Stren's 2012 CPU Water Block Roundup

[stren]

Look like alphacool will be making changes to the design:

We respect the high attention that is given to the waterblock in the forums, and we have taken a deeper look into the concerns of the “OUT”-thread which look like “half-cut”. The “OUT”-thread is designed 100% in accordance to the German Industial Standard (DIN), and we can guarantee that there are no problems with fittings also if they that have a 7mm long thread.
Nevertheless, in respect to the discussion and the concerns of the forum users we have decided to change the top of the waterblock so that the thread will be best for fittings with 5mm long threads. This design-change will be made with next production series.

[rge]

If I used a 7 mm long threaded fitting then even more reason to have threads for 360 degrees, since more of that fitting would then be sticking through hole, and no difference in having threaded acetal there as well.

I have no idea why even bring up 7 vs 5 mm, fact is threads should be 360 degrees for a minimum of at least shortest barb length, since barb will be blocking at least that much of the opening. To do anything else defies common sense in addition to being poor engineering. I would like to see alphacools reference stating 180 degrees of threads with only 1 mm of material on remaining 180 degrees is acceptable.

Glad they are changing it, but wasnt really clear how from what they said. But if the do decide to appropriately put 360 degrees of threads, hope they clearly mark the ones not defective.

[stren]

If I used a 7 mm long threaded fitting then even more reason to have threads for 360 degrees, since 5mm of fitting would then be sticking through, hence should be 5mm of threads for 360 degrees.

I have no idea why even bring up 7 vs 5 mm, fact is threads should be 360 degrees for a minimum of at least shortest barb length, since barb will be blocking at least that much of the opening. To do anything else defies common sense in addition to poor engineering. And chance of that passing any engineering standard is zero, but lets see alphacools reference stating 180 degrees of threads with only 1 mm of material on other 50% is acceptable.

Glad they are changing it, but wasnt really clear how from what they said. But if the do decide to appropriately to 360 threads, hope they clearly mark the ones not defective.

Yeah I took it to mean they would remove the 7mm long section on one half and make it 360 degrees for 5mm. Like you said it's not 100% clear though.

[stren]

Update

- Sounds like alphacool will be sending the full metal version for testing. I've requested that it have the design changes made to it so that everyone can see exactly what they'll get
- I've also made a facebook page. If you want to keep up to date with what I'm testing then feel free to like away:

https://www.facebook.com/ExtremeRigs

Also if anyone has a differential manometer for sale - please let me know!

[Martinm210]

I'd sell or give you mine, but I do have one collecting sickness and that is tools collecting...can't seem to let go. Even if I don't use it for watercooling, I may need it to calculate some HVAC velocities or for sync on some carburetors or...

Here you go..$135 for a new 475-5.
http://www.dwyer-inst.com/Product/Pr...es475/Ordering

You have to be careful to avoid water getting to the meter, but I actually prefer using the air only manometer as the air in the lines works as suspension in smoothing out the bumps in the pressure readings. I think Cam went with the hydro version so he could allow water to reach the meter without worry. I bought mine used on ebay for under $100 and went with the 477-5, they are $270 new:
http://www.dwyer-inst.com/Product/Pr...es477/Ordering

Or watch ebay and wait for a good deal. The 0-200" models will work fine for blocks and single pumps, but you'll need the 0-30PSI version for multiple pumps or really high end pumps like the RD-30 which is capable of around 24psi shutoff head. Anyhow, I've always admired the Dwyer's been really good reliable units for me even though I haven't taken very good care of them.

[stren]

Thanks again Martin. I understand hoarding tools - I do the same thing, I have some fairly obscure tools I only used once but refuse to get rid of. Interesting I'd only researched the price on the 477 because that's what you had used. I hadn't even realized that it needed to be kept dry so now I feel like a noob. It seems the 477 is just a more accurate version of the 475? I'd been keeping an eye on ebay but never saw one with the right range! If I don't see one soon then I'll settle for the 475 as that's much more affordable! With 0.5% accuracy and 30psi range it would give me 0.15psi accuracy which seems still "good enough" for flow curves on blocks while still being able to handle the Iwaki.

While you're here any recommendations on shutoff valves for testing the pumps? After the cpu block review is completed I was going to run through all my pumps (rd30, mcp35x2, pmp450s and pmp500) and make flow curves at max power. This would be to review the pmp500 while also giving some credence to the data by seeing if it matches yours on the other pumps.

[Martinm210]

I thought the 477 was just more features with the electronic zero button and backlighting. My older 475 I had used a metal knob on top to zero but the newer ones I am not sure. I would have kept my old 475 if it had been a mark 5, but it was a mark 3 with only 0-200inch range...good enough for blocks and smLl pumps but a bit short on range for larger pump testing. 475 is a great unit too though..it was a huge step up from my water tube manometer.

You will like pressure drop and pump testing, it is fast and pretty straight forward. Regarding the dry/wet..I just make sure the meter is upright and watch the water movement in the lines. Pressure is pressure no matter if its air or water, but I found having the compressible air in the lines does help smooth the readings and I could generally plot a smoother pressure drop or pump curve than I could with a hydro manometer, but either option should work the same..always just used the air type myself.

For valves, I liked using my valved king flow meter the best for simplifying the setup and the control on the valve is nice, however it is also fairly high in restriction which does limit data gathering a bit nearer the zero pressure line. I have also used brass globe (ball) valves which are the least restrictive but the lack of precise control makes those very frustrating to use and I also have a brass gate valve that worked pretty good with fairly good control and not too much restriction...it does trap air a bit but probably the tool I would suggest if you already have a flow meter.

Here is an older picture showing the gate valve setup.
http://martinsliquidlab.petrastech.c...-CPX-Pro-5.jpg

I just put the valve and flow meter on the pump outlet side after the manomter posotive pressure Tee. The large custom 4" pvc reservoir is really nice for pump testing too, bleeds almost instantly and very low in restriction which again help get you more points near the zero pressure line which improves/completes the curve a bit.

I also made my own 1/2" copper x 1/8" brass tee to reduce restriction and ensure I wasn't getting any odd turbulence issues in the Tee fittings. In theory any losses should cancel each other out using a differential manometer, but I liked my little custom T fittings..it was worrh the effort fabricating/soldering.

Regardless..there could always be calibration issues with my or anyones tools so I wouldn't worry too much about matching other bench data..more importantly you can repeat your own results to ensure relative accuracy is good...if you want to be really tidy use a baseline pump or block and use it as a data check in check out tool between batches of testing. I did that for fan testing because I didn't trust my anemometer calibration but I never saw any obvious issues with pump or pressure drop testing for the most part. You do need to measure voltage at the plug and constantly have to dial in 12.00v as you adjust restriction. Alsomake sure the loop is completly air free with no bubbles on the flow meter float. Other than that pretty consistent and repeatable testing....easy compared to thermal testing which is much more difficult in our variable ambient environments.

[stren]

I thought the 477 was just more features with the electronic zero button and backlighting. My older 475 I had used a metal knob on top to zero but the newer ones I am not sure. I would have kept my old 475 if it had been a mark 5, but it was a mark 3 with only 0-200inch range...good enough for blocks and smLl pumps but a bit short on range for larger pump testing. 475 is a great unit too though..it was a huge step up from my water tube manometer.

You will like pressure drop and pump testing, it is fast and pretty straight forward. Regarding the dry/wet..I just make sure the meter is upright and watch the water movement in the lines. Pressure is pressure no matter if its air or water, but I found having the compressible air in the lines does help smooth the readings and I could generally plot a smoother pressure drop or pump curve than I could with a hydro manometer, but either option should work the same..always just used the air type myself.

For valves, I liked using my valved king flow meter the best for simplifying the setup and the control on the valve is nice, however it is also fairly high in restriction which does limit data gathering a bit nearer the zero pressure line. I have also used brass globe (ball) valves which are the least restrictive but the lack of precise control makes those very frustrating to use and I also have a brass gate valve that worked pretty good with fairly good control and not too much restriction...it does trap air a bit but probably the tool I would suggest if you already have a flow meter.

Here is an older picture showing the gate valve setup.
http://martinsliquidlab.petrastech.c...-CPX-Pro-5.jpg

I just put the valve and flow meter on the pump outlet side after the manomter posotive pressure Tee. The large custom 4" pvc reservoir is really nice for pump testing too, bleeds almost instantly and very low in restriction which again help get you more points near the zero pressure line which improves/completes the curve a bit.

I also made my own 1/2" copper x 1/8" brass tee to reduce restriction and ensure I wasn't getting any odd turbulence issues in the Tee fittings. In theory any losses should cancel each other out using a differential manometer, but I liked my little custom T fittings..it was worrh the effort fabricating/soldering.

Regardless..there could always be calibration issues with my or anyones tools so I wouldn't worry too much about matching other bench data..more importantly you can repeat your own results to ensure relative accuracy is good...if you want to be really tidy use a baseline pump or block and use it as a data check in check out tool between batches of testing. I did that for fan testing because I didn't trust my anemometer calibration but I never saw any obvious issues with pump or pressure drop testing for the most part. You do need to measure voltage at the plug and constantly have to dial in 12.00v as you adjust restriction. Alsomake sure the loop is completly air free with no bubbles on the flow meter float. Other than that pretty consistent and repeatable testing....easy compared to thermal testing which is much more difficult in our variable ambient environments.

Thanks Martin - I assume those gate valves are standard issue at any hardware store?

For the T fitting - Yes I may have to do the same as I can't find any barbs small enough although possibly 3/16 tubing could be stretched onto 1/4" barb. It's a hassle though because I don't have the tools to make one of those yet either lol so I may try the 1/4" fitting and see how it goes. Lots to do!

[Martinm210]

Yes, gate valve should be available at most home improvment centers. Mine was a brass with 3/4MIP/NPT female threads and I used copper barbs, but you buy 3/4MIP to 3/4" Hosebarbs in brass or nylon. I have also seen some with the hosebarbs made into the casting. They cost around $10 plus a few for the barbs.

[Lutfi]

subbed to this thread - can't wait to find out how things will unfold for the alpahcool

@ stren+martinm210 - would you be able to recommend a block...

[babalouj]

Can you explain why the Raystorm didn't get recommended? It seems to perform very well (near the top) with only the mounting variation being a little high. I own one and I love it so I am probably a little biased. I know the mounting isn't as great as the Swiftech blocks I have owned but I feel like the Raystorm makes up for it in performance.

Anyways, Great roundup Stren! This must have been/continues to be a massive effort for you.

[stren]

subbed to this thread - can't wait to find out how things will unfold for the alpahcool

@ stren+martinm210 - would you be able to recommend a block...

I would choose one of the silver awards depending on your situation and what you like the look of:
- The MIPS block performs well and is very unrestrictive making it a great choice all around for a multiblock/rad loop
- The DT Sniper performs very well at high flows so choose this if you know you have tons of flow e.g. Iwaki or cpu only loop with a dual pump
- The Koolance CPU-370 if you know your flow is lower than a mcp35x with cpu only loop, e.g. single pump + multiple blocks/rads or
- The Apogee HD if you like a super easy mount and aren't worried about restriction

All those blocks are so close that you have to choose based on secondary characteristics and aesthetics. I don't believe any of those are bad choices though.

Can you explain why the Raystorm didn't get recommended? It seems to perform very well (near the top) with only the mounting variation being a little high. I own one and I love it so I am probably a little biased. I know the mounting isn't as great as the Swiftech blocks I have owned but I feel like the Raystorm makes up for it in performance.

Anyways, Great roundup Stren! This must have been/continues to be a massive effort for you.

The raystorm is a good block but I can't give an award to every block. The raystorm performs well but it's mounting system gives more variable results so why not choose one of the others? The alphacool was only given gold because it was the best on every thermal test despite it's poor mount variability. If you have a raystorm I wouldn't feel bad about the choice. The results are so close that effectively what the test says is, apart from the worst performers it's actually hard to choose.

[Lutfi]

hmm thanks for that really helpful advice. Decided to move away form getting a Supremacy plexi copper and onto a white swiftech Apogee HD.

[stren]

I think you forgot to include the sample size when calculating the confidence intervals. Confidence intervals are calculated from the standard error. The formula is

X +- 1.96*s / sqrt(n)

where n is the sample size, X the sample mean, and s the sample standard deviation.
The row should be

DT Sniper 41.40  0.25 0.25

Note: I am ignoring the fact that 5 is a very small sample size (statistically speaking) and thus should probably use the t-distribution, so we should preferably use 2.77 instead of 1.96.

Finally got round to adding the performance intervals - thanks to mdatmo and rrrhal for the info on how to do this. Maybe I made a mistake in not giving the raystorm an award, particularly given the price.



What this does show is just how close things are.

Also took some internal shots of the summit, cpu-370 and sniper. Hopefully I'll get those put up tonight.

Also compiled a price list with links but embedded google docs aren't working for me, any clues? It worked on OCN.

<iframe width='580' height='375' frameborder='0' src='https://docs.google.com/spreadsheet/pub?key=0AhqJsdm9xVf4dFdfSktQaDBpNEtVRVZDc3F4RmlEO EE&output=html&widget=true'></iframe>

[UrbanSmooth]

Thank you for the comprehensive report, Siren.

[stren]

Updated results post (#18) with

- DT Direct
- Swiftech Apogee HD under IX
- EK Supremacy Retest under MX2
- "Mystery Block" under MX2

The "Mystery Block" is a new block that is being released on monday. You may notice that it's the new top performer. Can you guess who it's made by?

I'll release more details of it on Monday including photos!

Don't forget to like my facebook page for exclusive updates and giveaways coming soon!

[stren]

Koolance CPU-380 has been released and is the "mystery" block that tops the charts.

Pics are here:

http://www.xtremesystems.org/forums/...=1#post5123408

Results are still in the same place:

http://www.xtremesystems.org/forums/...=1#post5123416

If you want an individual review (basically the edited combo of those two posts) you'll have to head to my website:

http://extremerigs.blogspot.com/p/ko...ck-review.html

Don't forget that you can follow me at www.facebook.com/extremerigs

[UrbanSmooth]

Glad to see Koolance back at it again. "NEEDS MOAR LEDs," just like the floss on the GTX 480 GPU blocks.

[stren]

Glad to see Koolance back at it again. "NEEDS MOAR LEDs," just like the floss on the GTX 480 GPU blocks.

Lol I had those illuminated Koolance 480 blocks too.

[Six]

Just one funny thing @ overclock.net
I wrote there:

Like you maybe noticed his thread has been closed because
Stren is not able or dont want to pay to overclock.net for his thread.

He had a lot of work with this and i am very interested that he has the possibility to continue his tests and publish the results here.

For that reason i ask you to make a small donation to overclock.net.

I would like to give 5$ to overclock.net for that.

If overclock.net accepts we need Paypal information.

If you are also willing to make a donation please write here.

Btw. Why all my posts get deleted without a comment ? This is a seriously offer.

[stren]

Just one funny thing @ overclock.net
I wrote there:

Like you maybe noticed his thread has been closed because
Stren is not able or dont want to pay to overclock.net for his thread.

He had a lot of work with this and i am very interested that he has the possibility to continue his tests and publish the results here.

For that reason i ask you to make a small donation to overclock.net.

I would like to give 5$ to overclock.net for that.

If overclock.net accepts we need Paypal information.

If you are also willing to make a donation please write here.

Btw. Why all my posts get deleted without a comment ? This is a seriously offer.

Hey Six

Thanks for the offer. It wasn't about the money for me as much as the principle of the thing and the exclusivity issue. Spotswood (and some members also) even offered to pay for the whole fee himself to keep the thread alive. I could afford to pay $29 this time, but what about for the next review and the one after that? Would I end up paying hundred's every year to OCN or begging people/companies to donate just because one forum got greedy?

Bear in mind I already payed them $29 out of my own pocket to post my sponsored build log. Even after I told the mods that spotswood offered to fund the review they made it clear that they couldn't accept the money because the review was cross-posted to different forums. Not only did they want money but they wanted exclusivity. I couldn't post this review on my own website or [H] or XS. In return for this I got nothing back. No front page promotion, nor a special sponsored review sub-forum. Not surprisingly this is the first time they tried to tax a review and the whole policy seems ill thought out.

So there you have it. OCN wants reviewers to pay them to do reviews and they want to own your content. It's a good business model I suppose if there weren't other forums out there who are more reasonable. However it gives me more reason to build a better website and become more independent.

Here's some updated results with Indigo Xtreme:

Since last time I ran the Apogee HD, the Raystorm and the Raystorm Full Copper. The CPU380 and the MIPS IceForce are next in line:





I also plotted the MX2 results relative to the IX results. I'm trying to show how good the mount/bow is relative to a "perfect" mount reflected by IX. I thought this was an interesting way to look at the data:

[Six]

I dont want that you pay anything !
My idea with this was to make people awake and take a look to their behavior and i hope that they can change their mind
and open this thread again.
I only sign in there because your interesting thread. I risk to get banned

[stren]

I dont want that you pay anything !
My idea with this was to make people awake and take a look to their behavior and i hope that they can change their mind
and open this thread again.
I only sign in there because your interesting thread. I risk to get banned

Oh I'd been arguing with them all week about it, there is no way they are changing their mind. Thanks for trying to stick up for me though. The decision was handed down from the very top as a direct violation of the terms of service and it was made clear that that decision would not change under any circumstances. OCN has banned plenty before for silly reasons and gotten away with it. So I suppose at least they didn't ban me too for being a trouble maker and kicking up a fuss about this.

[stren]

CPU-380 IX reflow went well and as predicted it's in the overall lead. Data vs flow/pump settings coming in the next few days.

[UrbanSmooth]

So there you have it. OCN wants reviewers to pay them to do reviews and they want to own your content.

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.