For example Cooler Master Z600 - a mediocre candidate on a SPCR test platform, can be a very powerful contender on an overclocked Q6600 or on an overclocked i920.
I think the point you are trying to make is that the relative ranking of the coolers can be different on a 65W CPU, a 95W CPU, a 125W CPU or an overclocked 175W CPU. (The "modernity" of a CPU is a red herring, as pointed out several times in this thread. It is the heat generated by the CPU that matters in evaluating the cooler.)

That is a fair point. But if SPCR tests all the coolers on an overclocked 175W CPU, how does it help the rest of us who use 65W CPUs, or even lower-powered ATOMs etc? I think a 95W CPU provides a happy medium, and people can extrapolate the results for their choice CPUs.

The reason SPCR uses an old Pentium 4 for tests is that it has been used in their tests of all the other coolers. Continuing to use it provides a fair comparison between the different coolers.

On the otherhand, SPCR provides an open forum, and the people that are interested in high-powered processors are welcome to share their results or point to other reviews etc. I don't see the problem. In fact, there has been a great deal of discussion here about the relative merits of Megahalems, TRUE and other coolers on i7 processors and we came to the conclusion that the SPCR results may not extrapolate linearly.

Uday

RicaNeaga
Posts: 7
Joined: Wed Sep 03, 2008 10:31 am
Location: Romania

Post by RicaNeaga » Sun Nov 08, 2009 4:46 am

reddyuday wrote:That is a fair point. But if SPCR tests all the coolers on an overclocked 175W CPU, how does it help the rest of us who use 65W CPUs, or even lower-powered ATOMs etc? I think a 95W CPU provides a happy medium, and people can extrapolate the results for their choice CPUs.

In fact, there has been a great deal of discussion here about the relative merits of Megahalems, TRUE and other coolers on i7 processors and we came to the conclusion that the SPCR results may not extrapolate linearly.
I think that those two bolded sentences are in a clear contradiction. And that is what I am trying to say here.

What do you want: a classification made on a medium test platform that isn't relevant for an overclocked modern CPU like i7 (where for example your higly-praised HR-01+ or TRUE don't stand a chance against Z600 with a low-rpm fan), or a classification that shows how the coolers behave on a more stressful platform?

I think that a reader of the latter kind of tests is inteligent enough to realize that if he buys a lower-heat-generating CPU he can also purchase a cooler that isn't exactly the best performer at low rpm on a more stressful platform.

But a reader of your tests isn't going to buy a let's say CM Z600 cooler for an i7, but rather a TR HR-01+, and then based on your findings and his experience, he will conclude that no cooler can cope with an i7 fanless since HR-01+ can't. And he resigns himself to that conclusion.

The time when an air-cooler wasn't enough for a ,,happy medium" is gone (when you started using that P4 for your tests), and now many coolers are suitable for silent-computing even in the case of an overclocked ,,not-so-happy-highly-generating-heat" CPU.

And from what Mike was saying earlier, about experimenting on other platforms and CPUs, a Phenom 2 processor won't make a difference for this argument: it is a known fact all over the internet that a Phenom 2 CPU is easy to cool down by even a medium-ranked air-cooler.

That's all. I hope I haven't offended nobody and I will remain a fan of SPCR when it comes to dB measurings. And I will continue to look for cooler tests elsewhere, since I don't consider the SPCR ones the most relevant for me. :cry:

Maybe this you should ask yourself - how many potential readers you loose or not fully inform with your current methodology of CPU cooler testing? Can you afford the luxury of continuing to argue on your forums with your readers and keep that methodology intact?

nutball
*Lifetime Patron*
Posts: 1304
Joined: Thu Apr 10, 2003 7:16 am
Location: en.gb.uk

Post by nutball » Sun Nov 08, 2009 5:30 am

My take on this is that the tensions exposing themselves in this thread have been growing here for some years now. I'm a little surprised it's taken so long to come to the surface.

When I started visiting here overclocking was the preserve of an almost invisible minority. These days it is the preserve of a larger and more visible (and seemingly vocal) minority, though still a minority as far as I can tell.

I can't see how one site can sensibly cater to both those who enthusiastically underclock and undervolt trying to save every last watt, and those who overclock and overvolt trying to extract every last FPS. There are many sites catering for the latter group on the internet, but very few catering for the first group.

Personally speaking I would prefer that SPCR continue to serve up reviews relevant to the section of the market not already done to death by others.

MikeC
Site Admin
Posts: 12285
Joined: Sun Aug 11, 2002 3:26 pm
Location: Vancouver, BC, Canada
Contact:

Post by MikeC » Sun Nov 08, 2009 10:20 am

But a reader of your tests isn't going to buy a let's say CM Z600 cooler for an i7, but rather a TR HR-01+, and then based on your findings and his experience, he will conclude that no cooler can cope with an i7 fanless since HR-01+ can't. And he resigns himself to that conclusion.
Not a good example -- the CM Z600 will lose at low airflow against the TR HR-01+ with any CPU and neither will cool effectively enough w/o a fan on an i7.

There's some merit in running a separate hotter CPU heatsink test rig, but I can't see it replacing the existing test setup. The 95W TDP of our Pentium D is just about right to cover most CPUs. It is not too far from the 125/130W TDP of the hottest CPUs (i7 & AMD Phenom) -- I think the 140W Phenom II 965 will not last long; the 125W version replacement is already out.

I mentioned this in an earlier post:
A brand new AMD II X4 965 3.4GHz w/125W TDP was installed in an Asus M3A78-T mobo running Vista. This CPU is extremely useful because it is multiplier unlocked, allowing the CPU to be ramped up and down in speed/heat at will, w/o stressing bus-speed dependent components on the mobo. With all stock setting and unconditionally good cooling for the CPU, VRMs and Chipset, the power draw at the 2x12V socket (which is where the board gets all the power for the CPU) was measured after 15 minutes at full load:

w/ prime95x4 -- 82W
w/ CPUBurn (on k7) x4 -- 107~108W
What I didn't mention is that the power draw of the 95W TDP Pentium D on the existing platform measures 88W (Prime95) with this new tool, 10W higher than with the old tool, which is now suspect.

Now, by increasing the Vcore to 1.35V in the BIOS w/o any other changes, I got the measured 2x12V socket power to 120W. This is higher than the stock 125W TDP Phenom II 965. I'll have to set up an i7 system to see what those CPUs really pull at the 2x12V socket.

More experiments are in order. The gist of what I am trying to determine:
1) Does C/W of a heatsink/fan stay the same as CPU power is increased?
2) If not, how much more power is needed for C/W to change? Is there a consistency to that change? ie, for every +10W, C/W decreases (or increases) by 10%, or whatever pattern can be discerned.
3) Does the ranking of SPCR's top low-airflow coolers stay the same when a CPU of higher TDP is used?
4) Finally, if the decision is made to add a high power CPU test, what should that power level be?

Post Reply