Overclockers Australia Forums

OCAU News - Wiki - QuickLinks - Pix - Sponsors  

Go Back   Overclockers Australia Forums > Specific Hardware Topics > Extreme and Water Cooling

Notices

Reply
 
Thread Tools
Old 8th January 2017, 3:14 AM   #16
Chardiieee Thread Starter
Member
 
Chardiieee's Avatar
 
Join Date: May 2009
Location: Clayfield, Brisbane, QLD
Posts: 993
Default

Quote:
Originally Posted by _slacker_ View Post
Fast
Cheap
Reliable
I am just trying to minimise the cost impact. I might be able to get away with the compression hardline fittings from Barrow, but I could use second opinions. I was not cheaping out entirely by going non-compression, but 250 in fittings from Barrow beats 350 in fittings from EK, and if they 'perform' the same then why not? 100 bucks pays for T-fittings and pass-throughs.


Quote:

run 2 d5s in serial and the loop in parallel and you wont have any worries ever.

You with one pump will lose a little pressure and hence flow rates suffer. If you eliminate the pressure problem you will have the flow. While that saying above is true that water will take the path of least resistance, you will be applying equal pressure to all 3 gpu blocks. GPUs really dont need huge amount of water to be tamed.

Its really not hugely important, no leaks is waaaaaay more important

Like I said, cost is a factor.

I understand how overkill flow -might- overcome the lack of flow as it goes up the chain here, but Im not really clear on how two pumps next to each other in serial is going to help (when I already have a fairly decent D5 at full speed).
Id understand using a booster pump halfway in the loop might help, but this doesnt seem applicable here...

The inlet pillar in the diagram is already the first obstacle in front of the pump, so IDK if it can get much better besides having an entirely better pump, not just two pumps going at the same speed. That seems more like a RAID 1 than a RAID 0 if you know what I mean.

Quote:
No-one will sponsor you because its been done to death, the romans understood fluid dynamics...
The only reason Id want 'sponsorship' (aka borrowing spare parts) is to see if this unique experiment would even work. I feel like a loop design as complex as this would benefit from a fair amount of parts AND testing, using decent flow meters and such to get some tangible evidence of the loop performance. Its not just so that I could have a flashy build and not pay for it.

It could provide valuable information for future builds which have similarly complex loops. Imagine if someone (me in the future, HA!) did a similar build, but with dual RAM blocks as well? What if there was another graphics card on top of that too?

Id love to conduct a full round of testing with different pumps and variables to obtain results that can stop the guessing.
__________________
My PC: [4930K] 4250MHz [GTX980 Classified] 1600/4200 [Gskill DDR3] 16GB 2133MHz 9/10/10/28 [X79 RIVE] Black Edition!
[Qnix 2710]
1440p @ 96Hz [TJ07] + watercooling = [EK Rad] 480 x 120 x 45 || [GPU: EK Clear Acrylic 980] + [CPU: EK Sup. Evo]

Last edited by Chardiieee; 8th January 2017 at 6:33 PM.
Chardiieee is offline   Reply With Quote
Old 8th January 2017, 3:20 AM   #17
Chardiieee Thread Starter
Member
 
Chardiieee's Avatar
 
Join Date: May 2009
Location: Clayfield, Brisbane, QLD
Posts: 993
Default

I feel like we should remember that a decent amount of flow in each block is entirely acceptable. It doesnt have to be exactly as good as the most optimal / efficient setup, as if it was done in serial or whatever.

If I lost a small amount of flow and minimal temp difference (compared to the best case scenario) then isnt it still worth it for something entirely unique? I know the cost is higher, but isnt that the consequence of all aesthetic choices?
__________________
My PC: [4930K] 4250MHz [GTX980 Classified] 1600/4200 [Gskill DDR3] 16GB 2133MHz 9/10/10/28 [X79 RIVE] Black Edition!
[Qnix 2710]
1440p @ 96Hz [TJ07] + watercooling = [EK Rad] 480 x 120 x 45 || [GPU: EK Clear Acrylic 980] + [CPU: EK Sup. Evo]
Chardiieee is offline   Reply With Quote
Old 8th January 2017, 4:16 AM   #18
_slacker_
Member
 
_slacker_'s Avatar
 
Join Date: May 2003
Location: hiding from census dude!
Posts: 1,240
Default

Quote:
Originally Posted by Chardiieee View Post

If I lost a small amount of flow and minimal temp difference (compared to the best case scenario) then isnt it still worth it for something entirely unique? I know the cost is higher, but isnt that the consequence of all aesthetic choices?

This is what Im trying to say, just do it as you picture it and it will definately work. I didnt mean either that Barrow are like cheap shit, they are a great fitting for the money, I just meant that if we all had like no budgets then how we do things would probably be inline with whats the absolute best we can run.

I spec out hydraulic fitups for trucks and so I do have a vague idea what im talking about

Im not here to discourage you, infact quite the opposite but you do need to listen to people who have done this before, there is no need for you to be a pioneer, that happened ages ago. All you will be doing is building on someone elses experience for your own benefit. procooling forums are a good place to go have a read if your interested in learning about pressure and flow and how they affect temps.

I also wasnt implying you just want free swag. But like the picture I posted before, thasts from EKWB, they have doen what you are saying is a maybe. They have pics to prove it So the biggest ( arguably ) manufacturer says its quite an acceptable fashion to do so just go for it bro.

If your interested in trialling some shit give me a PM and I'll loan you some gear for a month. Have a cpl of spare rads here and some pumps and res's's's's'ssss. they wont be in use for a while so if you want to science then yell out



Do you have blocks yet?
__________________
**Asrock X370 Gaming K4 ** AMD Ryzen 1700@ 4Ghz ** Team Group Dark Pro 3000 @2933 C12-11-11-11-25 ** XFX RX480 @1405/2055** Samsung Evo 960 M.2 + Evo 850 ** EKWB Watercooled ** Silverstone 750w Gold**
_slacker_ is offline   Reply With Quote
Old 8th January 2017, 9:17 AM   #19
havabeer
Member
 
havabeer's Avatar
 
Join Date: Dec 2010
Posts: 2,784
Default

Hardest part about testing t (accurately) is you would need a hand full of flow meters to show just how much of a temp difference it makes.

Yes you could just build the loop in parallel measure temps, and then rebuild in serial and measure temps and see which is better, but it would be interesting to see just how much flow drops in either case

Really I though it wouldn't make stuff all of a difference being a closed loop system, as all the loses are seen over the entire system and not per block
__________________
Current rig: CPU i7 5960X - Mobo MSI x99 pro carbon - GPU 2x Titan X (pascal) - Ram 16gig DDR4 2666 G*skill - Case Custom little devil V8 - Monitor Phillips 40" 4k
Cooling - Custom water loop - 360 rad x 2 - 480 x 1 - acrylic hard tube
havabeer is online now   Reply With Quote
Old 8th January 2017, 11:19 AM   #20
^catalyst
Member
 
^catalyst's Avatar
 
Join Date: Jun 2001
Location: melbourne
Posts: 11,569
Default

Overthinking it, do it in series, enjoy computer. You are not going to discover a magical loop configuration that gives you sub ambient or some shit. Also spending that much on fittings is clearly ridiculous.
^catalyst is offline   Reply With Quote
Old 8th January 2017, 6:28 PM   #21
Chardiieee Thread Starter
Member
 
Chardiieee's Avatar
 
Join Date: May 2009
Location: Clayfield, Brisbane, QLD
Posts: 993
Default

Quote:
Originally Posted by ^catalyst View Post
Overthinking it, do it in series, enjoy computer. You are not going to discover a magical loop configuration that gives you sub ambient or some shit. Also spending that much on fittings is clearly ridiculous.
Yeah its mostly an aesthetic way to build it, I know the laws of thermodynamics as much as the next guy, so no one is trying to cheat all of science, dont worry about that.

Spending this much is worth it if it means I have something really awesome, plus I can use all those parts for something else if I ever retire the build.

My ultimate goal is to have a really polished build, and this would just be the centrepiece.
__________________
My PC: [4930K] 4250MHz [GTX980 Classified] 1600/4200 [Gskill DDR3] 16GB 2133MHz 9/10/10/28 [X79 RIVE] Black Edition!
[Qnix 2710]
1440p @ 96Hz [TJ07] + watercooling = [EK Rad] 480 x 120 x 45 || [GPU: EK Clear Acrylic 980] + [CPU: EK Sup. Evo]
Chardiieee is offline   Reply With Quote
Old 8th January 2017, 6:29 PM   #22
Chardiieee Thread Starter
Member
 
Chardiieee's Avatar
 
Join Date: May 2009
Location: Clayfield, Brisbane, QLD
Posts: 993
Default

Quote:
Originally Posted by havabeer View Post

Really I though it wouldn't make stuff all of a difference being a closed loop system, as all the loses are seen over the entire system and not per block
I keep getting conflicting information but something tells me that youre right and it should be fine. I also really wont know until I do it, but Im still searching for some confident approval on the idea, or proven reasons for why it wont work, so that Im not wasting my time and money. Its a real doozy.
__________________
My PC: [4930K] 4250MHz [GTX980 Classified] 1600/4200 [Gskill DDR3] 16GB 2133MHz 9/10/10/28 [X79 RIVE] Black Edition!
[Qnix 2710]
1440p @ 96Hz [TJ07] + watercooling = [EK Rad] 480 x 120 x 45 || [GPU: EK Clear Acrylic 980] + [CPU: EK Sup. Evo]
Chardiieee is offline   Reply With Quote
Old 8th January 2017, 6:40 PM   #23
Chardiieee Thread Starter
Member
 
Chardiieee's Avatar
 
Join Date: May 2009
Location: Clayfield, Brisbane, QLD
Posts: 993
Default

Quote:
Originally Posted by _slacker_ View Post



If your interested in trialling some shit give me a PM and I'll loan you some gear for a month. Have a cpl of spare rads here and some pumps and res's's's's'ssss. they wont be in use for a while so if you want to science then yell out



Do you have blocks yet?
I have almost everything else ready to go, Im already running most of the loop until Im ready to install the motherboard blocks and redo all the tubing. I just need to exchange my 480 rad for 2x 240's, and Im considering doing a different res, since I only have a 150mm EK thats 5 years old now.


I did PM you.
__________________
My PC: [4930K] 4250MHz [GTX980 Classified] 1600/4200 [Gskill DDR3] 16GB 2133MHz 9/10/10/28 [X79 RIVE] Black Edition!
[Qnix 2710]
1440p @ 96Hz [TJ07] + watercooling = [EK Rad] 480 x 120 x 45 || [GPU: EK Clear Acrylic 980] + [CPU: EK Sup. Evo]
Chardiieee is offline   Reply With Quote
Old 10th January 2017, 1:26 AM   #24
Renza
Member
 
Renza's Avatar
 
Join Date: Dec 2004
Location: Adelaide, 5051
Posts: 4,600
Default

flow through each of the blocks will be inversely proportional to its flow resistance. I would say that it would be ok to parallel similar blocks, but parallelling blocks with vastly different flow resistance may cause flow starvation, especially with so many blocks.
__________________
i7 3770K @ 4.5ghz | AsRock Z77 Professional | 16gb Corsair Vengence 1600mhz | Gigabyte GTX580 | Seasonic X750 | Silverstone TJ07 | Crucial M4 256GB | 2x Dell U2011
Canon 5D Mark III | Canon EF 24-70 f/2.8L | Canon EF 17-40 f/4L | Canon EF 70-200 f/2.8L IS | Canon EF 50mm f/1.8 | Canon EX-580II | Manfrotto 190 Pro B
Traded over $22,000 on OCAU! - PM for recent traders
Renza is offline   Reply With Quote
Old 10th January 2017, 10:51 AM   #25
Chardiieee Thread Starter
Member
 
Chardiieee's Avatar
 
Join Date: May 2009
Location: Clayfield, Brisbane, QLD
Posts: 993
Default

Quote:
Originally Posted by Renza View Post
flow through each of the blocks will be inversely proportional to its flow resistance. I would say that it would be ok to parallel similar blocks, but parallelling blocks with vastly different flow resistance may cause flow starvation, especially with so many blocks.
The opinions here seem to learn more towards this idea working rather than having flow stagnation, although they are assumptions at best since no one can really know for sure. I do completely understand this statement that you and others have made, but there seems to be uncertainty on both sides.

Im not hearing much regarding the pump, unless everyone is in agreement that it is more than sufficient. I feel like this pump is more than powerful enough, beaten only by a DDC style pump in head pressure, which Im not sure I need to overcome this.
__________________
My PC: [4930K] 4250MHz [GTX980 Classified] 1600/4200 [Gskill DDR3] 16GB 2133MHz 9/10/10/28 [X79 RIVE] Black Edition!
[Qnix 2710]
1440p @ 96Hz [TJ07] + watercooling = [EK Rad] 480 x 120 x 45 || [GPU: EK Clear Acrylic 980] + [CPU: EK Sup. Evo]
Chardiieee is offline   Reply With Quote
Old 10th January 2017, 12:56 PM   #26
Chardiieee Thread Starter
Member
 
Chardiieee's Avatar
 
Join Date: May 2009
Location: Clayfield, Brisbane, QLD
Posts: 993
Default

Here is a new alternative design that Ive just mocked up which uses a serial flow path, but still retains the two central pillars which will prop up the GPU.

let me know what you guys think.

__________________
My PC: [4930K] 4250MHz [GTX980 Classified] 1600/4200 [Gskill DDR3] 16GB 2133MHz 9/10/10/28 [X79 RIVE] Black Edition!
[Qnix 2710]
1440p @ 96Hz [TJ07] + watercooling = [EK Rad] 480 x 120 x 45 || [GPU: EK Clear Acrylic 980] + [CPU: EK Sup. Evo]
Chardiieee is offline   Reply With Quote
Old 10th January 2017, 3:37 PM   #27
juzz86
Member
 
juzz86's Avatar
 
Join Date: Nov 2013
Location: Forth, TAS
Posts: 743
Default

Much better
__________________
7700K, 32GB, 980Ti, 960 EVO, PG278Q - DG-87
juzz86 is offline   Reply With Quote
Old 10th January 2017, 8:57 PM   #28
Renza
Member
 
Renza's Avatar
 
Join Date: Dec 2004
Location: Adelaide, 5051
Posts: 4,600
Default

Quote:
Originally Posted by Chardiieee View Post
The opinions here seem to learn more towards this idea working rather than having flow stagnation, although they are assumptions at best since no one can really know for sure. I do completely understand this statement that you and others have made, but there seems to be uncertainty on both sides.
Δp = Q2R, where p is the pressure loss over each block, Q is the flow rate, and R is the flow resistance.

If all blocks are parallel, then Δp over each block will be the same.

from there you should be able to figure out the relative flow rates through each block if you can come up with the resistance of each block.
__________________
i7 3770K @ 4.5ghz | AsRock Z77 Professional | 16gb Corsair Vengence 1600mhz | Gigabyte GTX580 | Seasonic X750 | Silverstone TJ07 | Crucial M4 256GB | 2x Dell U2011
Canon 5D Mark III | Canon EF 24-70 f/2.8L | Canon EF 17-40 f/4L | Canon EF 70-200 f/2.8L IS | Canon EF 50mm f/1.8 | Canon EX-580II | Manfrotto 190 Pro B
Traded over $22,000 on OCAU! - PM for recent traders
Renza is offline   Reply With Quote
Old 24th January 2017, 7:53 PM   #29
Jakusonfire
Member
 
Jakusonfire's Avatar
 
Join Date: Jun 2012
Location: Newcastle
Posts: 512
Default

For the original design in the OP as people have already said flow rate will vary depending on the restriction values of each block.
Current EK CPU and GPU blocks are of very similar values so you could regard them as essentially the same. The VRM block is normally quite a bit more free flowing, as much as half the restriction or less and the PCH turns out similarly.

That would give you the result of the flow in the CPU and GPU being significantly lower than the other two blocks.
The problems arise because in order to maintain a similar flow rate through 4 parallel blocks you need to quadruple the flow through the rest of the system. Flow restriction increases roughly with the square of flow rate. So even parts normally considered so free flowing as to be of little concern like (most) rads start to be real obstacles.

The Specific rad models of course are not provided, but unless the radiators are in parallel too then the flow rate achievable in the section other than blocks will be high, but not that high. I would give an educated guess at something like 8LPM at the very max with a single pump. In Ideal circumstances that would then be divided as 2LPM to each block but of course that won't be the case. The CPU/GPU pair will have around half the restriction of the VRM/PCH pair based on testing of similar blocks.
http://www.overclock.net/t/1501978/o...#post_22721397
http://www.overclock.net/t/1501978/o...#post_22799054

The system will split the flow so that each block has the same restriction. In tested blocks of similar types that can mean CPU and GPU getting half the flow rate or even less of the other blocks. If you had an 8LPM overall flow that splits to something like 1LPM for CPU/GPU and 2LPM for VRM/PCH.
Flow rates that low can have the effect of making the system painful the bleed and start to seriously impact temp performance.

In a conventional setup you would expect something close to 3.5LPM through each block.

These types of arrangements will always work, because the absolute minimum needed for blocks to function is very low. The Euro style of water cooling used low flow rates for years. It is just not getting the very best out of the system.
There is nothing to say that aesthetics shouldn't be just as important as bottom line temps.
__________________
Corsair 750D | Asus Sabertooth Z77 | 3570K | EVGA 780 | Samsung 840 Pro 256GB X 2 | EK Supremacy | EK FC-Titan | EK D5 X-Res w AquaC D5 USB
| Aquaero 6 Pro | EK Coolstream PE 360 & 240 | 16GB Corsair Vengeance Pro 2133 | NB Eloop B12-PS & EK Vardar F2 | Silverstone Strider+ 850W | Samsung S27A750 & LG E2290V | BF4:Jakusonfire

Last edited by Jakusonfire; 24th January 2017 at 8:29 PM.
Jakusonfire is offline   Reply With Quote
Reply

Bookmarks

Thread Tools

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump


All times are GMT +10. The time now is 8:37 AM.


Powered by vBulletin® Version 3.8.9
Copyright ©2000 - 2017, vBulletin Solutions, Inc.
OCAU is not responsible for the content of individual messages posted by others.
Other content copyright Overclockers Australia.
OCAU is hosted by Micron21!