Waterblocks on i7 - Part 5

[G-UNIT91]

THANK YOU SHANE!!

[ecat]

Many thanks for all your time and attention to detail, it really is appreciated

[Zehnsucht]

Very nice test

[Gadgeteer57]

Many thank's for this excellent test&efort

[beast200]

Hi

i have tried the dek v1, ek sumpreme, swiftech gtz and now using the HK3 LT
in order of temps
HK3 LT 37 -39
Supreme 38 - 40
GTZ 40-43
Fuzion 40-44 but temp mount and only @4.0
cant really take fuzion temps because of temp mount and only running @ 4.0
all others @4.2

so this follows this review in terms of order

cheers

mark

[PaulyD]

Excellent work there

So glad I went for the HK 3.0, just waiting for stock off it to come in

[*|T2D|*PopovS]

I'm talking about in Canada. But those prices are pretty much what I listed anyways.

Me too!

GTZ is not cheapest block here.

[WatZp]

Thanks for your great job.

[Golly]

I've run the Koolance 350 and now a HK Cu.

This test follows what i've seen in the two blocks I've owned. The HK beat the K350 by 1-2C even though I added a Koolance NB block to my Classified.

Great work (I already congratulated you over at OCN but double thanks is in order!)

[smanet]

Hello there, nice review...! Your graphs are outstanding.

I have a dubt, since I'm an Eclipse owner... I can't find mine so restrictive, but I'm using the one with larger hole, wich comes standard in the WB. Wich one is in test?

[santiagodraco]

Nice roundup.

Would love to see you redo the GTZ though based on the mounting issue Gabe mentioned. It's a little odd to see it perform as it did compared to other tests where it performs better comparatively.

How many other blocks might have been affected by the lapped processor height effecting block to cpu mounting pressure?

[HESmelaugh]

Hello there, nice review...! Your graphs are outstanding.

I have a dubt, since I'm an Eclipse owner... I can't find mine so restrictive, but I'm using the one with larger hole, wich comes standard in the WB. Wich one is in test?

I also used the midplate with the larger opening. I noticed that my results showed the block to be more restrictive than it was in other test, but it stayed the same after disassembling, checking, cleaning etc.
The Eclipse was one of the blocks with which I did quite a few more testruns than usual but I coulnd't find out why it was more restrictive for me than for others.

Nice roundup.

Would love to see you redo the GTZ though based on the mounting issue Gabe mentioned. It's a little odd to see it perform as it did compared to other tests where it performs better comparatively.

How many other blocks might have been affected by the lapped processor height effecting block to cpu mounting pressure?

I've asked Gabe about this. Apparently, the problem can't be fixed just by adding washers or the like. I'm willing to do a retest, but I want to wait for an "official" solution, otherwise the retest will be pretty much pointless.

I don't think many blocks should have a great disadvantage from being used on a lapped CPU. For me it's just one of the things with testing: I can never cover all the bases and no test result will ever be universally applicable or 100% correct. I just do my best to get results that are helpful to as many people as possible.

In my opinion, it's very valuable and important that there are several people doing these kinds of tests and everyone is using different methods and setups. That way, once you've read up on several different tests, you get a broader picture.

[santiagodraco]

I also used the midplate with the larger opening. I noticed that my results showed the block to be more restrictive than it was in other test, but it stayed the same after disassembling, checking, cleaning etc.
The Eclipse was one of the blocks with which I did quite a few more testruns than usual but I coulnd't find out why it was more restrictive for me than for others.




I've asked Gabe about this. Apparently, the problem can't be fixed just by adding washers or the like. I'm willing to do a retest, but I want to wait for an "official" solution, otherwise the retest will be pretty much pointless.

I don't think many blocks should have a great disadvantage from being used on a lapped CPU. For me it's just one of the things with testing: I can never cover all the bases and no test result will ever be universally applicable or 100% correct. I just do my best to get results that are helpful to as many people as possible.

In my opinion, it's very valuable and important that there are several people doing these kinds of tests and everyone is using different methods and setups. That way, once you've read up on several different tests, you get a broader picture.

Very true. Of course for the unitiated it can be somewhat confusing, ie what results "really" apply to me. But that aside I have to say I don't know how I missed your thread earlier, but great job and you seem to have covered the bases pretty well.

Another question. How accurately do you think one can apply the results of your test to say a loop with a GPU block, or multiple additional blocks, and expect the "pecking order" to remain the same? Would they? Or might we see the comparative performance of the blocks change by changing the loop configuration and adding the additional heat into the loop?

[Freaky Freezer]

Great tests and presentation as always HES. Thanks for the info.

The difference is due to the fact that Shane uses a lapped processor.

1/ The GTZ was designed to perform best with off-the-shelf (unlapped) processors
2/ More importantly: The GTZ mounting system simply does not work well with lapped processors because if has a fixed travel distance, so as one laps the processor down, the block loses perfect matting with the IHS.

I would love to see this proven or disproven by a future test.

[pingpong1455]

nice review

looks like the heat killer is my block

[ColonelCain]

nice review

looks like the heat killer is my block

If you can find it in stock.

It's been OOS in the US for a long time now... trust me, I know, because this little detail is holding back my new rig.

[DeathWalking]

wow, nice tests
ek supreme is pump killer

EK Supreme...IS SUPREME! FTW!!!
(I'm kidding -- but you should know who I'm referencing, though )

[HESmelaugh]

Another question. How accurately do you think one can apply the results of your test to say a loop with a GPU block, or multiple additional blocks, and expect the "pecking order" to remain the same? Would they? Or might we see the comparative performance of the blocks change by changing the loop configuration and adding the additional heat into the loop?

Adding heat to the loop won't make a difference. Since it's delta temps we're looking at, you wouldn't find a significant change due to different water temperatures.
It's the added restriction of extra blocks that will make a difference. That's where the Temps in relations to flowrates graph can be a useful, albeit not perfect, reference.
Basically, for a restrictive loop, you want a block that isn't very flow-dependant while in a less restrictive loop the block's flow-dependancy is less important.

Though I also want to mention that the difference flowrates make and even the differences from one block to another usually aren't dramatic.


Thanks to everyone for the positive feedback, by the way.

[gabe]

I've asked Gabe about this. Apparently, the problem can't be fixed just by adding washers or the like. I'm willing to do a retest, but I want to wait for an "official" solution, otherwise the retest will be pretty much pointless.

Well it's pretty much official now, see http://www.xtremesystems.org/forums/...&postcount=108

I don't think many blocks should have a great disadvantage from being used on a lapped CPU.

Mounting systems:

I concur. In fact, I would even go further: most of the blocks you have tested wouldn't have a problem. The explanation is simple: except for the GTZ, they use unlimited travel mounting systems. For the record, in my personal opinion and professional experience, I find that to the majority of users these systems are potentially dangerous, not to mention hard to adjust properly. To the small manufacturer, they are the only option available (I know, I've been there) because of the high costs associated with making custom hardware. But what about the larger manufacturers? Oversight, cost cutting measures, lack of understanding? lack of care? For example, I read in another post (http://www.xtremesystems.org/forums/...3&postcount=45) that a leading manufacturer recommends up to 74 lbf mounting pressure for example. This is way over Intel's 60 lbf spec for socket 1366 (see http://en.wikipedia.org/wiki/Socket_1366 and also http://download.intel.com/design/pro...nex/320837.pdf page 24). How responsible is that?

Anyways, back to subject of testing methodology. In an email dated 5/6/09, you wrote to me: "CPU-waterblocks: Yes, all the tests are done on a lapped CPU. I opted for lapping the CPU since it's usually warped in an asymmetrical manner and that could mean an unfair advantage or disadvantage for some blocks."

So I would like to to try to reconcile what you wrote to me with what you are saying in this post. In your email to me, you explain that you didn't want an unfair advantage or disadvantage for some blocks, and in the above post you say "I don't think many blocks should have a great disadvantage". I see a contradiction here. I am just curious if your latest statement reflects a change in opinion on your part lately.

[HESmelaugh]

Thanks for linking, Gabe. I missed that. Did I understand correctly that for a lapped CPU, we'd be using a "normal" screws and thumbnuts mount like with most other blocks? Can I use the GTZ's springs for this?

Concerning the lapping: I actually do mean the same thing with both statements though I guess it's not put very well (keep in mind that english isn't my first language...).
I do think that there is a possibility that an unlapped IHS would cause an unfair disadvantage or even advantage for a particular block. I don't know this for a fact, but here's something I can imagine might happen: Most IHS's I have lapped were convex to one side of the center ie they had a bulge that was off center. If this bulge connects with a bowed block there's a possibility that the block and the CPU will have a very small contact surface and if that contact surface happens to be where the cooling structure is hit with the least amount of flow an turbulence (say, the far side of the outlet on a diagonal-flow block) that would result in what I would imagine to be an unfair disadvantage for that block. An unbowed block would have a larger contact area and if it happens to have it's inlet port right above that surface, that's what I would assume to be an unfair advantage.

This is all theoretical, none of this has been tested. At least not by me.

Anyway, I imagine that lapping the CPU levels the playing field and while some blocks might potentially lose an unfair advantage that they otherwise would have had, none of the blocks should have a real or actual disadvantage on a flat surface.
This all looks more complicated written out than it is in my head, but anyway, I hope you see what I mean.

[gabe]

Thanks for linking, Gabe. I missed that. Did I understand correctly that for a lapped CPU, we'd be using a "normal" screws and thumbnuts mount like with most other blocks? Can I use the GTZ's springs for this?

Yes. M3x35 screw into backplate creating posts, then use thumb nuts and compress the GTZ springs by 6.35mm (.250") for 255N. You need a thumb nut with a wide flange, since our spring is 10.6mm in OD. I am preparing a kit. I had planned on sending you one.

Regarding methodology (lapped IHS or non-lapped IHS), it is evident that the vast majority of people do NOT lap their processor. A set of tests like yours, which is so incredibly comprehensive, automatically gains a massive audience. Because this audience is so broad, it would make sense that the tests be conducted to reflect the majority of use: non-lapped IHS.

[masska]

If you can find it in stock.

It's been OOS in the US for a long time now... trust me, I know, because this little detail is holding back my new rig.

here you go. if you cant wait for US suppliers, Highflow have them in stock.
Ordered mine a couple of days ago
http://www.highflow.nl/water-blocks/...3-0-s1366.html

[santiagodraco]

...
Mounting systems:

I concur. In fact, I would even go further: most of the blocks you have tested wouldn't have a problem. The explanation is simple: except for the GTZ, they use unlimited travel mounting systems. For the record, in my personal opinion and professional experience, I find that to the majority of users these systems are potentially dangerous, not to mention hard to adjust properly. To the small manufacturer, they are the only option available (I know, I've been there) because of the high costs associated with making custom hardware. But what about the larger manufacturers? Oversight, cost cutting measures, lack of understanding? lack of care? For example, I read in another post (http://www.xtremesystems.org/forums/...3&postcount=45) that a leading manufacturer recommends up to 74 lbf mounting pressure for example. This is way over Intel's 60 lbf spec for socket 1366 (see http://en.wikipedia.org/wiki/Socket_1366 and also http://download.intel.com/design/pro...nex/320837.pdf page 24). How responsible is that?

...

[Boogerlad]

i want a sapphire. (enzotech) It's performance is only 1. degree worse than the hk and is the most freeflowing of them all.

[gabe]

i want a sapphire.

the name alone is superb