View Full Version : Best Waterblock
EBFoxbat
11-02-02, 08:26 PM
Let's here personal opinions on the best CPU waterblocks out there. Give some justification please. I'm still considering which waterblock to buy, or perhaps making one
i like maskedgeek's block..
very nice design..
does not restrict flow..
nealric
11-02-02, 08:30 PM
Ive been using a maze3 for a while now... So far its been some good/some bad. At first, I had terrible leaking problems, so I had to go way overkill with the sealing (3 hose clamps and half a tube of silicone). It also crushed the core of my tbird 1.4- but after the installation headaches it seems to be performing niceley so im happy overall. One word of advice: get metal mounting hardware, the plastic stuff that usually comes with it is awful.
why do you need 3 hoseclamps?
and the crushing the core part is your fault.. not dd's faults.. :(
Since87
11-02-02, 09:43 PM
Cathar's Concept Waterblock (http://forums.overclockers.com.au/showthread.php?threadid=106836&pagenumber=1)
For justification click the link.
dj_unforgetable
11-02-02, 10:54 PM
I agree with Since87...Cathars block is VERY good...in design, creativity, temps...the works...
you do realize that not everybody is a member there..
Penguin4x4
11-02-02, 11:50 PM
http://forum.oc-forums.com/vb/showthread.php?s=&threadid=123929&perpage=30&highlight=Micro
:D
mulletron
11-03-02, 12:14 AM
leufen sold a direct die block that worked well-no probs.
now its on a 220w tec-xp1800 @ 1935. sites been down awhile...
mulletron
11-03-02, 12:17 AM
leufen sold a direct die block that worked well-no probs.
now its on a 220w tec-xp1800 @ 1935. sites been down awhile...
oops, i've been gone-new sig
Since87
11-03-02, 07:55 AM
Originally posted by Liss
you do realize that not everybody is a member there..
So, what's the big deal? Sign up for a membership. Some of the most creative stuff in cooling is going on there.
Originally posted by Since87
So, what's the big deal? Sign up for a membership. Some of the most creative stuff in cooling is going on there.
LAZY
nikhsub1
11-03-02, 11:07 AM
Originally posted by Since87
So, what's the big deal? Sign up for a membership. Some of the most creative stuff in cooling is going on there.
Very, VERY true. And yes, I believe Cathar's block will spank anything out there right now. I know that in his and other people's tests, it outperforms EVERY Cyclone block and the Maze 3 by 5 - 6C. I have 2 on order...
masitti
11-03-02, 11:11 AM
Running a Dtek Spir@l in 1/2" setup, I've had nothing but good things with this block. :)
Tecumseh
11-03-02, 11:11 AM
From a design viewpoint I like Cathar's WB. It's my kind of block. :D
Originally posted by nikhsub1
Very, VERY true. And yes, I believe Cathar's block will spank anything out there right now. I know that in his and other people's tests, it outperforms EVERY Cyclone block and the Maze 3 by 5 - 6C. I have 2 on order...
Wow, I was thinking most every block out there performed basically the same, but 5-6C better is actually significant. That's a nice block! Please please please someone test one against maskedGeek's s-type block!
JFettig
11-04-02, 03:02 PM
i dont even have my blocks yet, in 2 days i should... ill prob test it, but most likely itll woop it, just because its the best design...
nikhsub1
11-04-02, 03:06 PM
Originally posted by JML
Wow, I was thinking most every block out there performed basically the same, but 5-6C better is actually significant. That's a nice block! Please please please someone test one against maskedGeek's s-type block!
Well I have 2 of Cathar's blocks coming, could be here this week. If Masked wants to send me an S Type for testing, I will...
JFettig
11-04-02, 03:14 PM
i do know that the microfin block will beat it, unless theres some odd mirical.. ;)
YGPM nik
The Overclocker
11-04-02, 03:21 PM
Originally posted by nikhsub1
Very, VERY true. And yes, I believe Cathar's block will spank anything out there right now. I know that in his and other people's tests, it outperforms EVERY Cyclone block and the Maze 3 by 5 - 6C. I have 2 on order...
hmm, its closer to 3C (see here: http://forum.oc-forums.com/vb/showthread.php?s=&threadid=123929&perpage=30&highlight=Micro&pagenumber=2)
anyway, it is still very good, and shows water a good design can achive (instead of a monkey with a cnc following the same cross drill pattern)
i think the best waterblock has yet to be found, watercooling has really just started, we just need lots of crazy ideas so we can find a good one.
Silence
11-04-02, 03:26 PM
I would like to see how morphling's block stacks up, however since he lives in slovenia it would be hard to do comparisons. However I'm not sure if it really is worth the trouble just you get a little cooler.
JFettig
11-04-02, 03:33 PM
honestly I beleve the best waterblock is a direct die waterblock.... thats as good as it can get...
nikhsub1
11-04-02, 03:50 PM
Originally posted by The Overclocker
hmm, its closer to 3C...
No he has refined it further, and temps have gone down from that post, you must read this http://forums.overclockers.com.au/showthread.php?s=&threadid=95955&perpage=15&pagenumber=1 Now this 5 - 6C is on BARE CORE CPU's, not P4's with heatspreaders.
I'd vote Swiftech's MCW462 blocks... i have seen some really amazing overclocks with them (esp. with a TEC)!
nikhsub1
11-04-02, 05:48 PM
Originally posted by Stalka
I'd vote Swiftech's MCW462 blocks... i have seen some really amazing overclocks with them (esp. with a TEC)!
Votes don't count, only numbers:D TEC's are a different issue altogether. The Cyclone WB's and the Swifty perform very similarly with the Cyclone's edging the Swifty, and well, they are still 4 - 6C short of Cathar's design.
masked: make a microfin s block
JFettig
11-04-02, 09:56 PM
LOL that would be cool, but very very very, did i say very? difficult! cost like 200 dollars to just get machined IF they would do it..
can't do you it yourself?
Cathar's design seems more geared toward amd cpu's, where as masked's block will prob be more versatile. Testing on P4's and pelts would be nice to see on both of these.
NeoMoses
11-04-02, 10:55 PM
you ever used a 1mm endmill? those babies snap easier than a toothpick. I'll bet it would take at least 4 hours to cut those micro channels. when endmilling, a good rule of thumb is to cut no deeper than 1/2 your endmill diameter. So, for 5mm channels, that's 10 passes each! put 8 channels in a block and that's easily 80 passes!!!
Albigger
11-04-02, 11:29 PM
Originally posted by NeoMoses
you ever used a 1mm endmill? those babies snap easier than a toothpick. I'll bet it would take at least 4 hours to cut those micro channels. when endmilling, a good rule of thumb is to cut no deeper than 1/2 your endmill diameter. So, for 5mm channels, that's 10 passes each! put 8 channels in a block and that's easily 80 passes!!!
yes i know from experience milling channels that small would suck. that's why i went with 1/8" in mine, and that still was very hard. It wouldn't be much work to program in and NC machine though, but its probably not something you would want to leave unattended as the chance of breaking a cutter is fairly high...
and whoever said about a pelt comparison of masked/cathar's I would definately like to see that as well
ssjwizard
11-04-02, 11:41 PM
on a cnc 80 passes arent that many but the care for passes so close togeather and on cu sux
Originally posted by JML
Cathar's design seems more geared toward amd cpu's, where as masked's block will prob be more versatile. Testing on P4's and pelts would be nice to see on both of these.
Actually the block is geared towards cooling CPU heat dies of 15x15mm and smaller.
The Intel P4 CPU is 11.3 x 11.3mm. It has an IHS on top, however the IHS does very little to spread the heat when there's a good waterblock on top. >95% of the heat is still focussed in a 15x15mm area with the IHS on top of the P4 with the waterblock on it.
Due to the IHS, results for the P4 are not as impressive as for the bare die CPU's, however my own internal testing has shown improvements in the order of 3C+ over the other blocks I own (Maze 3, Cyclone 3, 4 & 5). I specifically blame the IHS for this with its second thermal interface, and Volenti's testing with and without the P4 IHS would appear to back up this theory. Volenti has his own version of my block and found that direct die cooling, and his semi-copy of my block (my production block does a few things better than his) were producing almost identical results.
Of course people may say my own testing is subject to suspicion, however I do have customers of my block seeing results very similar to mine. (+/- 1C), and I trust that when the first P4 blocks start shipping tomorrow that customers will again confirm my own results for the P4's. My own P4 doesn't overclock that high, so I'm even hoping that some customers will see better results that what I'm seeing on my P4.
My block is absolutely not geared towards peltier use. I am in the process of designing a peltier only block.
It's simply impossible to design a block that is optimised for both peltier (large die) and CPU (small die) use. If you want the best for both, you need to focus on the separate needs of each problem (and they are vastly different) and use a different block depending on what you're cooling.
Originally posted by ssjwizard
on a cnc 80 passes arent that many but the care for passes so close togeather and on cu sux
This is so true. This is why my block costs so much compared to something like a Maze 3. It's really stretching the limits of affordable machining and strives for the best balance of best possible performance with a price structure that isn't stupidly expensive.
gone_fishin
11-05-02, 08:24 AM
Originally posted by Cathar
Actually the block is geared towards cooling CPU heat dies of 15x15mm and smaller.
The Intel P4 CPU is 11.3 x 11.3mm. It has an IHS on top, however the IHS does very little to spread the heat when there's a good waterblock on top. >95% of the heat is still focussed in a 15x15mm area with the IHS on top of the P4 with the waterblock on it.
Due to the IHS, results for the P4 are not as impressive as for the bare die CPU's, however my own internal testing has shown improvements in the order of 3C+ over the other blocks I own (Maze 3, Cyclone 3, 4 & 5). I specifically blame the IHS for this with its second thermal interface, and Volenti's testing with and without the P4 IHS would appear to back up this theory. Volenti has his own version of my block and found that direct die cooling, and his semi-copy of my block (my production block does a few things better than his) were producing almost identical results.
Of course people may say my own testing is subject to suspicion, however I do have customers of my block seeing results very similar to mine. (+/- 1C), and I trust that when the first P4 blocks start shipping tomorrow that customers will again confirm my own results for the P4's. My own P4 doesn't overclock that high, so I'm even hoping that some customers will see better results that what I'm seeing on my P4.
My block is absolutely not geared towards peltier use. I am in the process of designing a peltier only block.
It's simply impossible to design a block that is optimised for both peltier (large die) and CPU (small die) use. If you want the best for both, you need to focus on the separate needs of each problem (and they are vastly different) and use a different block depending on what you're cooling.
Interesting on your P4 results. Can't post my development page here, but I updated it and there is a link to it at Procooling thread.
Cathar, do you have a performance curve yet for different flow rates?
Originally posted by gone_fishin
Cathar, do you have a performance curve yet for different flow rates?
My data points are these with an AthlonXP @ 1925MHz/2.15v.
I'll use 6.8lpm as a base-line.
0.9lpm => +3.0C
1.1lpm => +2.5C
5.0lpm => +0.25C (best guess) (Eheim 1048)
6.8lpm => +0.0C (Eheim 1250)
10.0lpm => -0.5C (Pondmaster 4200)
What happens between 1.1 and 5.0 I haven't defined yet. Looking at Les theoretical results I'd imagine that by 2.0lpm we'd be looking at something less than 1C worse than that 6.8lpm.
how much would it cost for someone in the U.S. to get one of your blocks?
JFettig
11-05-02, 05:01 PM
oh, about your guys crazy millling questions, you dont need to do 1/2 of the end mill diameter, you can easily do more... I have done .4inches with a 1/4 end mill in aluminum, i prob wont try to do that with copper tho..
NeoMoses
11-05-02, 05:45 PM
yeah, that's just a rule of thumb. Aluminum is much easier to mill than copper or steel, so you can take deeper cuts. I hate milling copper. It's so gummy. Give me steel or aluminum any day.
JFettig
11-05-02, 06:00 PM
hmm, i did mill .18inches with a 1/8 bit in copper also... till it snapped;) it was old and dull, and i was taking it too fast, otherwize it worked great
Originally posted by JML
how much would it cost for someone in the U.S. to get one of your blocks?
Anyone who wants one can send email to slf@connect.com.au.
There's a "user's results" thread that someone has started over here:
http://forums.overclockers.com.au/showthread.php?s=&threadid=109948
vBulletin® v3.8.7, Copyright ©2000-2012, vBulletin Solutions, Inc.