Aqua Computer or Dtek

[Top Nurse]

TN :

Yes (High flow doesn't cool unnecessary components)

Yes / No (Fittings aren't high flow, block design isn't low flow)

Yes (See point 1)

No (Good, relatively unrestrictive design for a chipset block, unless that is the GPU block....though aren't they the same?)

Actually they can be higher flowing blocks to cool RAM and VR's. Also many people use these blocks with DDC pumps and 6mm, 8mm, or 10mm tubing.

Remember that push-fits get a +1 over barbs in the Cathar overview of flow. Therefore 8mm tubing used with a push-fit is similar to a 10mm tube with barbs.

See point 1

That block is being used as a chipset block. Some people, like me, also use them for GPU blocks. However, most people use the full cover or TwinPlex Pro design for their GPU's.

 

[Ranker]

They're friggin noisy too when put next to an Eheim 1046.

Funny. I have an Eheim 1046 and the 1250. The 1046 is on the same level of noise as the DDC-1 (350) and pushes a pathetic amount of water. A D-Tek DB-1 does better than that and is quieter as well.

I can count two Aquaducts sold based on this thread. Lucky thing a new shipment is in the near future at www.sharkacomputers.com

AC will need all the help they can get. It's sad when you have to count how much have sold when I can't even count how many PA thermochill rads and Swiftech rads have sold in the past day few days. But hey, count the small victories whatever it may be. As long as you're initiating a person here and there into the AC cult right?

 

[Top Nurse]

Funny. I have an Eheim 1046 and the 1250. The 1046 is on the same level of noise as the DDC-1 (350) and pushes a pathetic amount of water.

Suggest you get your hearing checked. The high frequencies are the first to go with hearing loss.

I'm glad your Eheim only puts out a pathetic amount of water, but doesn't it do the job to keep your fish nice? My Aquastream pump gets the job done in a computer and that is what's important right? IIRC an Aquastream pump appropriately set (OC frequency approximately 74 Hz) has it's output exceeding a 1048.

AC will need all the help they can get. It's sad when you have to count how much have sold when I can't even count how many PA thermochill rads and Swiftech rads have sold in the past day few days. But hey, count the small victories whatever it may be. As long as you're initiating a person here and there into the AC cult right?

Actually any company needs all the help they can get and yours is also extremely appreciated. BTW, you have to sell a lot of radiators to equal 2 Aquaduct 360's. Some companies have to make it on the little stuff they import because they don't have the technical expertise, engineering know how, and capabilities to produce a product like the Aquaduct.

 

[Cathar]

Why are we all still arguing over the same old stuff three years on? I seriously would've thought that as a community the people here would've have (grudgingly) learned to just get along by now.

By totally closing our minds to different points of view, we're only harming our ability to accept the good elements that exist on both sides of the fence.

 

[Top Nurse]

Why are we all still arguing over the same old stuff three years on? I seriously would've thought that as a community the people here would've have (grudgingly) learned to just get along by now.

By totally closing our minds to different points of view, we're only harming our ability to accept the good elements that exist on both sides of the fence.

We weren't until a certain person decided to have at it on high flow and drag in all his flow buddies from XS. I guess perhaps I should have just ignored it, but the guy is really irksome.

My Feeding Frenzy project that never got finished is getting a revamp to a pair of DDC's powered with an AC Poweradjust controller. Now that I have a flow meter installed in my loop I can see various effects on cooling by using different flow, tubing sizes, and blocks. In my system doubling the flow only changed the CPU about 1.4 C which while a big deal to some wasn't a big deal to me considering some other issues.

So what's next from Australia or did you quit making blocks? Did you ever make that G5 with the nickel top?

 

[Justin_Diduch]

Why are we all still arguing over the same old stuff three years on? I seriously would've thought that as a community the people here would've have (grudgingly) learned to just get along by now.

By totally closing our minds to different points of view, we're only harming our ability to accept the good elements that exist on both sides of the fence.

Forum conflict is nerd crack. It will never end.
IMO the cooling forums would get a little stale if we didn't duke it out every once in a while. The flow debate is really an end user debate anyway. If thermal engineers and the companies they work for were flow cultists like many people here we wouldnt see companies like AC and Koolance switching to DDC pumps, or Swiftech making things like the Storm. Smart engineers and companies use what works (if they want to stay in business that is). Using you as an example Cathar, you have designed blocks that pretty much cover the entire range of flow possibilities at this point.
I guess what it comes down to is that people like conflict, I think it makes them feel less like sheep.

 

[Top Nurse]

 

[Cathar]

Yeah, I agree. Forum conflict is certainly entertaining in about the same way that a small child will deliberately do something to annoy the crap out of their parents just so they can get some pleasure from watching the drama unfold. After all, it's the reason why many of us watch TV shows isn't it? It's in human nature to thrive on drama.

Hmmm. Maybe that explains why I've been so distant. I guess I got to the point where I could see the good and bad points of both sides of the fence, and so got less enamoured by the drama of arguing about things which really are a matter of personal preference. It's not too different to arguing about religion.

As for what I've been doing. Things have been going quite slow of late. Been making the odd block, and exploring the odd new design style slowly, and at my own pace. Frankly I've been enjoying taking a break from water-cooling for a while. It's not that I don't like it, I do, just that lately other hobbies and family have been calling me more fervently than the w/c hobby.

There's a G7 Rev 2.
There's a wholly new block design not seen before
Still trying to finalise the dew-point TEC liquid chiller
Working on a wholistic system designed around the basis of 5/16" ID (1/2" OD) tubing (8mm/12.7mm) utilising push-fittings.

Currently have a broken right ankle (fractures in the talus) and broken right wrist (chipped tip of ulna, multiple fractures in top of radius requiring a plate). Am housebound and so have been unable to do my regular outdoors activities, and I guess that explains why I'm back here. I can't do the other things that have been distracting me, so instead I can do the things that don't require me to move around so much.

 

[R1ckCa1n]

Still trying to finalise the dew-point TEC liquid chiller

Working on a wholistic system designed around the basis of 5/16" ID (1/2" OD) tubing (8mm/12.7mm) utilising push-fittings.

These pre-orderable yet? I really am interested in your TEC solution and have been for years.

Currently have a broken right ankle (fractures in the talus) and broken right wrist (chipped tip of ulna, multiple fractures in top of radius requiring a plate). Am housebound and so have been unable to do my regular outdoors activities, and I guess that explains why I'm back here. I can't do the other things that have been distracting me, so instead I can do the things that don't require me to move around so much.

I hope your motorcycle faired better than you did.

 

[Top Nurse]

Sorry to hear about your recent visit to the trauma unit. Hope the pain pills are working and add a little cheerfulness to your life.

While we are on the idea of an unheard of block design, what's your take on this animal? The one I am thinking of getting is the G 1/4 design. They did say that the G 1/4 design would be slightly different. I assume that the fitting would have to be placed more central to the acceleration holes. From my understanding here it is has two totally separate cooling chambers on one block and is made for the dual or quad cores.

Yes AC has started making some of their blocks for bigger tubes. Seems like money really does talk.

Holistic design? More input please BTW, I have been edging more towards a 5/16" (8mm) ID design in Feeding Frenzy as it doesn't seem to make much difference in routing or aesthetics compared to the 1/4" tubing. But why would you use a 5/16" ID with a 1/2" OD? For big bore looks?

 

[Ranker]

While we are on the idea of an unheard of block design, what's your take on this animal? The one I am thinking of getting is the G 1/4 design. They did say that the G 1/4 design would be slightly different. I assume that the fitting would have to be placed more central to the acceleration holes. From my understanding here it is has two totally separate cooling chambers on one block and is made for the dual or quad cores.

Yes AC has started making some of their blocks for bigger tubes. Seems like money really does talk.

About time AC wised up and went where the performance is at and what the consumers want. They just need to drop the alu thing and then there will be peace.

In regards to the DI XT, Cathar's given his input on that thing over on XS: http://www.xtremesystems.org/forums/showthread.php?t=144102

 

[Top Nurse]

They just need to drop the alu thing and then there will be peace.

In regards to the DI XT, Cathar's given his input on that thing over on XS: http://www.xtremesystems.org/forums/showthread.php?t=144102

Never happen as they have tried making copper Aquatubes and it was a total disaster. They were also extremely expensive.

Your reading comprehension must not be very good to make a comment like that about Cathar in that thread. One of those at XS tried to assume what was inside the Cuplex XT DI by doing some photoshop magic. Last post I read was that Cathar wasn't making any comments till he saw what they actually did. After reading what he did say I'm glad I got one of the first ones hitting the shores here in the states.

 

[Cathar]

One of those at XS tried to assume what was inside the Cuplex XT DI by doing some photoshop magic.

Actually the photoshop mislead was done by one of the guys at nokytech, and really not the product of anyone on XS. Just someone with an overactive imagination.

What do I think of it though? I know from what I see of it that it's gonna be very restrictive. At a rough guess, without seeing the internals but in guessing that they're a variation on the Cuplex XT insides, I'd guess that we'd be talking around about a 2-6mH2O pressure-drop at 4LPM. It won't cut your flow-rates in half, but maybe to 75-85% of whatever they currently are with an XT in the same loop. Of course that won't affect the performance of the CPU block, that's what it's designed to run at, but everything else will run a little bit warmer.

Still, it's water-cooling. A little bit warmer for other components is still a lot cooler than air-cooling on those same components.

Also, assuming that Aquacomputer correctly gives you documentation to allow you to align the block so that the two JI areas line up over the dual cores, then it'd do a very decent job. Would it be the best? I have no idea. The best blocks are all fairly close, so I wouldn't want to make that call. Given AC's track-record, you can trust that it'd certainly be a very effective CPU w/b design.

The following is just my personal philosophy. I'm somewhat against tailor-made block designs though. If they're too tailor made for a particular CPU application that then just limits their re-usability when the market changes again. I'm still focusing on solving how to cool an entire IHS effectively and evenly in a manner that doesn't really care about what's underneath. I'm also against bowed block bases. That's an irreversible solution to a flawed mechanical design of the S775 socket.

 

[topcat989]

Why are we all still arguing over the same old stuff three years on? I seriously would've thought that as a community the people here would've have (grudgingly) learned to just get along by now.

By totally closing our minds to different points of view, we're only harming our ability to accept the good elements that exist on both sides of the fence.

The Phrophet has Spoken.

 

[Cathar]

BTW, I have been edging more towards a 5/16" (8mm) ID design in Feeding Frenzy as it doesn't seem to make much difference in routing or aesthetics compared to the 1/4" tubing. But why would you use a 5/16" ID with a 1/2" OD? For big bore looks?

1/2" OD for multiple reasons:

1) Easier/cheaper to source 1/2" push-fittings in this size
2) Easier/cheaper to source tubing in this size
3) Best trade-off between tubing flexibility without kinking (1" minimum radius) but still with minimal flexing torque. No need for elbows. No need for tubing guides/coils.
4) Mid-way between big-bore and small-bore, as suits my overall philosophy.

I accept that small-bore purists don't like OD's greater than 10mm. I agree that thinner tubing looks more elegant, but I can't find a way to turn decent corners without needing to use restrictive elbows when using tubing with wall thicknesses thinner than 2mm. The smallest tubing I'd go for would be 8mm/12mm, but the range of such metric sized tubing is both limited and expensive.

An 8mm or 5/16" ID is the smallest possible ID that I can accept while keeping with my wholistic philosophy of balancing the tubing/routing restrictions in line with the sorts of flow-rates that I want to achieve.

Input about wholistic design philosophy coming later. I might make it into its own thread.

 

[Halk]

And Cathar just sold me one of them

CPU is the only thing in my loop that's particularly heat sensitive. The other stuff just needs cooled.

I still need to know what AC mean when they say "NEW: The backplate adapter bolts allow the use of the typical A64 backplate for a better load sharing at the mainboard.
The cooler can be mounted alternatively without a backplate."

I have my Cuplex Pro mounted without a backplate. I'd prefer to mount it with a backplate, since my motherboard bends slightly using the nuts and bolts.

 

[Erasmus354]

I can agree with Cathar, 8mm ID 10mm OD tubing is nigh impossible to find. (at least on mcmaster....and they have everything don't they? )

 

[Top Nurse]

I can agree with Cathar, 8mm ID 10mm OD tubing is nigh impossible to find. (at least on mcmaster....and they have everything don't they? )

Page 110 => http://www.mcmaster.com at bottom of page, Item # 5958K15 @ $0.18/ft in semi-clear. Also available in opaque blue, green, orange, red, or yellow colors.

Also it is available from most any Innovatek or Legris distributor.

 

[Top Nurse]

I still need to know what AC mean when they say "NEW: The backplate adapter bolts allow the use of the typical A64 backplate for a better load sharing at the mainboard.
The cooler can be mounted alternatively without a backplate."

I have my Cuplex Pro mounted without a backplate. I'd prefer to mount it with a backplate, since my motherboard bends slightly using the nuts and bolts.

I think it means that they fixed the screwed up stuff they were shipping that didn't fit correctly to most of the back plates that came with the MB's.

Cathar sold you what?

 

[Top Nurse]

1/2" OD for multiple reasons:

1) Easier/cheaper to source 1/2" push-fittings in this size
2) Easier/cheaper to source tubing in this size
3) Best trade-off between tubing flexibility without kinking (1" minimum radius) but still with minimal flexing torque. No need for elbows. No need for tubing guides/coils.
4) Mid-way between big-bore and small-bore, as suits my overall philosophy.

An 8mm or 5/16" ID is the smallest possible ID that I can accept while keeping with my wholistic philosophy of balancing the tubing/routing restrictions in line with the sorts of flow-rates that I want to achieve.

Me likes 1" bend radius. Where can I get tubing and fittings?

 

[Erasmus354]

Me likes 1" bend radius. Where can I get tubing and fittings?

That's easy, mcmaster. I think what they call "instant fittings" are push-fits.

 

[R1ckCa1n]

I accept that small-bore purists don't like OD's greater than 10mm. I agree that thinner tubing looks more elegant, but I can't find a way to turn decent corners without needing to use restrictive elbows when using tubing with wall thicknesses thinner than 2mm. The smallest tubing I'd go for would be 8mm/12mm, but the range of such metric sized tubing is both limited and expensive.

I agree that a 10mm system optimal for the small bore purist. Now that AC and other small bore manufactures are starting to move to G1/4, you will see the shift. There is still something to say about 8mm OD and how clean it looks.........

 

[Cathar]

...and to answer TN's question about the fittings, they're like the ones on the old Swiftech MCW50 GPU waterblock. As to where you can get them in the USA, I'm sure some USA forum members on here can help out with that. You can get them with G1/4 ends on them for easy addition to blocks/pumps/radiators and so on.

 

[Top Nurse]

Those were the blocks that got me interested in push-fits.

 

[JC634]

...and to answer TN's question about the fittings, they're like the ones on the old Swiftech MCW50 GPU waterblock. As to where you can get them in the USA, I'm sure some USA forum members on here can help out with that. You can get them with G1/4 ends on them for easy addition to blocks/pumps/radiators and so on.

Those fittings are the ones I use to make my water-temp fittings. That and a whoooole bunch of marine epoxy.

 

[Cathar]

Been working on my wholistic system design models.

By wholistic, that means that we're factoring in things like the thermal resistance of the room as well as things like the pump heat-dump, and the heat dump of the fans on the radiator, the inherent thermal resistance and subsequent rise of water temperature as a factor of the flow-rate, and so on. As we use more powerful pumps and fans, we put more heat energy into the water and then ultimately into the surrounding air, and it's that air that gets used to blow through the radiator to cool our system. The rooms that we sit in can only dissipate so much heat as well - they aren't infinite heatsinks. Anyone who's used an electric heater can attest to that. Put heat in a room and the room will warm up. The effect is small, but I model that as well, because it does add an interesting curl to the whole idea of what is the "best" cooling system.

Also taking into account tolerable fan noise to determine what is an effective radiator C/W, and the practicalities of mounting radiators of various sizes.

Been modeling the theoretical best pump based upon common centrifugal pump characteristics, and using measured data from well known radiators and water-blocks on modern CPU's, and am starting to reach some conclusions which may be rather surprising to some people.

Almost invariably (high restriction system, low restriction, high/low heat load) the lowest water temperatures are achieved with flow rates between 2.5 and 5.0LPM. That's not the final answer, because even though the water may be getting warmer after 5.0LPM, the CPU waterblock may be cooling even better at higher flow-rates, and this can more than offset the penalty of an increase in the water temperatures.

As for what are the best flow-rates to use, there's no clear answer on that. This is because that all depends on the CPU block design as its flow-resistance and flow-performance curves.

What is coming out in the wash consistently though is that it is of no benefit to ever use a pump that draws more than around 20W of power, and that's at the extreme upper end where we're cooling the CPU, GPU's and so on, all over-clocked. For less fire-breathing systems, or when cooling just the CPU, a 10-12W pump is more than enough.

A pump like a DDC-2 or 3 with a modded top is extremely close to the be-all, end-all of desirable pumps, for high & low restriction systems alike. A pump like a DDC-1 with a modded top will also serve excellently, and is even more desirable than a DDC-2 if you're not overclocking everything heavily. Alternately, for maximum flexibility, a Laing D5 with its variable pot speed settings is another excellent choice.

There's a whole mass of variables, but there are strong trends emerging. It's a massive document I'm preparing. There's no STRONG case to favor either a high-flow or low-flow setup, but there is strong evidence to suggest that a low-flow high-restriction system can be made to perform just as well as any low-restriction high-flow system, so long as the correct pump, waterblocks, and tubing size is chosen. Some of the low-flow setups I see about are saddled with very weak pumps and overly small tubing, but on the flip-side, some of the high-flow setups are using stupidly over-powered pumps and needlessly oversized tubing.

The other thing that comes out is that it's extremely important to match components correctly.

I'll keep on editing and finalising the document. Might be a couple of weeks before it's ready.

 

[Top Nurse]

I feel like Neo believing that there was something not quite right with the water cooling world we live in. I couldn't put my finger on it, but I knew there wasn't something quite right.

From my use of AC components and their monitoring hard/software I am already to make a jump to 8mm ID tubing if I can find the right kind of fittings for my particular needs.

I really don't see a need for the DDC pump with my Aquastream pump running over what an Eheim 1048 does, but I am going to try one to see whether it makes a major difference or not with what I do with computers.

On my work box project I used a lot of 45 degree push-fits. Not for the reasons usually given, but just because the tubing was easier to route. Still had to use a few 90 degree ones due to design constraints of where I located certain components.

 

[Halk]

so long as the correct pump, waterblocks, and tubing size is chosen. Some of the low-flow setups I see about are saddled with very weak pumps and overly small tubing

Well, if I ever was going to change my fittings to move to a bigger tubing than 8mm OD, before I place my order would be the right time.

I don't suppose you'd like to comment on 8mm OD, and if it's overly small?

 

[Cathar]

I really don't see a need for the DDC pump with my Aquastream pump running over what an Eheim 1048 does, but I am going to try one to see whether it makes a major difference or not with what I do with computers.

I wouldn't bother too much with it, at least from what I'm seeing here. If you're running your radiator fans really slow, and not overclocking heavily (or at all), then what you've got is pretty close to optimal in terms of pumping power. You wouldn't make more than a 0.2C difference by moving to a DDC.

That was the other thing I was seeing in my calculations. There's a very broad plateau of acceptable pumping powers if we assume everything within 0.2C of the best possible pump choice as still being effectively equal.

While a 10-15W pump is near optimal (DDC1 or DDC1+ (aka DDC2)), a 5W pump or a 20W pump will still deliver final CPU temperatures within 0.2-0.3C of the optimal choices.

If you gave me/pointed me at a description of your full system, and told me your typical CPU/GPU loads and the typical speed you run you radiator fans at, I can quickly tell you how far away from optimal you are, but I just plugged some guesses into the calculator, and given your revved up 1046 Aquastream I can tell you that your system isn't likely to be suffering all that much for it.

 

[Cathar]

Well, if I ever was going to change my fittings to move to a bigger tubing than 8mm OD, before I place my order would be the right time.

I don't suppose you'd like to comment on 8mm OD, and if it's overly small?

It's my personal opinion, backed up by modelling analysis, that anything less than 8mm ID (not OD) is overly restrictive for most any setup unless the entire system's waterblocks are hyper-restrictive. You may not lose too much by going with smaller tubing than that, maybe 0.5-1.0C if going with 6mm ID (8mm OD) tubing instead of 8mm ID, but it's a simple choice. Tubing restriction is one thing that is easy enough to solve if you pay attention to it. Very small tubing adds needless restriction.

Actually, 9mm ID would be better. Anything above 3/8" ID (9.6mm ID) becomes largely irrelevant for all but the least restrictive of setups with the most open of flowing blocks on the highest heat-loads. IMO, 3/8" ID is "ideal", while 8mm ID (5/16" ID) is at the lower end of acceptable. Smaller than that and the tubing really is becoming a needless source of restriction in the system.

 

[Halk]

It's my personal opinion, backed up by modelling analysis, that anything less than 8mm ID (not OD) is overly restrictive for most any setup unless the entire system's waterblocks are hyper-restrictive. You may not lose too much by going with smaller tubing than that, maybe 0.5-1.0C if going with 6mm ID (8mm OD) tubing instead of 8mm ID, but it's a simple choice. Tubing restriction is one thing that is easy enough to solve if you pay attention to it. Very small tubing adds needless restriction.

Actually, 9mm ID would be better. Anything above 3/8" ID (9.6mm ID) becomes largely irrelevant for all but the least restrictive of setups with the most open of flowing blocks on the highest heat-loads. IMO, 3/8" ID is "ideal", while 8mm ID (5/16" ID) is at the lower end of acceptable. Smaller than that and the tubing really is becoming a needless source of restriction in the system.

Thanks Cathar.

Essentially what you're saying is that 10mm ID is ideal, anything above that adds little or nothing. Anything below that has an adverse effect. And also that 6mm ID (AquaComputer) does take a performance hit from the tubing size, but realistically it's about 1 degree.

For one degree I'll keep my current fittings, and stick with the easy mode tubing

Edit : Reading between the lines. You're either saying that the big gains in temps are from overall loop efficiency, or you're saying that the only way forward is with more restrictive blocks... Now of course I'm guessing here. But since pumps and tubing size has no "real" effect on temperatures (nothing significant) then it comes down to better contact with the core, better thermal paste, better blocks, overall loop efficiency. I'll see when you put up the big thing you've been working on.

As much as I'd like to wish you to get well soon. Erm, how about you don't get well until you've completely answered every watercooling question there is and can design the ultimate complete watercooling system. (Nah, seriously get well soon).

 

[Cathar]

Edit : Reading between the lines. You're either saying that the big gains in temps are from overall loop efficiency, or you're saying that the only way forward is with more restrictive blocks... Now of course I'm guessing here. But since pumps and tubing size has no "real" effect on temperatures (nothing significant) then it comes down to better contact with the core, better thermal paste, better blocks, overall loop efficiency. I'll see when you put up the big thing you've been working on.

Well, it's a little more involved than that. You're basically right, depending on how much value you place on the last few 0.1C points. I'm more-or-less tackling this issue down to the last 0.2C, as I'm sure I'll be able to show that the "best" setup really doesn't need to involve being excessive. There is a real penalty with getting overly restrictive with blocks because radiators do start to take a really noticeable performance hit once you get below 2LPM flow-rates. Below 2LPM, there's no (realistic) combination of factors that overcomes the radiator penalty, even with hyper-restrictive blocks, a very low heat load, the most balanced pump choice, and quiet radiator operation (all of which are factors which favor low flow-rates). In short, if your flow rates are less than 2LPM, then your system is not even close to optimal. If you have to stick a more powerful pump into the system to get your flow rates above 2LPM, but the pump is fighting against the restrictive tubing size, then the stronger pump is likely to make things worse because it's dumping more heat into the system. At this stage the tubing is what is holding us back. Bring the tubing ID in balance with the pump power IS a significant factor as it then influences what performance the radiators and water-blocks can achieve.

I'm doing this all independently of the waterblock mounting. That's really outside of a system's design to control. Yes, it's important, but changing tubing size won't affect to any great degree how well your waterblock is mounted.

So yes, it really is about balancing the whole system and bringing that to an optimal level. Once we do that independently of the water-blocks, we are then ready to select (or design) the blocks that suit.

 

[Cathar]

As much as I'd like to wish you to get well soon. Erm, how about you don't get well until you've completely answered every watercooling question there is and can design the ultimate complete watercooling system. (Nah, seriously get well soon).

LOL. Don't worry. I'm pretty much house-bound for the next 9 weeks at least.

 

[Narisatu]

Well, looks like ill be putting off most of my watercooling purchases for the next few weeks then. I really want to read that work that you are doing, and use it to find an optimal solution to what im trying to achieve with my setup in the fall.

Thats ok, I have another project brewing. RFID or Biometric starter for a motorcycle.. maybe even with use of a tablet PC or a custom built mini pc with a touch screen to do things similar to a Carputer in a tuner...

 

[Cathar]

Crunching some maths.

Given the following setup (chosen only because statistics were readily available).

Swiftech Apogee GTX
Thermochill PA120.3 with 3 x Yate-Loon fans @ 12v
2.0 meters of tubing and fittings to suit (see below)
CPU = dual-core Conroe, heavily overclocked and under heavy load generating 100W of heat

Independent of tubing size, the optimal pump to use works out to be a Laing DDC1 (10W) with a custom top. Using a ~5W pump (e.g.. Aquastream thingy), predict pretty much a fixed 0.15C CPU temperature penalty is seen regardless of tubing used. Using a ~20W pump (e.g. Laing DDC1+ 18W with custom top), a pretty much fixed 0.06C CPU temp penalty is seen. i.e. not much difference depending on what pump is used.

Now to the tubing/fitting impact on final CPU temperature:

6.35 (1/4") ID with quick-fit = +0.53C
8mm ID over 8mm barbs = +0.45C
8mmID with quick-fit = +0.19C
3/8" ID over 3/8" barbs = +0.13C
3/8" ID with quick-fit = +0.04C
7/16" ID stretched over 1/2" barbs = +0.01C
1/2" ID over 1/2" barbs = 0.37 + 0.24 = +0.00C

Hands up all those who expected that 8mm ID tubing carries less than a 0.2C penalty to your CPU temps.

Please note that the above differences are small because we've carefully balanced the pump to the radiator & waterblock. This in turn minimises the impact that tubing changes have on the final CPU temperature.

 

[Top Nurse]

Not surprised at all on your calcs. This is pretty much what a lot of us had been observing in our low medium to medium flow rigs. So what changes when you figure in the Aquastream at 8 watts?

 

[Cathar]

Not surprised at all on your calcs. This is pretty much what a lot of us had been observing in our low medium to medium flow rigs. So what changes when you figure in the Aquastream at 8 watts?

I'm really just guessing 'cos I don't have the actual PQ curve for the pump, but by extrapolating a 1046's curve, that at 8W that you'd be losing out to a top-modded DDC1 by around 0.1C. i.e. not much.

 

[Top Nurse]

If you gave me/pointed me at a description of your full system, and told me your typical CPU/GPU loads and the typical speed you run you radiator fans at, I can quickly tell you how far away from optimal you are, but I just plugged some guesses into the calculator, and given your revved up 1046 Aquastream I can tell you that your system isn't likely to be suffering all that much for it.

My worklog is called Feeding Frenzy Interlude. Here are some pics so you can see what I'm basically up to on the blcoks. I use the restrictive Cuplex XT directly from the pump => Twinplex NB block (silver one) => TwinPlex GPU block (copper one) => MB VR block => 360 radiator (HWL thin one) => flow meter (Innovatek) => Aquatube => pump.

I'm currently running the Athlon 3500+, 939 pin, CPU at a stock 2.2 Ghz and the fans are set up for a linear output tied to the input and output of the radiator.

 

[R1ckCa1n]

Hands up all those who expected that 8mm ID tubing carries less than a 0.2C penalty to your CPU temps.

[RAISES HAND]ME[/LOWERS HAND]

This is exactly what I have been experiencing for the past few years of messing with high end blocks with smaller ID (6/8mm) tubing. While my test bed has been pretty much the same minus my radiator is a PA120.2 I don't think that impacts the results too much. I really like the DDC1 w/ alphacool modded head

Given that GPU's now create more heat than a CPU, how does this apply when introducing a GPU (or two) block into the loop? My guess is the less restrictive of a GPU block the better for the smaller tubing.

 

[Narisatu]

i would assume that as long as the gpu blocks are less restrictive than the tubing itself, it shouldnt make much of a difference..