The impact of tubing sizes

[NaeKuh]

Well, your studies merited enough for me to try it out.

So thats just what i'll do. If it works great for me, then hell, i'll keep it, and recomend it. But if it doesnt work out great, then i'll just tear it apart, and rebuild it to how i like it.

Its just that simple for me. :]

[Cathar]

I will not subscribe to stuff spouted out as widely as my leaky garden hose connection to the faucet. This is the same guy, not too long ago, that swayed public opinion from 3/8 to 1/2 and then back down to 7/16, and now its back to 3/8 and he's even suggesting 5/16.
As far as I am concerned, he's just another product innovator, of which there are a baker's dozen. He's made his mark via a product, where the product's intellectual property was sold/transferred. He couldn't even yield commercial success out of the product. Reminds me of Xerox, which invented just about everything that one can mention, but could not generate any commercial success out of any of its inventions. Visions of the phrase "washed up" come to mind.

Sorry, you'll have to do better than that if you want to get a rise out of me.

There are historical reasons for the tubing sizing progressions, which I actually explained in an earlier post. Once again you're just demonstrating your ignorance for the history of the subject, rather than making any valid point.

There was no sale or transfer of intellectual property. That has never taken place.

I think from pretty much the beginning that I've always stated that I never held any aspirations to make any money out of water-cooling, and to this day I've purposely operated in just such a manner, and it is exactly why I've managed to maintain a status of commercial independence on forums. I don't compete with the commercial block makers. Never have, never tried to. I've never ran the water-block production as a business, and never even tried to. I've always operated as a hobbyist paying for expensive small batch runs and merely asked people to cover costs. Someone offers to make them for me, and I look at the time I'm spending on making blocks as opposed to my real job, and I think to myself: "Why not?".

Yet again you demonstrate your ignorance for history.

I'm just a hobbyist like everyone else on these forums. I give of my time and knowledge freely. I just believe that it speaks volumes for inherent bitterness in the character of various individuals who demonstrate a problem with that. Such people probably just need a good hug, but then they'd likely reject that too because they'd either be suspicious of the person giving the hug, or believe that they weren't worthy of it.

[Cathar]

I still would like to see a test done with at least another block. I vote Fuzion + a GPU block. Why? Because I think the differences might become more apparent as we introduce more restriction into the blocks.

I'll run a scenario like in the OP for you when I get some time. Have played around with it casually, and the tubing choice is about as much of a factor as for a single CPU block system.

[ira-k]

Nice testing Cathar!! I come through every now and then to see if you have left any new posts...Glad I came through today...Appreciate you taking the time to share this with us..I have to stay with my 1/2" barbs but its nice to know its not always needed...You and Bill Adams, Robotech and many others have greatly enriched our hobby with your passion for research....Thanks Again....

[Okda]

I'll run a scenario like in the OP for you when I get some time. Have played around with it casually, and the tubing choice is about as much of a factor as for a single CPU block system.

i had the same idea like Burn

thanks for the info Cathar

[Cathar]

i had the same idea like Burn

thanks for the info Cathar

Ok. Give me a suggestion for a CPU heat-load, and a GPU heat-load that you'd like to see me plug in.

Spent the last half-hour extending the software to accommodate a CPU+GPU system. Should be ready in a bit.

As for running the Fuzion, I don't have test data for it. From what I understand the Fuzion is reasonably close to the GTX, so the scale of the tubing impact should be fairly similar for both blocks. It's not like it's going to suddenly be a massive difference between the two. They are both mini-pin block designs.

One thing for the Fuzion though is that a smaller ID fitting effectively creates an impingement jet ala Scott's testing, and would actually boost performance.

[Okda]

Ok. Give me a suggestion for a CPU heat-load, and a GPU heat-load that you'd like to see me plug in.

i don't remember the exact numbers but what about an Intel Quad Core @ 1.45~1.5vcore with a single HD2900XT or 8800GTx or even SLI or CF setup

is that possible ??

personally i am interested in a loop with a Quad Core cpu overclocked and a single high end vga card cause this will be my future setup by Q4 2007

[kamongear]

What evidence could you have that such people wouldn't already know about? Any testing data you look at, they compiled. Any phase change guide you follow, they wrote. Any idea you think of, they thought of it first. By all means, feel free to argue with whoever you want, but let IanY's experiences with Cathar be a lesson to you. Mess with a legend, and you get put in your place

And that is why you my friend will never become a "legend".

The world has an infinite amount of knowledge that no one can master.

[SiGfever]

Alright, my system has been getting boring and redoing my loop sounds like fun. Now which 1/2" push fittings to use?

Thanks again Cathar this definitely needs to be stickied.

Hey Alex, how's the 3/8"id x 1/2"od Tygon stock?

[Cathar]

i don't remember the exact numbers but what about an Intel Quad Core @ 1.45~1.5vcore with a single HD2900XT or 8800GTx or even SLI or CF setup

is that possible ??

personally i am interested in a loop with a Quad Core cpu overclocked and a single high end vga card cause this will be my future setup by Q4 2007

Okay, did a bit of poking around. There's no hard and firm figures about, but I'll model for a 120W CPU heatload (overclocked dual-core at full-load) with a 150W GPU power (overclocked 8800GTX), which seems to be in the rough ballpark for heat dump. This seems to be more of a standard sort of system that more people would own.

An overclocked quad-core system with SLI 8800GTX overclocked, would be more like 200W CPU heatload, and 300W heat-load for the GPU's. That's just a freaky heatload in my opinion. If anyone's going to spend that sort of dosh and go to that extreme, just go buy an RD30, or 2xDDC2's in series, and use 1/2", and never mind about trying to be efficient. A quick run of the simulator says to just go all out with such a beastly machine. The radiator is going to be the major limiting factor in such a setup, what with ~550W of heat load including the pumps. Spend your money on getting as much radiator power as possible, and don't worry so much about the pump heat load. The old tried and true rules of excess is best dominates this scenario.

So yeah, I'll model for the more general case, that being a CPU+GPU setup that's affordable and doesn't generate more heat than a small nuclear explosion.

[HiJon89]

And that is why you my friend will never become a "legend".

I can't wait to prove you wrong, friend.

[Grinch]

I know if cathar does some testing/benching...you can take it to the bank!

[afireinside]

This is the same guy, not too long ago, that swayed public opinion from 3/8 to 1/2 and then back down to 7/16, and now its back to 3/8 and he's even suggesting 5/16.

Did you even READ the original post? He completely explained it. Heatercores and more primitive block designs benefited much more from the higher flow 1/2" tubing and we were using pond pumps. A maze 3, heatercore, and eheim 1250 setup was a completely different the researched and designed components you're buying off the shelf for the sole purpose of computer cooling.

Plus no one has ever even done thorough 1/2 vs 3/8 and smaller tests. How can you be claiming he's a sort of hypocrite when, up to this point, there was no thorough data of any kind comparing different tubing sizes other than 1/2" and 7/16".

[serialk11r]

I can't wait to prove you wrong, friend.

HiJon, he's right...the "OMFG I MUST WORSHIP THE GODS" attitude is exactly what's causing a lot of problems.

Thank you Cathar for sharing this with us.
But I think there's a problem, there is simply no way the pump dumps ALL the heat it generates into the water, but here you've done pretty much exactly that.

[Grinch]

IanY...trust me...this is an uphill battle...you are dealing with a PIONEER of watercooling...he is not from [H]...he is [H]ardcore.

[HiJon89]

HiJon, he's right...the "OMFG I MUST WORSHIP THE GODS" attitude is exactly what's causing a lot of problems.

I trust Cathar's testing methods and when he gathers empirical data I tend to listen, it has nothing to do with worshiping Gods and everything to do with respect.

[Grinch]

I trust Cathar's testing methods and when he gathers empirical data I tend to listen, it has nothing to do with worshiping Gods and everything to do with respect.

exactly what I was thinking..

[serialk11r]

For example, you don't disagree with Hipro5, OPB, Bigtoe, or Eva2000 when it comes to RAM. You don't disagree with Chilly1, PC Ice, or jinu when it comes to phase change. You don't disagree with k|ngp|n, Kinc, Sampsa, Lardarse, or OPPAINTER when it comes to 3D benching. And you sure as hell don't disagree with Cathar when it comes to water cooling. (or Maxxx or Marci for that matter)

Ahem...
Please don't say this kind of stuff...that's what I was referring to. If you have a valid point, and valid support for that point, then by all means disagree with them if they said something you think is wrong. The support is the big thing, if you make up random crap then SOMEONE will probably have pointed it out, and hopefully the people on the forum figure it out for themselves :P

[Cathar]

But I think there's a problem, there is simply no way the pump dumps ALL the heat it generates into the water, but here you've done pretty much exactly that.

Actually I didn't. I've measured the ratio of power draw to heat dump for a number of pumps. In general most pumps dump 85-90% of their power draw into the water directly as heat, and I actually factored that into the calculations.

I model the heat-dump of the pump into the water, and the related heat dump of the pump into the air as two separate entities.

[dinos22]

fantastic thread

somehow i don't think IanY is accepting anything Cathar says keep it up guys

i do understand what the argument is that IanY is hiding behind which is not only insulting Cathar but anyone that accepts any of his research. It's all good though we'll just do the usual blind following of Stew's factual work and enjoy the odd "ownage" comment . I must say it's becoming poetic almost.

i do question members if i feel they may have it wrong myself......but not like this....this is just comedy central

[edit]Stew i should have ran it by you before i posted on OCAU about your findings lol.....ah well what's done is done >>> talk about getting owned hahhahah

[HiJon89]

Ahem...
Please don't say this kind of stuff...that's what I was referring to. If you have a valid point, and valid support for that point, then by all means disagree with them if they said something you think is wrong. The support is the big thing, if you make up random crap then SOMEONE will probably have pointed it out, and hopefully the people on the forum figure it out for themselves :P

In the case that you believe a well-known member is mistaken and you believe your evidence is sufficient to prove your point, then by all means go ahead. But proceed with caution, because 99.9% of the time it's just someone running their mouth who thinks they know everything when in fact they're completely wrong.

[Cathar]

Okay, accept this as a completely estimation. Haven't cross-checked the validity of this data, and it's what is spat out of the first stage of the estimation software process. I haven't mapped this back onto an actual pump and flow-performance curves yet, but in general the scale of the results between the first stage simulator and the final real-world validation don't differ by much at all.

I modelled a 120W CPU, and a 150W GPU. I used the Swiftech supplied data for the Apogee GTX, and the MCW50. I applied a fudge factor of 3x, as in I multiplied the Swiftech supplied C/W values by 3, purely because the supplied C/W values seem to be indicative of Tcase to block, and not Tdie to block. In multiplying the Swiftech C/W values by 3, I seem to arrive at CPU/GPU temperatures that are also more in-line with real-world seen die temperatures.

Incidentally, this scenario is where something like a RD30 would come into its own, or possibly 2 x DDC2's (unmodified). For this scenario I'll assume a level of pump power akin to a top modded DDC2.

i.e. The setup is:
C2D CPU overclocked & under load @ 120W
8800GTX GPU overclocked & under load @ 150W
Apogee GTX on CPU
MCW50 on GPU
top-modded DDC2 (21W pump)
Thermochill PA120.2 with 2 x Yate-Loon
2m of tubing

i.e. I've applied a fudge on a fudge here, and in doing so have potentially amplified the nature of the differences (as opposed to reduced them).

1/4" QF : CPU = 62.33C, GPU = 52.28C
5/16" barbs: CPU = 62.31C, GPU = 52.27C
5/16" QF: CPU = 61.48C, GPU = 51.39C
3/8" barbs: CPU = 61.56C, GPU = 51.47C
3/8" QF: CPU = 61.13, GPU = 51.01C
7/16" barbs: CPU = 61.03C, GPU = 50.91C
1/2" barbs: CPU = 61.03C, GPU = 50.91C

Using an "optimal" pump (~34W estimated) would knock around 0.1C off the temps of both CPU/GPU. In fact, after running these for a while now, it seems that the "optimal pump power" is roughly proportional to your CPU+GPU heat load divided by 8. May be a dumb rule of thumb, but it seems to be holding roughly true.

To do the full deal will require more time. I have to attend to my job now.

[Okda]

WOW
so the sweet spot is the 3/8" QF tubes
Thanks Cathar for the info

[serialk11r]

Okay I know this is a little off topic, but here it goes.
Petra calculated that the DDC-2 is about 15% efficient. The other 85% has no place to go except to be turned into heat. Common sense says that part of that heat is going into the air, the PCB, the plastic housing, and eventually...into the air from those. Another part is going into the water. Here it doesn't matter that much because its only a few watts at most. And I'm sure plenty of people have reported that their pump's cases get a little warm during operation, meaning a good deal of the heat is NOT going into the water...

[gosmeyer]

I love to see a man on a mission.
Remember, You can lead a horse water but you can't make them drink.