Water block Writeup with the big dogs!!!!

[webmedic]

Well I'm working on one and have quite a few to work with. Updates on the blocks below.

From my own stash.
2 modified swifties for flow
1 maze3
1 maze2
1 of my own blocks.


I will have also 1 of these:
http://forums.procooling.com/vbb/sho...&threadid=3929

1 of these
http://forums.procooling.com/vbb/sho...&threadid=3873

1 #rotor block.

1 of the new spir@l's ( not gemini, borrowed from fragenstien )

I will be requesting blocks from a few others.

Becooling: A black edge for the testing.

Silverprop: I will be getting a cyclone 5 in a few weeks.

Liquidcc.co.nz will donate a surge to the testing.

Ocpc: I will try to get one of jessfm's blocks from the uk. No reply yet.

Dtek customs: No reply on the tc-4 yet.

Gemini: Peter wrote me back and said he would send an all copper version and a regular version with the plexi top.

Innovatek: No reply from the Innovatek people here in the usa yet.

If anyone would like to be in it let me know. Testing starting next week with individual block writeups to be posted. When I get all that I can I will post a final write up.

Any interested in getting stuff tested email or pm me. cooler@water-cool.com

[Sanjuro]

Finally! I've been waiting for something like this for a long time! GOGOGO webmedic!

[webmedic]

Yea me too. I got tired of waiting and decided to take the bull by the horns so to speak.

[Roscal]

Some explanations on the protocol of measure ?

[webmedic]

Ok I want it to be as acurate as possible under normal conditions. I've been talking to billa about getting a simulated cpu and setting it up but the cost is to prohibitive for me to get into that for now. What I will be monitoring is the built in xp diode. I will try at least to keep everything at .1 but will probably not be able to achive .01 at this time. A few other tests will be for general use but here again not in a laboratory or at laboratory level.

I hope to be as exact as posible but at the same time it will be verry much a test for those out there that are oc'rs and system builders. It will fit thier needs. I would love to have better granularity (is this the right word) but am not able to achive this at this time. I will be testing with multiple gph pump to get an idea of how the blocks perform at different flow rates.

Any suggestions are welcome in the testing methodoligy. As a matter of fact that is my next step to post a general idea of the testing to take place and ask for suggestions.

You, billa and others are welcome to add any input to keep me on track with results that should be for the most part repeatable.

[Pinkster]

so you will be using the internal diode? great, i have lost faith in external temp probes, even those flat ones touching the side of the core. When cathar on overclockers australia compared a cyclone 5 and maze 3 using a probe touching the side of the core he got virtually the same temps but with the cyclone 5 he got 20mhz higher overclocks

[#Rotor]

probe placing is the most critical.... when you plan on testing multitude of sinks, be that water or air, you have to have the probe mounted on the under-belly of the cpu for external... on-die diode, is obviously the king here.... use that for the electronic data gathering, and then also use digidocs, mounted like this , as a 2nd point of refference....

makes for nice pictures too

[webmedic]

Here's a reply to a pm I had. If anyone can think of anyting else please add it:


I think the first order would be to decide exactly what test I will run and maybe post that on my site for input on what others may like to see.


How's this for startes for the testing

1) Flow test with the blocks using the rotor bucket flow test. I know not completely scientific but as someone else pointed out in the thread that for our purposes it would be more than close enough.

2) for lack of a better way to put it temp monitoring over an extended period. This is what I will derive my c/w ratings from. This is the part I was refering ot doing at least 3 times.

3) the last part was purely subjective but it will be highest stable oc. The reason for this is that some blocks seem to give the same temps but allow a higher oc. This is hard to quantify but really if it will do an oc at such and such speed and the other won't what else can I say.

4) Now that you mention it I could do 150gph pump, 370gph pump, and for those out there cazy enough I can test a 700gph monster that I have. Do you think this whould be good in the pump testing. If I need more I do have them but these will be the easiest to use.

As for the as3 yes I have plenty and can get more easily. I know the rumors floating arround but have not found most of them to be true while testing as3. I have not found any settle in period whatsover. I'm pretty sure billa would be able to confirm this but I have not talked to him about it. Yes the block would be cleaned and new applied for each test.

[#Rotor]

your pump selection sound perfect from where I sit... will give a real life measure of what most people.....and odd crazy one... is using...

[jaydee]

From what I have been reading the built in probe is as good as the old one. Everytime a bios upgrade happens the temps change aswell!! That probe is as good as the bios makers make it. I don't like the idea of using a actual computer in the first place to test. I was going to setup a test bed that would consist of pelts with known wattages/voltages/consistant temps and you can put your own (accurate) probes in. If you can get the heat source to remain constant and consistant you will get better results. Also you will rule out any install errors that may happen.

I maybe way off base but that is how I would test water blocks for accuracy.

[redleader]

From what I have been reading the built in probe is as good as the old one. Everytime a bios upgrade happens the temps change aswell!!

I suspect those boards don't really read the probe.

Regardless I'd get an external reader so you can be sure its measureing properly. If you ask MeltMan he might be nice enough to make you one.

[webmedic]

Not true as it's totaly hardware. Most of the people griping about it are frustrated because it reads so much higher than the thermistor when under load. I've tested it well enough to know that the results are pretty good already. On the epox they usea winbond chip to read the temps from the xp diode. So under windows the readout is coming direct from the winbond chip.

What they may be experiancing is that the timings are different on some bios to either slow or speed up the system. To slow down in case of compatibility problems and to speed up for performance. Some bios's work better than others depending on your actual hardware. Some motherboards are closer to spek than others and can handle the timeings others have a dud part somewere that causes problems so they use a bios with relaxed timings. Due to the timings your cpu may actually be running at a higher clock than the bios indicates. This could cause the temp differences. Asus are know to run there boards oc'd from the factory at default settings to make them look faster.

I agree about the simulated cpu but when I do it it will be done right and will allow for much more exact testing.

[FRAGN'STIEN]

Originally posted by webmedic



for those out there cazy enough I can test a 700gph monster that I have.

Yeah you'll probably need that 700GPH pump for rotors block. That thing looks like it would eat that 700GPH and just laugh.

[bigben2k]

Good start...

Are you going to measure the coolant temp? How? Are you going to use straight water, or add something to it?

Remember to report the room temperature and humidity, for each testing session.

What rad did you have in mind? What fan(s)?

How are you going to handle the different barb sizes?

Since you made the PM public, I guess I'll finish it here, if that's OK (I assume yes).

1) AGREED. You're all set there.

2) measure the coolant temp.

3) As for the different OC results with same temps, I think that it may come from an irregular contact with the CPU core, where the block contacts the hottest area of the core better than where the diode is. I'd guess that the L1 ram is a great heat source. There's no fix for it, and I don't believe that there's any kind of testing that will help any, unless you start with some kind of theory, or idea why the result would be different. I'd suggest lapping the core, and either lapping the block, or at least checking it for flatness. You'd need a micrometer that's mounted on a table.

4) I'd measure the baseplate thickness, and post results for each pump. The big pump may not be necessary, but if you want to put in the time, then by all means do: it might turn out to be very interesting.

5) I agree. I think that the AS3 rumors are unfounded. I'm not so sure about the settling in period, but in any case yeah, BillA can set you up. Just please, describe how you are going to apply the AS3 (i.e. razor blade scrape after the application) in the highest level of detail as is possible. I can't stress this enough, because whenever someone has bad temps, and half the noobs out there do, it's because they didn't apply this paste right. It also happens to be the single most performance restrictor that there can be, throwing off temps by 5 deg C, very, very easily. Clean using what the AS website recommends, maybe even drop them an e-mail to double check. Another note, once it is applied, it is always there, no matter what you use to clean it, so it's a one time shot kind of thing, for the most part. If someone sent you a block on which they applied some cheap stuff, then you should ask them to either send you a new one, or advise them that it's going to be lapped again.

[#Rotor]

I have to defend my poor little block, by giving this link....


link

what is often not realized, is that my design will thrive on even the slowest of flow-rates, because the liquid to copper interface are so large and the ratio between surface area vs. block volume are very large as well. where as a more open, easy flow design, will get to a point, that if the flow is any slower, it just can't scrub the hot lair of water touching the side-walls, and therefor those designs go for a nosedive at low flow levels.

[FRAGN'STIEN]

Well I'm hoping to be at webs shop when he tests that block. I have got to see that work of art in action

[webmedic]

The coolant will be distilled water with hyper lube super coolant in it. It's like ww but without the nasty smell. I will post the exact ammounts that I used for the testing in the review. I will be able to report room temp but not humidity as I have no device to measure this. You are right this would affect the cooling. I hadn't even thought of it.

For the rad I will be using a geo tracker core with an ebm 213 cfm fan. The tracker core performs extreamly well as does the fan. It should not in any way be the bottleneck in the system. I also have regular big mama style heater cores available that I could use if for no other reason than that they are in wide use and would be simmilar to what others are using. As a last option I have a black ice rev2 that I could throw in. My idea was not really to test rads so I will save that for another time. Whatever rad I choose I think it would be best to use it for all the testing but this is negotiable.

As for the barb sizes. I do believe every block I have for testing is either 1/2 or 5/8. I'm using 1/2id tygon tubing and it will work on both. So I will not need to use reducers and the like.


Ok here goes point by point.

2) I have inline temp probes and will measure inlet and outlet temps from the waterblock. What exactly are these temps for and what do they show?

3) No the oc results will be in the last part of the report when I compare results. It's not ment to be in any way temp related just simly which block gives the highest oc that is stable. Like everything in this oc sport this will not be able to be reproduced by most as the equipment I have now can oc beyond what most people will have. I just got the right combination of parts right now. I've never even had parts that oc'd like this before. As for a micromiter I don't have one so ther is no way for me to do this. I do however lap using glass/mirror and things should be flat.

4) I can try to measure this but allot of blocks are soldered and wont allow this. I will try to find this data from the manufacturer.

5) In my own testing I never get as exact as that. As3 simply isn't as picky as that but smetimes I get a tube that will be funny. Yes the whole tube. however for the sake of repeatability I will use a consistant method of applying it. Maybe billa will have some input on this or even if as3 should be used.

[jaydee]

Originally posted by webmedic
Not true as it's totaly hardware. Most of the people griping about it are frustrated because it reads so much higher than the thermistor when under load. I've tested it well enough to know that the results are pretty good already. On the epox they usea winbond chip to read the temps from the xp diode. So under windows the readout is coming direct from the winbond chip.

What they may be experiancing is that the timings are different on some bios to either slow or speed up the system. To slow down in case of compatibility problems and to speed up for performance. Some bios's work better than others depending on your actual hardware. Some motherboards are closer to spek than others and can handle the timeings others have a dud part somewere that causes problems so they use a bios with relaxed timings. Due to the timings your cpu may actually be running at a higher clock than the bios indicates. This could cause the temp differences. Asus are know to run there boards oc'd from the factory at default settings to make them look faster.

I agree about the simulated cpu but when I do it it will be done right and will allow for much more exact testing.

Well I will take your word for, but some people report 9-10C differences after switching to a newer bios. I can't see that being all timmings as those are huge temp changes. I have not looked into what is what on the new probes and how they are read so I will not argue and your word is usually good and you definatly have been way more involved with computers than I have the last 5 months.
In any event it really shouldn't matter if the probe is off as long as you never change ANYTHING to make it read different. That way however off or on the probe is it will have the same offset on every block tested and the results will still be usable. For example if the probe is 5C off it will be 5C off for every block tested. That way we can still see which cooles better. The overall temp means jack as every system cools different but the difference between the overall temp is the key to seeing what can do what. The biggest concern I have with using an AMD system is it is so easy to mount a block wrong and get bad results. I know as I have done it a few times. I simple remount can get you 5-10C cooler temps than before. Thats the worst thing about the 4 hole hold down on the tiny amd core. If one corner is to tight it pulls the pother 3 corners up and you get higher temps. Even the springs don't make a lot of difference but they do help a little. What I want to see is air temp, water temp, cpu temp. Humidity? Sure why not. And yes a picture of the thermal paste install on every block would be a bonus. I agree with the last poster as thermal paste not being used correctly can cause bad temps. Been there done that! I also agree with the last poster on lapping the core. It will be a tidious task but I think it would be a good idea. lapping the block though I don't know. I personally would like to see results from retail blocks as they are when you would get them as if you where the customer. Then maybe lap it after the first test and see if it does any better, but of course that adds in extra time and energy.

Non the less your efforts will be appriciated by us all. Your website is looking real good!

[webmedic]

Originally posted by #Rotor
I have to defend my poor little block, by giving this link....


link

what is often not realized, is that my design will thrive on even the slowest of flow-rates, because the liquid to copper interface are so large and the ratio between surface area vs. block volume are very large as well. where as a more open, easy flow design, will get to a point, that if the flow is any slower, it just can't scrub the hot lair of water touching the side-walls, and therefor those designs go for a nosedive at low flow levels.

Thanks for that link rotor It I couldn't remember exactly how you did the test.

[Can O' Beans]

I would say:

1 - Ease of installation, quality of all components....

2 - Performance on a lower flow system - smaller pump, smaller lines, small rad... A system for someone who just wants to watercool for silence.

3 - Performance on a high flow system - big pump 1/2" + hoses, higher low rad (heater core, etc..) - A big system for those who try to Overclock as fas as they can go.

[#Rotor]

you are welcome.... now remember one thing with that method...... I aimed it at being as simplistic as possible. It is to give an overall idea of block flow rate, But more-so be repeatable, by anyone in the world, at absolute no cost, as far as testing equipment....


the resolution of the method is probable in to 20.0 to 60.0 LPH I was however able to get a noticeable difference, in testing the same block with 3/8 fittings, so it will be able to show a difference, if that difference is large enough to warrant a mention. And that was all the experiment was intended for.

of coarse you can make it very very accurate, just by using a much larger container.

[webmedic]

Originally posted by jaydee116

The biggest concern I have with using an AMD system is it is so easy to mount a block wrong and get bad results. I know as I have done it a few times. I simple remount can get you 5-10C cooler temps than before. Thats the worst thing about the 4 hole hold down on the tiny amd core. If one corner is to tight it pulls the pother 3 corners up and you get higher temps. Even the springs don't make a lot of difference but they do help a little. What I want to see is air temp, water temp, cpu temp. Humidity? Sure why not. And yes a picture of the thermal paste install on every block would be a bonus. I agree with the last poster as thermal paste not being used correctly can cause bad temps. Been there done that! I also agree with the last poster on lapping the core. It will be a tidious task but I think it would be a good idea. lapping the block though I don't know. I personally would like to see results from retail blocks as they are when you would get them as if you where the customer. Then maybe lap it after the first test and see if it does any better, but of course that adds in extra time and energy.

Non the less your efforts will be appriciated by us all. Your website is looking real good!

I will be using the swiftech type mounting on all blocks accept the one by paul as his is built into the block. With this system there is a standoff that only allows about 1/4 inch to tighten down. The springs will be 12 pound springs on each corner. In this way I can ensure as close as possible to the exact same pressures for each block.

[tim-may]

very nice so far! ill be waiting to see what comes out of this massive project

[webmedic]

Update: Peter from gemini wrote me and said he would send out an all copper version along with one of the regular plexi versions of his spiral blocks.

[bigben2k]

How about the TC-4? You really have to get one of those!