New CPU Waterblock from Aquacomputer

[Church]

.. As said a slightly different variant for larger fittings will follow.And I still think using such large screws could get you into new trouble. Make sure that you can screw them completely into the bolts since they do not offer much space for larger screws.If you use the block without the rubber plate you will lose cooling performance. The rubber plate is necessary to get a proper pre-load. ..

Hmm, when reading this, it seems that possibly short screws included are intentionally, as in to get specific mounting strength when screwing them fully in. I recall reading few times before about importance of mounting strength for water block performance, skinnee - maybe you guys should take a note of this, when testing this block, not just replace with easier to use longer screws ..

[Nickel020]

Well if its, then it is still bad design since it makes no sense and is unnecessary. Same thing goes for the rubber plate being needed for optimal perfromance as Shoggy said, which makes even less sense. The cooler itself pulls at the nuts, there's no reason why the combination between nuts backplate and screws needs to be as tight as testers say it is. Just take a look at the Megahalems mounting system, where the backplate and mounting bracket fit very loosely - a tight fit as with AC's system just isn't needed, the only thing it achieves it making the block harder to mount.

@Shoggy
You guys still made some new G1/8 stuff a year ago and still have a lot of products that are G1/8 (only). That's why I said you're slowly moving away from it, you're almost there, but not yet fully. E.g. if I want board cooling for a RIIE, I will need to use G1/8...

It also might fit the 1/2" OD compressions you guys are selling, but Vapor said that it doesn't fit the commonly used BP/KL 1/2" OD compressions:

EDIT: Those wondering about the compression fittings: it doesn't fit any of the ones I have here (smallest I have are BP and KL 3/8-1/2 and it doesn't fit those)

[Church]

Well if its, then it is still bad design since it makes no sense and is unnecessary.

BTW, by your very same Apogee XT example, changing inlet placement wasn't that unimportant as you've told. If with inlets closer it had flow restriction less then GTZ, it was more restrictive with inlets more apart. And judging by so far seen tests this new block especially shines in flow area - while getting thermal performance among top ones, it manages to show much better flow then most of them by far. So - make it compatible with with all fittings, and rob performance? Imho most users would take another choice, better performance + one 45deg fitting. So i'll still disagree on bad design and about unnecessaryesness and senslessness bits aswell.

[Nickel020]

Inlet placement has a slight impact on flow, it would be interesting to test to see if the slightly lower performance of the alternate Apogee XT configuration is only due to the lower flow or if the turbulence in the water resulting from an angled inlet patch has any negative impact on performance. It should be easy to test with the Apogee XT (test alternate inlet then test normal inlet and lower the flow rate to the alternate inlet flow rate with a valve). The difference is probably so low though that we can't really measure it. Since AC is apparently working on a version for large compression fittings as well, we'll soon know if it really makes a noticeable difference...

I also don't know where you got the idea that this block has a low pressure drop, because it certainly doesn't:
http://www.awardfabrik.de/kuhlung-wa...x-kryos-6.html

And for the record, my post above is obviously about the mounting system, not sure why you quote me as if I were talking about the inlet placement...

[Church]

hmm, yup, my mistake, why did i thought that mehr is "less" (kicking case for not using some online translator to manually translate text on graphs aswell) :/
But even so, in your apogee xt link on swiftech page flow graph (see "Pressure Drop vs. Flow Rate") it got worse for about 0.5-1gpm (1.5 times worse). Imho noticeable enough (and even more so in single loop scenarios with other waterblocks in loop).
EDIT
damn, i should be more attentive at even reading simple things like graphs .. it's only 1.2 times worse, for about .25gpm flow drop.

[Waterlogged]

I'll stick with my FuZion V2's thank you.

[Nickel020]

In Vapor's testing the alternate inlet drops the flow from 1.38 GPM to 1.27GPM with a DDC 3.2 and thus slightly increases temps. While measurable, it's not noticeable. So if they release a version of the Kryos with further apart connectors and you want to use large tubing you might as well go for the "worse" version. 1/2" ID vs. 3/8" ID tubing might even make more of a flow difference than the inlet location...

[Church]

Nickel020: but current version one +45deg CF rotary with 1/2" ID will show even better results then, no?

[Nickel020]

Yes, but my point was that you won't notice the difference anyway, it's measurable, but not noticeable. I also think that you'll need two 45° rotaries.

[NaeKuh]

In Vapor's testing the alternate inlet drops the flow from 1.38 GPM to 1.27GPM with a DDC 3.2 and thus slightly increases temps. While measurable, it's not noticeable. So if they release a version of the Kryos with further apart connectors and you want to use large tubing you might as well go for the "worse" version. 1/2" ID vs. 3/8" ID tubing might even make more of a flow difference than the inlet location...

were not looking at the same base.
Also the design is not the same... wash area is different.

Nickel020: but current version one +45deg CF rotary with 1/2" ID will show even better results then, no?

No... why would u get better results with a 45 at the inlet?

Yes, but my point was that you won't notice the difference anyway, it's measurable, but not noticeable. I also think that you'll need two 45° rotaries.

Depends on the design.

the plate itself on the kryos is limited so you cant rotate it 90 degree's.

[NaeKuh]

Update... gonna have some numbers posted within an hour....

Cross comparision with a KL-360 only.... gonna be a very basic test, to give you just a simple grasp of where the block would rank @.

Update:1

Ok... so mounting the Kryos was painful.
The original stock mounts would not work properly... however i got an update from AC saying they already fixed that.

Also, the mickey mouse mounts which i had to use could not provide enough block pressure, so im sure it can do better then this.


Once again... i'll would honestly wait for vapor's real test as he probably figured out a better way to mount the block then i did.

Anyhow, One very very big problem i found is if u use 3/4OD tubing:


It was very difficult to get the koolance clamps over the tubing with that much gap. AS u can see it was Painful....

SO here are the numbers:
Ambient was recorded @ 24C
Both had a coolant temp of 25.7C under load.

Single CPU Loop:
Pump: Dual DDC-2 in Single Top.
Rad: PA120.3 Original
Tubing: Feser 1/2ID 3/4OD

As i said it was very tiny... and with better mounts, i think the kryos would probably be tiny better.

KL-360:


Kryos:


Last words:
IF I HAD BETTER MOUNTS FOR THE KRYOS, I WOULD ASSUME IT WOULD DO BETTER THEN THE 360 (Marginally) It would be very very close.
This block is definitely classed in the upper tier tho... so id say within the top 5.

Thank you for your time.

I may redue this... i need to find time to goto the hardware store for different mounts.

[Nickel020]

Thanks for doing this, but what made you use Intel burn test? Results are useless, since the CPU temp is all over the place in that test, use something like Core Damage or Prime in-place 8k-8k FFTs where the CPU temp pretty much stays constant after a small heating up period.

[Waterlogged]

<snip>



<snip>



<snip>

OK, who are you and what have you done with the real NaeKuh? There's no blinding flash in those pics so you can't be the real one.

[bomberboysk]

Thanks for doing this, but what made you use Intel burn test? Results are useless, since the CPU temp is all over the place in that test, use something like Core Damage or Prime in-place 8k-8k FFTs where the CPU temp pretty much stays constant after a small heating up period.

When running test for max temperatures i prefer to run a few hours of LinX with realtemp running to record the max temp.

[Nickel020]

That's good for you, but that still doesn't make the results repeatable/meaningful. Such results are much more unreliable than logged average core temps.

[bomberboysk]

That's good for you, but that still doesn't make the results repeatable/meaningful. Such results are much more unreliable than logged average core temps.

Results from LinX are repeatable, granted you are utilizing the same problem size between tests. And as i said, max temps, not average.

[NaeKuh]

Thanks for doing this, but what made you use Intel burn test? Results are useless, since the CPU temp is all over the place in that test, use something like Core Damage or Prime in-place 8k-8k FFTs where the CPU temp pretty much stays constant after a small heating up period.

im recording max temps... not averages..

i dont have the equiptment sensitive enough for it to be a solid test.

Also i would need my 975 on the bed so i can crank the voltages up to 1.5v.

the only thing i am testing to see what place the block would rank.

And IMO it would fall somewhere between the KL-360 and XT.

It wont beat a HF P1. But it will get dayam close.

[Vapor]

Results off my testbed aren't done rolling in yet (each block goes through 21 2.5hour tests)....but when the dust settles, I'd be very surprised if the Kryos XT with the thicker o-ring isn't the best performing block I've ever tested, beating the Supreme HF P1

[UrbanSmooth]

This is a good-looking block.

[Shoggy]

Say hello to a new variant



Will be available in the next days and costs 74.90 EUR (~77.00 USD w/o VAT).

The middle part has been extended to one side so you can use larger fittings like BP 16/10 now.

A few more pics...

Intel:


AMD:

[wesley]

Say hello to a new variant

Awman, they look really pretty.

Wes

[Church]

Shoggy: will new revision be named somehow to distinct it from old one?

[mk-ultra]

the amd brackets make it look like some kind of bug

[Shoggy]

Shoggy: will new revision be named somehow to distinct it from old one?

Yes, there will be HF hat the end of the name.

[dejanh]

Yes, there will be HF hat the end of the name.

Can you please offer fully nickel plated variant (top and base) of the new revision of the block you just posted above?

Please, please, please