Review Index:
Feedback

CPU Water Block Comparison on Haswell-E

Manufacturer:

Water Block Design

Outer block design

View Full Size

View Full Size

View Full Size

View Full Size

View Full Size

From the top down view of the blocks, you can see that they all share a similar "X" style hold-down layout, matching that of the standard Intel socket design parameters. The big differences come with the block top materials, the thickness of the top hold-down plate, and how the entire block assembly is held together. With the exception of the Koolance CPU-360 block, all blocks are of a mixed metal / acrylic construction. The Koolance CPU-360 block is an all metal construction. The all-metal construction of the CPU-360 makes for a more durable design, especially when inserting and removing barbs from the block. The acrylic tops have more give to them, but their barb holes can become worn over time and lead to cross-threaded holds which don't hold the barbs as securely.

The XSPC Raystorm block is equipped with the thickest hold-down plate, measuring a full 0.25 (1/4) inches thick. Both Koolance blocks come in at a close second with their plates measuring 0.1875 (3/16) inches thick. The two Swiftech block are equipped with a hold-down plate measuring a thickness of 0.625 (1/16) inches. While all hold-down plates are more than up to the task of holding the water block firmly seating to the CPU, the thicker plates are less likely to warp or bend under pressure over time. Hold-down plate warping could lead to non-optimal interface between the block and CPU surface over time, translating to escalating CPU temperatures and instability in overclocking situation.

The hold-down plates for all blocks with the exception of the XSPC Raystorm block are bolted to the bottom plate, intrinsic to the integrity of the block and internal fluid channels. The XSPC Raystorm block's hold down plate is removable, sitting atop a groove along the outside edge of the block's top. The removable hold-down plate of the Raystorm block is a mixed blessing, positive in that it can be easily replaced without taking apart the block if it breaks. In its default configuration, the Raystorm block comes with an acrylic hold-down plate with a metal liner. After repeated mounts and dismounts of the block, the hold-down plate can crack (as happened to me). The metal version of the hold-down plate is much more durable and was easily replaced without much effort because the plate is not bolted into place. However, the hold-down plate can shift much more easily during block board mounting (again, because it is not bolted to the block), making for a trickier CPU block installation process.

Inner block design

View Full Size

View Full Size

View Full Size

View Full Size

View Full Size

When the blocks are taken apart, the design differences between them become more apparent. The internal block design approaches differ in through distinct areas: number of fluid inlets/outlets, positioning of fluid inlets/outlets, and fluid delivery into the block's base plate. With the exception of the Swiftech Apogee HD block, all the blocks have a single inlet port into the block and single outlet port leaving the block. This single inlet/outlet design promotes more directional flow through the block's base plate channels. Swiftech took a different approach with the Apogee HD block, designing it with a single inlet port and up to three outlet ports. The block can be used in the traditional single outlet port configuration. However, it can be used to split the outlet flow between three outlet sources, increasing the fluid flow rate through the block by reducing the block's internal resistance. However, flow rate through the side outlet ports may be reduced in comparison to the port in line with the inlet port because coolant will more naturally flow towards the in line outlet port.

Out of the tested blocks, only a single block has its outlet port offset from its inlet port - the Koolance CPU-360 block. The Koolance CPU-360 outlet port is positioned to the right of the inlet port, forcing the fluid to split through the micro-channels to either side of the inlet port and come together via the outlet port. This splitting and reforming action forces the coolant to more optimally flow through the entire surface of the block's base plate but reduces the coolant flow rate through the block. All the other blocks were designed with their outlet ports in line with the inlet ports. The in line port design increases flow rate through the block, but can lead to non-uniform flow rate through the block and base plate channels.

Three of the blocks (the two Koolance blocks and the XSPC Raystorm block) use a jet impingement plate and micro-channel design while the two Swiftech blocks use a straight-thru micro-pin design for fluid heat absorption. The jet impingement plate and micro-channel design affects fluid flow rate through the block adversely, but makes for more even distribute of the coolant medium over the base plate's cooling channels. The straight-thru micro-pin approach preferred by Swiftech allows for high fluid flow through the block, but can lead to "hot zones" within the base plate. These "hot zones" are caused by the fluid tendency to take the shortest path through the base plate if not directed (as in the jet impingement / micro-channel design).

Block fluid channel design

View Full Size

View Full Size

View Full Size

View Full Size

View Full Size

The block's base plate design is critical to the cooling potential of the block. The base plate design criteria include base plate material and cooling channel layout. All of the blocks use copper-based base plates, a material that absorb transfers heat very well between the CPU surface and the coolant flowing through the base plate channels. The two Koolance blocks are nickel-plated copper with the other three blocks being naked copper. The nickel-plating gives the base plate corrosion and scratch resistance, but can lower the heat absorption capabilities of the copper. Untreated copper can become discolored and corrode over time as the top layer of the copy interacts with the atmosphere or liquid coolant.

The tested blocks have two types of channel layout - micro-channel and micro-pin. The micro-channel layout consists of small vertical channels running from to back along the base plate's surface, normally fed liquid via a jet impingement plate to force the coolant evenly across all channels. The micro-pin layout consists of rectangular pins oriented in a front to back uniform layout, resembling a copper forest. The micro-channel design leads to more even distribution of the coolant across the base plate, while the micro-pin design passes the liquid straight through the copper pin forest leading to the formation of "hot zones" on the base plate. These "hot zones" are caused by the tendency of liquid take the shortest route possible when flowing between the inlet and outlet ports. The micro-pin design alleviates this "hot zone" effect to some extent be causing flow turbulence as the liquid passes through the pin "forest".

The Koolance and XSPC water blocks use a jet impingement-fed micro-channel design with the two Swiftech blocks using a micro-pin-based design. The Apogee HD block uses a hybrid approach with micro-channels feeding coolant into the micro-pin forest, as well as micro-channels leading from the forest to the main and auxiliary outlet ports. Both Swiftech blocks as well as the Koolance CPU-380i block have coolant collection channels at the top and bottom of their channels. These collection areas sit over the inlet and main outlet ports for the Swiftech coolers, assisting with the coolant flow through the block. On the Koolance CPU-380i block, the upper collection area sits over the outlet port, forcing the liquid from the lower collection point to flow back to the upper one. This could cause further restrictions to coolant flow through the block as well as dissuade the coolant from traveling to the lower collection point. The Koolance CPU-360 and XSPC Raystorm blocks do not have any defined coolant collection areas, relying on the channels cut into the block top as well as the positioning of the outlet port to define coolant flow through the base plate channels.


September 18, 2015 | 01:36 PM - Posted by Anonymous (not verified)

Thanks for the review.

Here's an unused digital download code for Metal Gear Solid V for whoever gets there first:

MVNA-OXLL-QWBP-AJIK-DPRI-RVMC

September 18, 2015 | 01:46 PM - Posted by bburnham37 (not verified)

Thanks broheim, awesome sauce.

And, thanks for not mentioning the method of redemption in the post since I'm pretty sure that's why it was still active when I redeemed it ;-)

September 18, 2015 | 02:34 PM - Posted by Anonymous (not verified)

No problem man, it expired in two weeks and i didn't want it to go to waste. Cheers.

September 18, 2015 | 01:56 PM - Posted by Anonymous (not verified)

These water blocks always seem way over priced. There really isn't much to them compared to something like a large heat pipe air cooler. Why should these cost so much more?

September 19, 2015 | 02:04 AM - Posted by Anonymous (not verified)

well the copper is expensive. so you pay for that. nickel also expensive. Blending the materials together. meaning a custom metal also adds to cost. Then each one has to be routed by a very expensive multi axis CNC machine
so yeah there isnt much to them.
but the design and building is where its at

September 19, 2015 | 07:07 PM - Posted by Anonymous (not verified)

I still think they are overpriced. I suspect the margin on these are a lot higher than a big heat-pipe based air cooler. I think that this is somewhat based on the idea that it provides better cooling, so it should be more expensive, even if the BOM does not justify the MSRP.

I think another part of the problem is that these are designed much more expensive than they need to be. If you look at an all-in-one block, they mostly don't have the parts used for mounting machined out of the same piece as the block. There isn't too much point to doing this except for looks. The mounting bracket could easily be made out of much cheaper material. The block could also be designed to need a lot less machining. I suspect if you made it as a casting, it could be finished with only a small amount of machining. I have wondered if one of the all-in-one blocks could be modified to be used in a custom loop. These have to be significantly cheaper than the custom loop blocks since they are a lot cheaper than buying all of the components for a custom loop. They may be designed in such a way that they cannot be easily used with the pump removed.

September 19, 2015 | 07:17 PM - Posted by Anonymous (not verified)

I see they have some built economically (copper base, sheet metal mounting brackets, and plastic top) for only about $20. It would be interesting to see how these cheap blocks compare to the really expensive models tested here.

September 19, 2015 | 07:07 PM - Posted by Anonymous (not verified)

I still think they are overpriced. I suspect the margin on these are a lot higher than a big heat-pipe based air cooler. I think that this is somewhat based on the idea that it provides better cooling, so it should be more expensive, even if the BOM does not justify the MSRP.

I think another part of the problem is that these are designed much more expensive than they need to be. If you look at an all-in-one block, they mostly don't have the parts used for mounting machined out of the same piece as the block. There isn't too much point to doing this except for looks. The mounting bracket could easily be made out of much cheaper material. The block could also be designed to need a lot less machining. I suspect if you made it as a casting, it could be finished with only a small amount of machining. I have wondered if one of the all-in-one blocks could be modified to be used in a custom loop. These have to be significantly cheaper than the custom loop blocks since they are a lot cheaper than buying all of the components for a custom loop. They may be designed in such a way that they cannot be easily used with the pump removed.

September 19, 2015 | 07:07 PM - Posted by Anonymous (not verified)

I still think they are overpriced. I suspect the margin on these are a lot higher than a big heat-pipe based air cooler. I think that this is somewhat based on the idea that it provides better cooling, so it should be more expensive, even if the BOM does not justify the MSRP.

I think another part of the problem is that these are designed much more expensive than they need to be. If you look at an all-in-one block, they mostly don't have the parts used for mounting machined out of the same piece as the block. There isn't too much point to doing this except for looks. The mounting bracket could easily be made out of much cheaper material. The block could also be designed to need a lot less machining. I suspect if you made it as a casting, it could be finished with only a small amount of machining. I have wondered if one of the all-in-one blocks could be modified to be used in a custom loop. These have to be significantly cheaper than the custom loop blocks since they are a lot cheaper than buying all of the components for a custom loop. They may be designed in such a way that they cannot be easily used with the pump removed.

September 19, 2015 | 07:18 PM - Posted by Anonymous (not verified)

Why do I get triple post when posting from an iPad sometimes?

September 18, 2015 | 02:02 PM - Posted by H1tman_Actua1

Where is the EKWB EVO?

September 18, 2015 | 02:31 PM - Posted by Chaitanya Shukla

My thoughts exactly, why there is no EKWB CPU Block in comparison?

September 18, 2015 | 02:52 PM - Posted by Morry Teitelman

The review was done with the samples I had on hand.  The Koolance and XSPC blocks were review samples provided by the manufacturers a while back.  The swiftech blocks were from my personal collection that I have used in my own systems.  EK did not provide us with any samples...

September 18, 2015 | 02:25 PM - Posted by 4TwizZzle (not verified)

It's highly unlikely that there is any bit of laminar flow or promotion of a relevant laminar boundary layer on any of these blocks within the channels. Ideally you want a slight bit of turbulent flow near the surface area to extract the heat more rapidly.

Laminar flow has a zero velocity on the surface which ideally limits heat transfer and would be a poor design. In essence, Laminar flow would be a bad design choice if it was attainable.

Your argument for pressure drops due to an excess of turbulent flow is however a valid point and would restrict flow. I do believe that material choice and oxidation is the main culprit to the difference in the Delta T's between the limited sample size.

I really enjoyed this article, thanks!

September 18, 2015 | 06:55 PM - Posted by KingKookaluke (not verified)

Once again outstanding post by Morry! Makes me consider going back to running my own water cooling loops again.

September 18, 2015 | 08:04 PM - Posted by Branthog

Man, I've wanted to get in on water cooling for so long, but just can't justify the cost and hassle. I don't need to super-overclock my systems and water cooling doesn't make for a quiet or dust-free system, so the only benefit I'd really get from it is the fun of doing it and having it.

Which is enough when you're young and have all the time in the world, but hard to justify when you're old and busy. :D

September 19, 2015 | 03:00 AM - Posted by Tim Verry

If you are interested in getting into it without ponying up the full cost, keep an eye out on forums like [H] and PC Perspective's for users selling off their old water cooling gear. I was able to get everything I needed minus a silver killcoil and distilled water for less than $300, would have been a lot more retail. The best part is that you can usually get help putting it together at the same forums :).

September 18, 2015 | 08:12 PM - Posted by quest4glory

Morry,

Why do you say the CPU integrated graphics processor has been disabled when you're testing Haswell-E (5960X) here?

September 18, 2015 | 11:38 PM - Posted by Morry Teitelman

Good point, overlooked in the review. Thanks for pointing it out, fixed...

September 18, 2015 | 08:50 PM - Posted by Master Chen (not verified)

Only GODLIKE Swiftech, only hardcore!

September 19, 2015 | 02:06 AM - Posted by Anonymous (not verified)

To Be Honest I don't really see Pc Per as a over clocker/water cooling type site.
The impression I get about the site is. average looking aggregate/ review site, just like pretty much every other site abt this.

and this ad looks quite out of place here.

My 2 Pesos That's all.

September 19, 2015 | 11:31 PM - Posted by quest4glory

Does a site need to be dedicated to overclocking / water cooling to therefore be qualified to provide a review of water blocks?

I would have to save no to that.

This is not your "average looking aggregate / review site." Yes they aggregate links to content from the best of the PC enthusiast websites, but that's a service and convenience to the readers.

I am in no way affiliated with PC Perspective. I simply value the additional "perspective" they bring to all things involved with the personal computer enthusiast and consumer space.

September 24, 2015 | 02:22 AM - Posted by Dr_b_

The site can review water blocks for sure, but if you are only comparing it to a few other WBs, and not including one of the top selling WBs (EKWB), then making any conclusion about the winner in this test is of limited value, unless of course your only options are are the ones reviewed here. The problem with the limited review size is that its not inclusive enough, and not as thorough as other sites (extreme rigs)who do a lot more tests on liquid cooling parts.

While nice to see something different here, don't make buying decisions about a water block based on this review alone, because the gold winners here didn't perform as well as the EK blocks on other review sites.

Also, if you are liquid cooling your PC you are likely going to liquid cool your graphics card and maybe get a block for your motherboard as well, and likely those will all be from the same brand.

Here's my all EK block loop on Z97:
http://m34t.net/system1.jpg

September 19, 2015 | 07:23 AM - Posted by Rustknuckle (not verified)

Morry did you test if turning the Koolance 380i mounting 90 degrees yields lower temps like it does when using it with Ivy Bridge/Sandy Bridge E CPUs?

September 19, 2015 | 02:16 PM - Posted by Morry Teitelman

No, we did not test in that config unfortunately. The 380 block was oriented so that the micro-channels ran perpendicular to the core.

September 19, 2015 | 09:22 PM - Posted by Rustknuckle (not verified)

Here you can see the difference rotating 90 degrees can make with that block and a Sandy Bridge E CPU: http://i.imgur.com/9OSXVwK.png

September 19, 2015 | 11:01 AM - Posted by XSPSI (not verified)

No air cooler as a base test?

September 19, 2015 | 02:18 PM - Posted by Morry Teitelman

No, but you can check out other reviews for air-based temperature comparison since we include the XSPC kit as a comparison in those reviews.  This review was more to compare the performance of the various water blocks only without introducing other variables into the review.

Post new comment

The content of this field is kept private and will not be shown publicly.
  • Lines and paragraphs break automatically.
  • Allowed HTML tags: <a> <em> <strong> <cite> <code> <ul> <ol> <li> <dl> <dt> <dd> <blockquote><p><br>
  • Web page addresses and e-mail addresses turn into links automatically.

More information about formatting options

CAPTCHA
This question is for testing whether you are a human visitor and to prevent automated spam submissions.