Full cover vs. core-only GPU waterblocks

[Rodknock]

I know some of you are very adamant that core-only GPU blocks are the way to go for performance considerations. My question is why. Why are core only blocks better? Is it a flow-restriction thing that full cover blocks inherently have? I'm guessing that routing the water around the full cover block to cool the core *and* the ram is the issue, but I'm not entirely sure. Besides the obvious, are there any unseen benefits to a full cover block? What do you use to cool the ram if you go the core-only route?

 

[aigomorla]

Originally posted by: Rodknock
I know some of you are very adamant that core-only GPU blocks are the way to go for performance considerations. My question is why. Why are core only blocks better? Is it a flow-restriction thing that full cover blocks inherently have? I'm guessing that routing the water around the full cover block to cool the core *and* the ram is the issue, but I'm not entirely sure. Besides the obvious, are there any unseen benefits to a full cover block? What do you use to cool the ram if you go the core-only route?

http://www.xtremesystems.org/f...ghlight=fullcover+core


basically it was shown to give a hugh increase.

Read up on the link i sent you. If your still lost, ask questions, and i'll help you walk though it. :]

 

[Nemesis 1]

aigomorla maybe I am reading the link you posted wrong. But are you sure the differance is hugh. I only see 4c differance. I may be reading that wrong tho.

4c isn't going to improve O/C. As long as Gpu is under 50C looks good to me.

 

[aigomorla]

Originally posted by: Nemesis 1
aigomorla maybe I am reading the link you posted wrong. But are you sure the differance is hugh. I only see 4c differance. I may be reading that wrong tho.

4c isn't going to improve O/C. As long as Gpu is under 50C looks good to me.

nah he's pretty close to what i was estimating.

i ran into while working on a rig, and also because another member brought it to my attention.

MCR320 + Q6600(B3) + 8800GTX (Full cover) = no work :[ expecially if you Dis. Comp.

But if you change that to a MCW60 it works. :T

Ment that something was pushing another 80-90W extra as heat to overload that mcr320 because the top works on a PA120.3 which has about a 100W headroom with same fans over the mcr320.



And no for 4C id keep the blocks, even for 7C for the ease of application and bling factor. If your performance orientated, you would of went dual loops.

 

[Nemesis 1]

Watercooling CPU And 8800GTX Using MCW60 With Heatsinks:




CPU Idle: 27°C
CPU Load: 39°C

8800GTX Idle: 36°C (Diode)
8800GTX Load: 44°C (Diode)


Watercooling CPU And 8800GTX Using EK Full Cover Block:

http://www.xtremesystems.org/forums/...d=11744510 64

http://www.xtremesystems.org/forums/...d=11744510 99

CPU Idle: 30°C
CPU Load: 41°C

8800GTX Idle: 39°C (Diode)
8800GTX Load: 48°C (Diode


Now I didn't read anything in the thread but first post. But this is what I am seeing.

I am feeling fairly good today . So I am getting made at self for not seeing what your saying.

 

[aigomorla]

Originally posted by: Nemesis 1
Watercooling CPU And 8800GTX Using MCW60 With Heatsinks:




CPU Idle: 27°C
CPU Load: 39°C

8800GTX Idle: 36°C (Diode)
8800GTX Load: 44°C (Diode)


Watercooling CPU And 8800GTX Using EK Full Cover Block:

http://www.xtremesystems.org/forums/...d=11744510 64

http://www.xtremesystems.org/forums/...d=11744510 99

CPU Idle: 30°C
CPU Load: 41°C

8800GTX Idle: 39°C (Diode)
8800GTX Load: 48°C (Diode


Now I didn't read anything in the thread but first post. But this is what I am seeing.

I am feeling fairly good today . So I am getting made at self for not seeing what your saying.

he has a dualcore, not a quadcore B3.... different ball league.

And when i say it doesnt work, i mean it will be on par with air, so its not "worth" the migration.

You would need some pretty loud fans for it to work, and thats not what people look at when they think watercooling. You watercool to run away from the loud roaring fans.


Lastly look at his comment:
"By the way the 8800GTX at full load added about 7-8°C the the CPU core temperature. " <-- this is where i got my proof that my estimate was almost on par with his result. You want a performance h2o setup with a 8800gtx, you MUST dual loop. :T

 

[Nemesis 1]

Ok now I see what your saying. Its because of the differant cpu's.

I would like to point 1 thing out however and this is 100% dependent upon the Pump used and the rad.

I know for fact this works With a PA120.3 and an MCP 655 Pump because A friend of mine who didn't want to spend the kind of money to do dual loop asked me for help.

Use a Y with an 1/2" inlet and 2 3/8 outlets to the cpu and gpu. Now many will say the water will take the path of least resitance. This is true . If you have a Hi flow cpu block it becomes even a bigger problem . But its really not a problem simply install a ball valve in the line with least restance and with that pump. You should beable to balance the flow so that each block recieves 2 1/2 GPM flow . and that is enough flow for each block. With the heat dump from the pump and the 2 blocks the Rad mentioned does the lob nicely . He is within 2c of my wifes very expensive 4 loop set up. It worked for him and he is using a q6600 vs my wifes highly o/c dual 6800. GPU never goes above 50c and I won't say what cpu is as I don't need to be starting a war.

 

[DerwenArtos12]

Originally posted by: Nemesis 1
Ok now I see what your saying. Its because of the differant cpu's.

I would like to point 1 thing out however and this is 100% dependent upon the Pump used and the rad.

I know for fact this works With a PA120.3 and an MCP 655 Pump because A friend of mine who didn't want to spend the kind of money to do dual loop asked me for help.

Use a Y with an 1/2" inlet and 2 3/8 outlets to the cpu and gpu. Now many will say the water will take the path of least resitance. This is true . If you have a Hi flow cpu block it becomes even a bigger problem . But its really not a problem simply install a ball valve in the line with least restance and with that pump. You should beable to balance the flow so that each block recieves 2 1/2 GPM flow . and that is enough flow for each block. With the heat dump from the pump and the 2 blocks the Rad mentioned does the lob nicely . He is within 2c of my wifes very expensive 4 loop set up. It worked for him and he is using a q6600 vs my wifes highly o/c dual 6800. GPU never goes above 50c and I won't say what cpu is as I don't need to be starting a war.

Let me make sure I'm understading you right before i pass judgement. You're saying that your friends system that has an MCP655 choked down to a pair of 3/8" lines, one of which has a ball valve to further restrict flow, is working as well a set-up that has individual loops for each processor. What radiators are on your wifes supuercomputer? MCR120's?

 

[Nemesis 1]

Could you explain to myself how it is that you believe 2 3/8 lines are choking down a1/2 single line.

Wife has 4x pa 160. 2 pumps I listed above. The pumps run in parallel to the rads. 2 raids in series from their they go to very expensive 4 loop flow contol unit that is used in Injection molding industry.

She is very proud of her none super computer. Its only dual core.

1 loop to cpu

1 loop to gpu

1 loop to chipset

1 loop to her 4 raptors.

Case Lian Li PC 343 case

 

[DerwenArtos12]

Originally posted by: Nemesis 1
Could you explain to myself how it is that you believe 2 3/8 lines are choking down a1/2 single line.

Wife has 4x pa 160. 2 pumps I listed above. The pumps run in parallel to the rads. 2 raids in series from their they go to very expensive 4 loop flow contol unit that is used in Injection molding industry.

I'll submit I did not explain I meant choking down velocity, not flow.

Why would you have radiators running in parallel to the pumps instead of inline with each loop?

What type of flow controllers are you using? I helped design and build the flow control systems for my nephews shop. They specialize in Injection moulding and build in-house prototype molds for MANY big companies for both injection moulding and vacumn moulding. When shopping around for his flow controllers for both the injection and cooling sides of his injection moulders and for the coolant flow on their CNC machines we couldn't find any reputable companies making combined, self contained controlers for more than two lines. Beyond that you have to get a divorced controlling software and hook up all the valves to a system that very closely resembles a CCTV set-up.

 

[Nemesis 1]

Sorry it took so long to replt .Every since I installed Beta of SP3 for XP I have hard time posting I get booted off net , DAMIT.

I don't nowhere we got the links from it was one of the forums. The guy posting it showed his sert up and wife loved so she bought all the stuff he used except case.

Heres is the link . I am linking to the setup she bought but their are cheaper ones at same site. Others make these also . I don't know why you couldn't fing these but here is a little quality for ya.

http://www.emiplastics.com/water_regulators2.htm

Now its expensive as I said but so what. The Guy I am buying My Dark Star PC from uses same setup . So I believe he was the orginal poster I spoke of.

Note each loop shows water temp for that loop on the return side . Cool stuff, Its large but so is wifes case.

 

[Nemesis 1]

Originally posted by: DerwenArtos12

Originally posted by: Nemesis 1
Could you explain to myself how it is that you believe 2 3/8 lines are choking down a1/2 single line.

Wife has 4x pa 160. 2 pumps I listed above. The pumps run in parallel to the rads. 2 raids in series from their they go to very expensive 4 loop flow contol unit that is used in Injection molding industry.

I'll submit I did not explain I meant choking down velocity, not flow.

Why would you have radiators running in parallel to the pumps instead of inline with each loop?

What type of flow controllers are you using? I helped design and build the flow control systems for my nephews shop. They specialize in Injection moulding and build in-house prototype molds for MANY big companies for both injection moulding and vacumn moulding. When shopping around for his flow controllers for both the injection and cooling sides of his injection moulders and for the coolant flow on their CNC machines we couldn't find any reputable companies making combined, self contained controlers for more than two lines. Beyond that you have to get a divorced controlling software and hook up all the valves to a system that very closely resembles a CCTV set-up.

When you talk about velocity are you referring to head , Because velocity and flow are the same. threw tubing=sized tubing .Increased Head would only push the water further threw a more restictive system befor losing flow. We measure this resistance at input and output of flowcontrol unit with pressure guages.

 

[Nemesis 1]

Why wouldn't we use 4 loops 1 for each rad, Fair enough question. We only wanted to use 2 pumps. Since running in parrallel with 2 pumps =2x flow but head stays the same its all that is required. Head does not increase with parallel pumps. Pumps in series increases head but flow doesn't increase.
So 2 pumps in parrallel would give us a little more than 10 GPM flow it was detremined by us, That it was all we needed.

 

[DerwenArtos12]

Ok, the way you had it before i thought you had radiators in parrallel to the pumps, you have two pumps in parallel with radiators on those, so essentially you're combining two loops into one then using the flow control to break it out into four loops.

Trust me, those flow control valves are cheap compared to remotely monitored and controlled flow control valves like the Sierrra M100L units we ended up using. Could you immagine having to monitor and control 127 manually? It would take a staff of 4-6 people and the plant runs 24/7.

I still think you would get better temperatures by dividing out the radiators for each loop, putting them after the flow control and before the blocks then feeding back into a pair of y's to each pump then through another y back into the flow control.

And what I meant by velocity is the combination of head pressure and total flow. Going 1/2 to dual 3/8 is going to drop head pressure by about a third but, it's also going to divide flow, IMHO with a large enough radiator the flow and temperatures would be better with the two inline, cpu then gpu.

 

[Nemesis 1]

Ya for a hugh shop like your talking about those controllers would be cost effective. But a small shop it foolish waste of money. The Machine operator can monitor the mold temps just fine with the simplier flow control system I linked. Gettinga mold to cold isn't a good thing /same as to hot. Just right is operator controlled. The shop I worked in was small and used ball valves only. I complained about it but owners were cheap scapes. What type of cooling towers does your cousin use in his business. I hated working on those . Its one of the reason I became a nurse. But alast 1 month after becoming a nurse . A drunk driver changed my life for ever. I am not bitter about it. But I wish people would think befor drinking and driving.

I was always good with machines so I wanted to see if I could work the same kind of magic with sick people. Thats the only thing that really bothers me . As I will never know the ans. to that. God in his goodness must have other plans for myself.

 

[aigomorla]

Originally posted by: Nemesis 1
Why wouldn't we use 4 loops 1 for each rad, Fair enough question. We only wanted to use 2 pumps. Since running in parrallel with 2 pumps =2x flow but head stays the same its all that is required. Head does not increase with parallel pumps. Pumps in series increases head but flow doesn't increase.
So 2 pumps in parrallel would give us a little more than 10 GPM flow it was detremined by us, That it was all we needed.

nemisis your getting the concept mixed up in h2o cooling.

Benifits of paralelling your radiators is that you get less flow restriction then having them in serial. True...

However headpressure is what dictacts preformance more.. why?

because headpressure is what gives you water pressure. This pressure is the main driving force for pushing water though the blocks. Each time water hits a block it takes a reduction in pressure. Flow is kept uniform however.


Now dual loops is a must. A proper paralell setup is very costly, and very difficult to maintain. If one section of the pathway has less restriction, more water will travel down that route and starve the other path.

This is why i prefer to just dual loop it. Indpendant control of both cpu and gpu is something you definitely want on a performance loop.

 

[Nemesis 1]

To be perfectly Frank without trying to be Rude I doubt very much that you can educate me I fluid dynamics. I understand very much the relationship between flow/head in relationship to restriction of flow. I have worked and maintained huge hydraulic systems.

Point out in my post were it is that you think I don't have a full grasp of this relationship. In my wifes system she has a total rise of total 2.9' . Which would reduce total flow to about 71/2 GPM. Now if you ran everthing she is cooling in a single loop. The rise would be close to the same but I would have to measure that to know exact rise.

But in that single loop you would be preheating the water more and more as it goes threw each block and resistance to flow would increase. Even with 2 pumps in series. My wifes system would have more pressure at retun to inlet at pump . not only that the water temp would be lower their also.

 

[aigomorla]

Originally posted by: Nemesis 1
To be perfectly Frank without trying to be Rude I doubt very much that you can educate me I fluid dynamics. I understand very much the relationship between flow/head in relationship to restriction of flow. I have worked and maintained huge hydraulic systems.

Point out in my post were it is that you think I don't have a full grasp of this relationship. In my wifes system she has a total rise of total 2.9' . Which would reduce total flow to about 71/2 GPM. Now if you ran everthing she is cooling in a single loop. The rise would be close to the same but I would have to measure that to know exact rise.

But in that single loop you would be preheating the water more and more as it goes threw each block and resistance to flow would increase. Even with 2 pumps in series. My wifes system would have more pressure at retun to inlet at pump . not only that the water temp would be lower their also.

i dont disagree with you on a paralell setup, however performance driven systems are ones which hare heavily dependant on pressure and velocity at which the water is shot though the block.

if you use accelerators, and you have inadquate head pressure to allow for such accelerators.

Also its a known and proven fact, that when you have more head pressure on accelerators, your performance increases as the velocity of water and turbulance is increased.

Lastly, i never said paralell setups were bad, i said they were extremely difficult to setup and maintain. You need proper flow reducers with valves on each loop + a flow meter. Also you need to adjust your loop every month or so to keep your flow uniform. Something in the system might throw the entire loop astray.


This is why i said dual loops is a must for performance based systems. Setting the gpu on one loop and forgetting about it, while tuning the CPU loop indpendantly is what most performance addicts in this hobby go for.

Example of an accelerator block. Very hard to get good performance out of it without a pump that has good head pressure. Water will be majorly restricted from passing though the block.

http://i125.photobucket.com/al...aigomorla/IMG_0842.jpg


Also here is some numbers for you in terms of pressure reduction though nozzles and accelerators.
http://www.xtremesystems.org/f...8551&highlight=nozzles

And i use an RD-30 as my main drive pump. Head pressure output @ 30ft for a cpu loop only config. I use this because the head pressure on the pump is king in its class/size. This head pressure gives me a TON of benifit on blocks which use accelerators.

If you have a great working paralell setup, then thats wonderful. I would like to see pictures, however paralell is something i would never recomend anyone unless they really wanted to try one. Its just too difficult to maintain compared to dual looping it.

Also.. if you think im wrong somewhere feel free to correct me. If you have a valid point where im stuck at answering, i'll even goto people who i feel know more then me and ask them direct also. This is all about learning, not proving who is right or wrong.

 

[Nemesis 1]

Its a fact. Flow is given at a certain rise . It depends alot on how the specs are used. If you have 5gpm flow at 13' rise. Do an experament to prove to yourself. You will find few pumps that are rated at 5gpm at 13' rise that would actually deliver 5gpm @ 13' rise that are used in PC's .

Many companies are now removing this confusion . A pump that is rated @ 5gpm is the amount of water delivered at zero rise. as you increse the rise flow decreases.

Companies use some pretty strange wordings on This relationship between Flow rise and length of run . Here is how it works check it yourself its easy to do .

Take a pump that is specied the confusing way and try this.

5 gpm flow and measure flow for 1 min. by filling a 5 gallon pale. At zeto rise. You should get 5gallons in 1 min.

Than do the same thing at 3 ' of rise . Flow will be reduced. At 13' flow will be a trickle. @ 14 ' no flow as the pump can't overcome the rise.

 

[aigomorla]

Originally posted by: Nemesis 1
Its a fact. Flow is given at a certain rise . It depends alot on how the specs are used. If you have 5gpm flow at 13' rise. Do an experament to prove to yourself. You will find few pumps that are rated at 5gpm at 13' rise that would actually deliver 5gpm @ 13' rise that are used in PC's .

Many companies are now removing this confusion . A pump that is rated @ 5gpm is the amount of water delivered at zero rise. as you increse the rise flow decreases.

Companies use some pretty strange wordings on This relationship between Flow rise and length of run . Here is how it works check it yourself its easy to do .

Take a pump that is specied the confusing way and try this.

5 gpm flow and measure flow for 1 min. by filling a 5 gallon pale. At zeto rise. You should get 5gallons in 1 min.

Than do the same thing at 3 ' of rise . Flow will be reduced. At 13' flow will be a trickle. @ 14 ' no flow as the pump can't overcome the rise.

i understand this concept fully. however each time you hit a block, its like adding another 2-3feet of tubing straight up vertical. The restriction of the block minics added increase climbing height.

Thats why headpressure is important when setting up performance loops.


But saying flow is more important then headpressure, is a myth in h2o. Thats why people use DDC-3.2's instead of D5's because the 3.2 with a top gives more head pressure, yet the D5 provides more flow.

 

[Nemesis 1]

Here is an example of a correctly spec. Chart. Notice Total head is given . Also notice flow is a zero rise(head).

http://www.bluelineaquatics.co...ter_pump/velocity.html

Now If you read about these pumps you could get excited . But I bought 2 their heat dump is crazy. But I had to know.

I would never ever recommend the type of set up my wife has to anyone . Because of cost. But as far as maintaining flow to each block its the easiest system there is . In the fact everthing is right there in front of you for easy adjustment.

You would have to ask my wife for pics. After that remark a poster made about me last week and my medication . good luck with that. It didn't bother me at all . But the rest of my family was ready to go yo war . Ya they over reacted. It was a comfort to me to know they love me that much to worry about my feeling .

 

[aigomorla]

Originally posted by: Nemesis 1
Here is an example of a correctly spec. Chart. Notice Total head is given . Also notice flow is a zero rise(head).

http://www.bluelineaquatics.co...ter_pump/velocity.html

Now If you read about these pumps you could get excited . But I bought 2 their heat dump is crazy. But I had to know.

I would never ever recommend the type of set up my wife has to anyone . Because of cost. But as far as maintaining flow to each block its the easiest system there is . In the fact everthing is right there in front of you for easy adjustment.

You would have to ask my wife for pics. After that remark a poster made about me last week and my medication . good luck with that. It didn't bother me at all . But the rest of my family was ready to go yo war . Ya they over reacted. It was a comfort to me to know they love me that much to worry about my feeling .

Nah Nemesis, i think i understand how you type now. So i know when it seems like your attacking, its really your asking or trying straighten something out.

I got no quams with you bro, and you do give out a lot of interesting facts.


And yeah.. i know about those pumps, my past hobby was reef tanking, and i went all out. Only issue is it takes more effort and time to service and maintain it.

But thats why i love water.

RD-30 Specs.
http://i125.photobucket.com/al...aigomorla/IMG_0658.jpg

Size Comparision to a D5
http://i125.photobucket.com/al...aigomorla/IMG_0661.jpg

Inlet Comparision:
http://i125.photobucket.com/al...aigomorla/IMG_0660.jpg


I like monster eq. :T

 

[Nemesis 1]

Well thank you for that . Its so hard on the net. I don't care what people say or do as long as it has no direct effect on me or mine.

Water cooling a pc is really a hard thing to grasp in actual fact . Because there are so many things that were stated to be fact by the big guns in the water cooling pc world that just weren't true. Size of line was one that had everbody confused and caused many a forum war. Than you have these water blocks and design . Were fanbois of one products vs another is just retarded. If you use restrictive blocks ya have to do the right things . A high flow block ya can do something else. This is were the confusion comes from many of the so called gurus can do the math . But that doesn't mean they understand the concepts.

One that really still isn't understood buy many . is the flow threw a rad. I perfer my water running threw the rads at a slower rate. In racing we have to actually use restrictors to slow the water down to cool our engines. So I like a rad like the PA160 as water can move threw it slower while still maintaining good flow without a big pressure drop.

Each system has its own little tuning points. So is what I am saying is there is no 1 right way. Each setup is like humans were all differant.

If you have a system thats not working correctly you have to tune it to get max eff. parts many time have to be replaced.

 

[aigomorla]

LOL of course!

You always look at application b4 you start on any project.

It makes no sense to buy all the hardware and then work on application for it.

Its just like the financial world. You only get things for what you need and not the other way around.

 

[DerwenArtos12]

I'm glad there are some people here that understand the difference between taking sides on a subject and pecking a fight. We can all learn more that way.

:beer:s all around!