Stren's 2012 CPU Water Block Roundup

[rge]

Wait a minute, waaaiiit a minute. This is terribly backwards. Someone does a review and website wants the reviewer to pay them for doing all of the work? It should be the other way around. You did the work, they should be paying you to use your information on their site. I feel sorry for you that you actually paid them already. I'm glad that I no longer frequent that site.

He didnt pay them, which is why OCN made him delete all the information in his review thread. But the OCN owner is clearly brilliant, just have to follow his logic.....take all independent reviewers/testers which bring a lot of traffic to the site, charge them and make them hand over rights to their work, which should effectively eliminate all of them from the site, and permanently drive down traffic. Then watch the money come pouring in.

[stren]

Wait a minute, waaaiiit a minute. This is terribly backwards. Someone does a review and website wants the reviewer to pay them for doing all of the work? It should be the other way around. You did the work, they should be paying you to use your information on their site. I feel sorry for you that you actually paid them already. I'm glad that I no longer frequent that site.

He didnt pay them, which is why OCN made him delete all the information in his review thread. But the OCN owner is clearly brilliant, just have to follow his logic.....take all independent reviewers/testers which bring a lot of traffic to the site, charge them and make them hand over rights to their work, which should effectively eliminate all of them from the site, and permanently drive down traffic. Then watch the money come pouring in.

Yeah rge is right. I didn't pay them although Spotswood was kind enough to volunteer to pay the fee. However OCN wouldn't accept the payment even though they had asked for it because I wouldn't give them exclusive access and delete the same info from XS/[H] and my own website lol

I think their theory is that they are big enough to push people around like that. While they do get a lot of views now, it won't be sustainable if there isn't any good content.

BTW anyone know how I close the poll? Seems silly to still have it running.

[jogshy]

Stren, you should test the ALPHEOS waterblock from Coollaboraty ( the creators of Liquid Ultra/Pro ):



http://coollaboratory.com/shop/produ...bde54903ba3d7f

[stren]

Stren, you should test the ALPHEOS waterblock from Coollaboraty ( the creators of Liquid Ultra/Pro ):

Interesting - I hadn't seen that before - thanks I'll write to them and ask for a sample.

Meanwhile I got done with the CPU-380 data vs flow/pump setting - it's still kicking butt, in fact kicking so much I had to readjust all my axis' to have more range:

[bds71]

wow, color me suprised! even though the 370 was already at the top of the chart - to have that much of a decrease in temp (in the 380) and still scale that well to flow? incredible.

something interesting caught my eye though - in Skinnee's EK Supreme (Full Nickel) review, he had the Supreme about 1C better than the 370 from 1 to nearly 2.5 GPM (IX) using an i7 930 at 1.5V. i knew the trick with the supreme was contact (more pressure on regular paste would get stellar temps - or IX). with the supremecy, it almost looks like EK took a step backwards. i would like to see how the Supreme (with jet plate 6) fares in this comparison on IX. i don't think it will have the flow characteristics to beat the 380 - especially at high flow rates - but i think it would beat the supremecy? (based on Skinnee's review and it's relative placement with the 370 here)

question: what speed are you running your fans at? i was just curious what your overall (full load) dT is at high flow...

[felix_w]

I wonder where would my D-Tek Fuzion V2 fit in that chart ?

[stren]

wow, color me suprised! even though the 370 was already at the top of the chart - to have that much of a decrease in temp (in the 380) and still scale that well to flow? incredible.

something interesting caught my eye though - in Skinnee's EK Supreme (Full Nickel) review, he had the Supreme about 1C better than the 370 from 1 to nearly 2.5 GPM (IX) using an i7 930 at 1.5V. i knew the trick with the supreme was contact (more pressure on regular paste would get stellar temps - or IX). with the supremecy, it almost looks like EK took a step backwards. i would like to see how the Supreme (with jet plate 6) fares in this comparison on IX. i don't think it will have the flow characteristics to beat the 380 - especially at high flow rates - but i think it would beat the supremecy? (based on Skinnee's review and it's relative placement with the 370 here)

question: what speed are you running your fans at? i was just curious what your overall (full load) dT is at high flow...

Yes I was surprised how well the 380 did vs flow too. It's really hard to find a fault with the block so far.

In the MX2 testing the Supreme HF JP6 was ~0.4C worse than the Supremacy. This was within expectations for EK for the 3930K. The big surprise to them was that both did so poorly in my testing. It's not just skinnee's review that my data disagrees with on EK's performance but also Martin's (who also tested on a 3930K unlike skinnee I believe), so I'm not sure what to make of it. I did retest the Supremacy for EK as they thought I may have polished the IHS so much over time that it had skewed data towards later blocks, but the retest results were close to the original data. EK have said they may send a 2nd Supremacy in case that one was a particularly poor sample.

For fans I use 2150rpm gentle typhoons in push/pull on a 560 rad with 120->140mm adapters, though as I'm comparing to water temps then it shouldn't matter much. However Delta T of water to Radiator Air Out =~0.4C, for the Air In it's more like ~3.5C. Though I think the radiator air in sensor placement is particularly poor. I didn't worry though because it was more of a sanity check than a real data point. Ideally if you were going to believe the air numbers you'd want multiple sensors per fan placed accurately near the fan intake.

[UrbanSmooth]

Stren, you should test the ALPHEOS waterblock from Coollaboraty ( the creators of Liquid Ultra/Pro ):



http://coollaboratory.com/shop/produ...bde54903ba3d7f

Curious.

[stren]

Curious.

Yeah I wrote to them, we'll see what they say

[bds71]

yea, i meant water:ambient (air in). 3.5C is pretty nice!! as for the results of the supremecy, could it have the same love for tension that the supreme had? what tension did you test at for the paste? how much difference was there between paste and IX? i'll be the first to admit: i usually just go straight to the IX results because pressure can have so much influence on the results that my results could vary greatly from yours (sometimes as much as 5-6C) - whereas the consistency with IX is what i chose to look at first. anywho, i would still like to see the supreme results with IX (to compare both with the 380, and the supremecy). honestly, the results of the supremecy on IX were rather dissapointing. are you sure you had a good flow lol? (j/k)

[stren]

yea, i meant water:ambient (air in). 3.5C is pretty nice!! as for the results of the supremecy, could it have the same love for tension that the supreme had? what tension did you test at for the paste? how much difference was there between paste and IX? i'll be the first to admit: i usually just go straight to the IX results because pressure can have so much influence on the results that my results could vary greatly from yours (sometimes as much as 5-6C) - whereas the consistency with IX is what i chose to look at first. anywho, i would still like to see the supreme results with IX (to compare both with the 380, and the supremecy). honestly, the results of the supremecy on IX were rather dissapointing. are you sure you had a good flow lol? (j/k)

On the supremacy the first run I bottomed out the thumb nuts. On the retest I did 5 turns on the screws which was the recommendation by EK.

Here's a plot showing the delta between the 5 MX2 paste runs and IX at one fixed pump setting:



It's not complete yet obviously.

I was going to do the Supreme HF and the IceForce HF next but I dropped and bent my QDC so I have to order another one

I also finally ordered a differential manometer so once I'm done with IX I'll bang out the flow/pressure aka restriction plots

Had to settle for the 475 series but it should be good enough for now

[rge]

Strens testing suggests the problem with EK supremecy is in the internals. The indigo test (which nearly eliminates contact/bow/mount differences) showed the supremecy was just as mediocre performing relative to others as mx2 test (where bow/contact is more important). Would be interesting to see if EK sent another one to test.

[bds71]

The indigo test (which nearly eliminates contact/bow/mount differences) showed the supremecy was just as mediocre performing relative to others as mx2 test...

i agree - i think that sending another sample will only result in a similarly dissappointing result. looks like ill just stay with my supreme for now (depending on the IX showing for the supreme once Stren has a chance to run it). i do like the 380, i just don't think it will be enough better to justify buying one vs using my already-owned supreme.

[st0ned]

I beg to disagree, It really depends upon your socket and might depend also upon your sample. Supremacy is far from a poor performer.

In my testing with 1155 socket Supremacy was the best performer for exemple. Though I didn't had the opportunity to test as many blocks as stern did, sadly lot of important brands didn't want to send samples or didn't even reply.

[bds71]

you may very well be correct about the supremecy's performance on 1155/1156. but my statement stands, as it was intended towards socket 2011 (which was tested here - and which i own and have a personal vested intetrest ). i would readily accept that the tested block was a "freak" and that it is flawed - and would greedily accept new information saying it was the "be-all/end-all" of water blocks after it blew the competition out of the water!! (really, i would!!) however, i simply don't think that's the case - i think its performance here is typical, and that retesting with a different block might (or might not) be better - but, at best i still doubt it would be anywhere near the performance levels of the 380. ie: it still falls short on the platform i care about, and which this review was intended.

Stoned: this was written with a grain of humor that may not come accross in reading - i very much respect your opinions and your reviews. please don't take this to be insulting to you in any way

[stren]

you may very well be correct about the supremecy's performance on 1155/1156. but my statement stands, as it was intended towards socket 2011 (which was tested here - and which i own and have a personal vested intetrest ). i would readily accept that the tested block was a "freak" and that it is flawed - and would greedily accept new information saying it was the "be-all/end-all" of water blocks after it blew the competition out of the water!! (really, i would!!) however, i simply don't think that's the case - i think its performance here is typical, and that retesting with a different block might (or might not) be better - but, at best i still doubt it would be anywhere near the performance levels of the 380. ie: it still falls short on the platform i care about, and which this review was intended.

Well we'll find out soon - the new block is shipping from ppcs today!

[stren]

Today I received a package from PPCS - a new EK supremacy and a new Koolance QDC to replace the one that dented. Dear Koolance - please drop test VL4N's in future I also received my brand new manometer for the flow/pressure data. That will have to wait until I'm done with IX testing though as I don't have a spare pump.

I just reflowed IX on the Supreme HF so I'll take the data on that, and then get the Supremacy MX2 done next.

[stren]

Time for an update - I got done with the testing vs flow/pump setting with Indigo Extreme. It showed some interesting results. The CPU-380 was the clear leader with MX2. But certain blocks did much better with IX than expected including the HeatKiller and the Phobya blocks. Apologies for those of you who are color blind. I'll break this out into separate plots vs 3 settings soon:



Also vs flow:



You'll note that I also tried rotating the CPU-380 and it does 0.5-1C better. In my testing I kept the block orientation so that the manufacturer's logo/text was correctly orientated when mounted in a "normal ATX tower case". Therefore I suspect that all blocks with channels would benefit from having the channels aligned "vertically" rather than "horizontally". I'm also assuming that blocks with pins don't suffer the same problems which may or may not be fair depending on the block. I made a list of the blocks and their configurations:



I also included a count of the port threads after alphacool's was singled out. It seemed right to check into all of them.

I then went back and compared the IX results with the MX-2 results at the 39% pump setting which was common to both. I found some interesting results:



I expected all the blocks to be similar affected by IX and MX-2. There are particular blocks that give strange results. I'm not sure if this reflects on poor MX2 data on the few or poor IX mounts on the good ones. The HeatKiller I've retested this week on mx-2 and I'll be going back to test the Phobya, the M6 and a few more. I also want to spot check a few more blocks vs rotation. However I won't get to doing most of these til January as I have to close down my "lab" in the spare bedroom for holiday visitors!

[rge]

Great info, thanks again Stren for all the testing. So looks like koolance cpu 380 channels aligned parallel with core length, phobya, and alphacool have best internals (and heatkiller). Also, did phobya data get converted/graphed wrong in one of the 2 graphs looking at 380 90 deg as control for 2 graphs?

[stren]

Great info, thanks again Stren for all the testing. So looks like koolance cpu 380 channels aligned parallel with core length, phobya, and alphacool have best internals (and heatkiller). Also, did phobya data get converted/graphed wrong in one of the 2 graphs looking at 380 90 deg as control for 2 graphs?

I don't think so - I'm not seeing what you mean?

Yes the phobya and heatkiller had the correct orientation. I'm not sure why there is such a difference in the mount though 5C is a huge difference just on TIM. As usual it makes me second guess everything and I have to retest a bunch to confirm lol. No wonder this has taken 6 months so far!

I've also updated all the block internal photos on my website (extremerigs.net), the only one I'm missing now is the rasa and that is a low priority.

[Martinm210]

Wow, thanks so much for all the work you put into this. That is amazing work and it raises the bar in how to do CPU block testing right. My hat is off to you sir!!

[bds71]

wow - i hadn't looked at this chart in a while....the phobya? where the heck did that come from? (note: i'm looking at the IX vs flow rate/pump setting - chart). i assume it only goes to 1.75 GPM due to its restrictiveness (like the Rasa and Apogee HD)? but wow - that's some impressive numbers!!

question: since the Heatkiller block was practically identical to the 380 (both flow rate and pump setting chart) in its normal configuration, did you also turn the Heatkiller 90 degrees to see if it would have an effect? hard to believe a 5 yr old block keeps up with the brand new (and quite impressive) 380 (though it does fall off after 2 GPM)

also, why wasn't the phobya recommended? i didn't see any specific comments about it, and (per the chart i am looking at anyway) it seems to be better than the 380 (the best in fact!! - and by half a degree to boot!!) and cheaper to boot? granted - the pump setting chart has it neck and neck with the inverted 380 so the actual temps probably would not be better....at least until you get to the higher flow rates where it starts pulling ahead....

[rge]

The koolance 380 and Phobya through most users pump settings would be a performance tie. The "GPM flow chart" is actually an "unfair test" since to achieve equivalent flow in the more restrictive Phobya, you are actually adding more pumping power to that block than Koolance...nobody in real world sets their pump speed a notch or 2 higher on one block vs another to equal flow through a more restrictive block...the restriction is part of its performance.

[stren]

wow - i hadn't looked at this chart in a while....the phobya? where the heck did that come from? (note: i'm looking at the IX vs flow rate/pump setting - chart). i assume it only goes to 1.75 GPM due to its restrictiveness (like the Rasa and Apogee HD)? but wow - that's some impressive numbers!!

question: since the Heatkiller block was practically identical to the 380 (both flow rate and pump setting chart) in its normal configuration, did you also turn the Heatkiller 90 degrees to see if it would have an effect? hard to believe a 5 yr old block keeps up with the brand new (and quite impressive) 380 (though it does fall off after 2 GPM)

also, why wasn't the phobya recommended? i didn't see any specific comments about it, and (per the chart i am looking at anyway) it seems to be better than the 380 (the best in fact!! - and by half a degree to boot!!) and cheaper to boot? granted - the pump setting chart has it neck and neck with the inverted 380 so the actual temps probably would not be better....at least until you get to the higher flow rates where it starts pulling ahead....

The Phobya block did do very well with IX, but not so well on MX-2. I'm re running it with MX-2 to make sure the data is good.

Yes it's restrictive so 1.75 is the most I could get.

Hmm my diagram says that the Heatkiller is the wrong way round in testing, but I think it was the correct way round. I think that might be a mistake. So I don't think it will do better than the 380 rotated. I'll need to check. Yes for an old block it still does very well, probably because it's been revised a few times. It does have the worst clearance for fittings though.

If the Phobya data is repeatable then it's hard to recommend because the MX-2 data is much worse. Most people use regular TIMs and not IX. RGE has it right looking at data vs flow makes highly restrictive blocks look better than they are.

The koolance 380 and Phobya through most users pump settings would be a performance tie. The "GPM flow chart" is actually an "unfair test" since to achieve equivalent flow in the more restrictive Phobya, you are actually adding more pumping power to that block than Koolance...nobody in real world sets their pump speed a notch or 2 higher on one block vs another to equal flow through a more restrictive block...the restriction is part of its performance.

Bottom line those two are ~ performance tie, imo, since performance is same at a given pump setting. And the Phobya is not recommended for a reason. Though many have complained about the cheap quality of the Phobya...the issue is the fact it is only 50% threaded, and is easily stripped. Though manufacturer said they were fixing it with new block release....so maybe on next one.

See pic at post 87 in this thread.
And if that pic doesnt work, I couldnt get it to, also posted here: http://forum.alphacool.com/viewtopic.php?f=44&t=2653

Actually only the alphacool has the bad threading, although the bases look similar the internals aren't the same and neither is the threading. The phobya has a full 3 turns of thread at least.

[rge]

oops...yeah got those two confused...I edited out my post. thanks stren.