Direct CPU cooling

[D_o_S]

Well, I must admit I have never seen a waterblock like this:

http://www.vodnechladenie.sk/index.php?page=pag7

This is for direct cooling of your CPU (AMD 64 or Pentium 4). It has no bottom, so the coolant makes contact with your CPU's heatspreader directly. Any opinions on this? Is is a good water block?

[koei]

That's like walking on the edge of a knife. I don't see how people can enjoy their machine having to constantly pray that nothing leaks.

[killzone]

thats pretty sweet id try it just for the heck of it i wish they would show some results with that

[D_o_S]

That's like walking on the edge of a knife. I don't see how people can enjoy their machine having to constantly pray that nothing leaks.

For you guys, the following is not an option, but on their web they say that if you bring your PC there they will install the block and give you a lifetime guarantee. Also, if you're using a non-conductive fluid, I guess you have nothing to worry about.

[Psyche911]

But using a non conductive fluid usually means higher temps, right? Which would probably cancel out any benifit this would offer.

[koei]

A bit too much trouble and risk involved for potentially a few more degrees of cooling.

[MaxxxRacer]

i highly suggest not doing that.. for two reasons.. for one hte posibility of leaks is HUGE. especially for someone with little watercooling experience..

second is that it would crack eventually.. or get cracks in it.. thats what acrylic does.. my acrylic tdx top is already getting micro fissures in it.. its just a matter of time before they spread.

[Wicked Klown]

Direct die watercooling is kinda old. I did it three eyars ago with a celly 800 running at 1066. Temps load and idle never went above 21.7C. At the time there was one guy that did it to a T-bird and it lasted about 6 months my lasted 1year. Looking back on it sure it was nice cause of bragging rights, but not worth the hassle of protecting the mobo.

[Butcher_]

Any evidence to suggest direct die is actually better than a block? You loose pretty much any benefits of jet impingement with a direct die block, is it still any better than a normal block?

[MaxxxRacer]

well one of the guys around here is designing such a block that is diret die.. hopefully he doenst kill his comp.. and hopefully it gives some good temps...

[Wicked Klown]

Any evidence to suggest direct die is actually better than a block? You loose pretty much any benefits of jet impingement with a direct die block, is it still any better than a normal block?




Lets see you have the cool water hitting the heatspreader instead of the bottom of the block. So there for it's not having to mess with a 1/4" of copper or silver, it's directly cooling what needs to be cooled.

[MaxxxRacer]

wicked Klown dont make asumptions. Ancient wouldnt bring up that point unles there was some validity to it. it brings up the question if whether or not the water will be able to absort the heat better from just a flat surface such as the die, or will the water be better able to absorb heat from a design like hte mcw6000, tdx, or storm.. all of which have considerably more surface area than the face of a heat spreader...


would you like to revise your previous statement now Klown?

Im not saying which one is better, im just laying out some plain old facts here... Facts are much easier to deal with than conjecture.

[Wicked Klown]

It's not an asumption and no I would not like to revise my previous statement. As said before this is something I have done in the past.At the time I went from a all copper maze design to direct die watercooling. My temps went down roughly 5C. As I also said before at the time my idle and load temps were the same. Thing it is better cooling but is it worth the risk. After all one mistake and you rig is pretty mucj dead.

[caLume]

imho the block looks great.
logically it should have some really nice performance?

but i think on the whole thing its too dangerous. ^^

[SewerBeing]

I am joining Wicked Clown on this one. Yes even though there is more surface area you are still dealing with the fact that even silver has resistance to moving heat and then the resistance from the water. Also the fact that even with thermal paste you cannot get a perfect seal so there is more room for error. All in all direct die is better. Also the extra surface area doesn't help all that much.

[[AK]Zip]

I did it on my T-Bird a few years ago. Lasted for about 7 months. Temps were definatly better then using copper. Temps were only about 2*C higher then room temp at load. Would I do it again? yea maybe, but I don't have any time anymore to make another block for the P4 system. I used epoxy to keep it from leaking and it worked great. I ended up selling the entire system so I don't know what happend after the 7 months that I had it. If I find some pics of it I will post it up.

-Alex-

[Wicked Klown]

[AK]Zip by any chance were you the one on the OCF that did this??

[Holst]

Your better off and safer to take your heatspreader off and use a normal mounting copper waterblock.

Interesting idea, although not something I think is a good idea (or will give good performance)

www.spodesabode.co.uk

Wonder if that site still works?

[[AK]Zip]

[AK]Zip by any chance were you the one on the OCF that did this??

I did post there, but I wasn't [AK]Zip though. I forgot my old password so was unable to get back on. New name on there is [AK]Zip

[Holst]

Lets try another more specific link
http://www.spodesabode.com/content/article/directdie3

Very crazy direct die cooling there... read the other articles as well.

Allot of people tried DD watercooling at AOA a few years aggo.. quite a few good solutions but nobody made anything that was better than a block, or that was reliable.
IHS does make the job easyer... but its not the same thing IMO.

[Wicked Klown]

Your better off and safer to take your heatspreader off



Not really safer as the heatspreader is soldered to the core. So most likely you try and take the new heaspreader off you will break the core.

[[AK]Zip]

Not really safer as the heatspreader is soldered to the core. So most likely you try and take the new heaspreader off you will break the core.

I have taken it off. Its not soldered. Looks to be some sort of glue.

[Wicked Klown]

Really cause I remember reading and people saying it was soldered on.

[Butcher_]

It's not an asumption and no I would not like to revise my previous statement. As said before this is something I have done in the past.At the time I went from a all copper maze design to direct die watercooling. My temps went down roughly 5C. As I also said before at the time my idle and load temps were the same. Thing it is better cooling but is it worth the risk. After all one mistake and you rig is pretty mucj dead.
2 points:
1. A maze and a jet impengement block are not the same thing by a long way. Current impegement blocks a several degrees better than the old maze style blocks.
2. You were measuring on an old motherboard before the time of on-die sensors. Additionally stock sensors are useless at the best of times.

I'd like to see a proper test of direct-die vs a storm or similar block before making any judgements about whether or not it's worth it. And by proper test I mean the sort of test accuracy pH manages, not just slap a cpu in a motherboard and use mbm to take readings.

Having seen this debated before a lot of people with significant amounts of knowledge (cathar for one) are of the opinion that the heat density of a cpu is too high for direct-die to work effectively. You need a waterblock to spread the heat to a larger surface area for effective cooling.

[MaxxxRacer]

Butcher: I was just about to ask you if Cathar had made any comments on this. Knowing cathar he would have designed a block if it was effective cooling compared to other things.

And you make a very good point about the temp readings... we need a very large mallet to hit ppl with to get the fact that bios/mbm/any software tool's temp readings are so far off half the time that its scary and the other half of the time its in the general vecinity (sp).

[Wicked Klown]

When I did mine I din't use the mobo sensor for temps. I placed a temp die directly on the core. Seeing as temp sensor or with in .1% give or take I would say my temps were pretty damn close. Why not do the test with a direct die jet impengement style top. However I agree with your Maxxxracer as we should take a nice steel mallet and bash the skulls in on those who think they know everything. I'm willing to admit I don't but from what I have seen and done in the past, diert die is better.

[MaxxxRacer]

Klown, i was refering to the senors.. but it works for most things also.. lol... the mallet that is.

interesting way of measuring temps. usually thats not possible... I wonder how accurately it measures temps.

[Wicked Klown]

I mean we have those who swear that their mbm or mobo temp reading is right on the money. Sad thing is they can be off by as much as 10C. Now with the heatspeaders theres really no way to place a temp sensor close to the core. Also A why would they make a direct die pelt cooling if direct die wasn't better. After all a 172 or a 226watt pelt puts out more heat then any CPU out.

[Ancient_1]

Butcher: I was just about to ask you if Cathar had made any comments on this. Knowing cathar he would have designed a block if it was effective cooling compared to other things.

And you make a very good point about the temp readings... we need a very large mallet to hit ppl with to get the fact that bios/mbm/any software tool's temp readings are so far off half the time that its scary and the other half of the time its in the general vecinity (sp).
Yes Max he did have something to say about it in this (http://forums.procooling.com/vbb/showthread.php?t=10987) on ProCooling.

Here is a direct link (http://forums.procooling.com/vbb/showpost.php?p=132110&postcount=50) to his reply.

[The Mofo]

About 4 yrs ago someone did a direct die cool on the thunderbird core of the socket a Athlon, which was a pretty hot chip. 1C gain was measured.. Idea behind the terrible gain was pointed at the absence of pressure to the die.

[MaxxxRacer]

thanks for the link ancient.. i think that settles some things

[Butcher_]

Klown, i was refering to the senors.. but it works for most things also.. lol... the mallet that is.

interesting way of measuring temps. usually thats not possible... I wonder how accurately it measures temps.
Not very. Anything other than a built-in sensor is highly suspicious WRT to CPU temps. Typically an external probe will be placed on the edge of the core, which is typically much cooler than the rest of the core and not representative of the core temp itself. Direct-die only compounds the problem by making it almost impossible to mount a sensor anywhere near the core without having the water cool it directly as well (which would skew results completely).

[SewerBeing]

to the increase in surface area helping temps: There are limits to that, just look at the G5 it has most of the cups over the core and only a few over the other parts. So the difference cannot be that great otherwise Cathar would have made the cup region larger. Saying that I do not think that minor increase in surface area counters the thermal resistance of the materials. Also please remember Cathar is human and what he says should not automatically be accepted as the word of god...

[MaxxxRacer]

Butcher I already thought of that.. I didnt want to mention it as... well i was too lazy to type it i guess and didnt want to start more stuff with Klown... but since you mentioned it.. lol...

Sewer, cathar ascended a while back. didnt you hear?

[SewerBeing]

nope they dont let me out of my cage much :D

[MaxxxRacer]

lol :p:

[Butcher_]

Butcher I already thought of that.. I didnt want to mention it as... well i was too lazy to type it i guess and didnt want to start more stuff with Klown... but since you mentioned it.. lol...

I say it as I see it, Max. ;)

[MaxxxRacer]

lol.. its cool man.

[Wicked Klown]

Just remember tho Cathar in the begining said direct die was on par with the top of the line watercooling. However towards the end he goes against that. As said before tho he is human and his words are not written in stone. After all blocks like the G4 are as close as you can get to direct die with out having to worry about leaks. Don't worry tho Maxxx I just love getting under your skin.

[MaxxxRacer]

ahhh!! get him out!!! the klown is in my skin!! i can see him moving around!!

(my spanish teacher told us about this guy who thought bugs were under his skin and tore his skin off and killed himself in the process.. he was on crystal meth)

well DD used to be on par with cpu coolers at the time. now days we have about 3-5C advantage over the early blocks of the time.

and cathar was fairly consisten in the post. he mentioned that DD at its best was about as good as a WW clone. And the clone probably was not as good as the origional. and now we have higher performing blocks so it is safe to say that his post was consistent.

[Wicked Klown]

ok your kidding about the guy and the bugs right? Correct me if I'm wrong but doesn't his G4 and G5 cool directly over the core?

[MaxxxRacer]

they cool directly over it yes.. just like the tdx, rbx, WW, cascade, nexxosxp, and EVERY other impingement block known to man..

whats your point?

and no im not kidding.

[Wicked Klown]

Point is it's doing the same as direct die, just that it's safer. After all if direct die could handle the heat from a pelt, why couldn't it handle the heat of a prescott. Also man thats a heck of a way to go.

[MaxxxRacer]

u dont understand is the problem.. the way the G5 works is WAY WAY WAY differnt than direct die. Cathar would be utterly insulted if you said that to him and bill would tear you a new one.

im not going to get into the thermo/hydro dynamics of the G5, but i can garuntee you it does not do the same thing as direct die. if you made a DD block where you drilled mico holes in the die with little pins in them and had jets from the block that when into the cups then yes it would be essentialy the same thing. but you dont. with DD you have water spraying at a flat surface that isnt the greatest at heat transfer in the first place.

[Butcher_]

Point is it's doing the same as direct die, just that it's safer. After all if direct die could handle the heat from a pelt, why couldn't it handle the heat of a prescott. Also man thats a heck of a way to go.
Pelts are lower heat density so work better with direct die than CPUs. There is more at work than just the total heat output.