Overclock.net › Forums › Case Mods & Cases › Builds & Case Mods › Build Logs › [Build Log] CaseLabs STH10 - Tri EVGA - SR-X - Dual LGA-2011 - H20
New Posts  All Forums:Forum Nav:

[Build Log] CaseLabs STH10 - Tri EVGA - SR-X - Dual LGA-2011 - H20 - Page 31

post #301 of 353
Thread Starter 
No need to run the CPUs in Parallel too?
SR-2
(13 items)
 
  
CPUMotherboardGraphicsRAM
Dual - Xeon x5650 EVGA SR-2 Dual - EVGA GTX 470 / GT 240 Corsair Dominator GT 12GB (6 x 2GB) 
Hard DriveOptical DriveOSMonitor
C300 / WD RE4 Dual / BluRay RW Windows 7 Pro Dual - Dell 2709W 
KeyboardPowerCaseMouse
MS 6000 V3 Silverstone ST1500 Mountain Mods Tower MS Sidewinder 
  hide details  
Reply
SR-2
(13 items)
 
  
CPUMotherboardGraphicsRAM
Dual - Xeon x5650 EVGA SR-2 Dual - EVGA GTX 470 / GT 240 Corsair Dominator GT 12GB (6 x 2GB) 
Hard DriveOptical DriveOSMonitor
C300 / WD RE4 Dual / BluRay RW Windows 7 Pro Dual - Dell 2709W 
KeyboardPowerCaseMouse
MS 6000 V3 Silverstone ST1500 Mountain Mods Tower MS Sidewinder 
  hide details  
Reply
post #302 of 353
Quote:
Originally Posted by asg View Post

No need to run the CPUs in Parallel too?

Yes on the pump arrangement. The difference won't be huge, but it will help having one of them mid-loop.


You definitely want the CPUs in series and will also want the GPUs in series if your CPU flow is under 1.3gpm when the GPUs are in parallel. As I said before, the XTs are all about low flow/high pressure so doing parallel reduces flow a lot through each block. If you decide to run D5s instead the story changes because they are about high flow/medium pressure. XTs use a lot less power (almost 4 times less than a D5 on 12v), but even completely unrestricted they cap out at about 1.8-1.9gpm. That's why I think the GPUs should probably not be parallel unless they absolutely need to be to help the overall flow rate.

Dual D5s would give much better flow, but also uses 4 times more power (on max). MCP35X2 would be even more and you could put everything in parallel with either and they would still have around 1.5gpm in all blocks/rads (MCP35X are crazy powerful). The XTs look really cool though and give you all sorts of software options.
post #303 of 353
Thread Starter 
Thank you for the help...

Back to the drawing board for fittings. I think having the Rads in Parallel will help clean up the plumbing anyway.

Blocks in Series and Rads in Parallel....

Time to order a few more fittings.

I liked the XTs for the same reason.
SR-2
(13 items)
 
  
CPUMotherboardGraphicsRAM
Dual - Xeon x5650 EVGA SR-2 Dual - EVGA GTX 470 / GT 240 Corsair Dominator GT 12GB (6 x 2GB) 
Hard DriveOptical DriveOSMonitor
C300 / WD RE4 Dual / BluRay RW Windows 7 Pro Dual - Dell 2709W 
KeyboardPowerCaseMouse
MS 6000 V3 Silverstone ST1500 Mountain Mods Tower MS Sidewinder 
  hide details  
Reply
SR-2
(13 items)
 
  
CPUMotherboardGraphicsRAM
Dual - Xeon x5650 EVGA SR-2 Dual - EVGA GTX 470 / GT 240 Corsair Dominator GT 12GB (6 x 2GB) 
Hard DriveOptical DriveOSMonitor
C300 / WD RE4 Dual / BluRay RW Windows 7 Pro Dual - Dell 2709W 
KeyboardPowerCaseMouse
MS 6000 V3 Silverstone ST1500 Mountain Mods Tower MS Sidewinder 
  hide details  
Reply
post #304 of 353
Quote:
I liked the XTs for the same reason.
They actually could improve the Eheim pump flow rates with a better designed impeller and chamber. That's why I suggested adding a D5, though; because you keep the Eheims visible in the chamber and bury a D5 down in the radiators or something. Well, it's worth trying at least to see what kind of flow you get with just the two for starters; you can always add a pump later if you need it.


You mean sell some fittings, don't you?


[Edit] Oops, I forgot the last tube. 4-way split on the bottom, one goes up, one goes down, and two connect the rads.
Edited by Electrocutor - 8/2/12 at 9:13pm
post #305 of 353
Thread Starter 
It seems like with all the fittings I have, I never seem to have the right combination. frown.gif

I will try it and see what happens. I actually like the look of the D5 pumps with the trim kits and the right tops.

I wish there was room to hide some of the pumps, but with push pull fans and 2 radiators per compartment, there is no room between them.

I have had to take one of the ones out of the psu compartment so I could hook up the power cables to the psus.

Too much stuff stuffed into this giant box. frown.gif
SR-2
(13 items)
 
  
CPUMotherboardGraphicsRAM
Dual - Xeon x5650 EVGA SR-2 Dual - EVGA GTX 470 / GT 240 Corsair Dominator GT 12GB (6 x 2GB) 
Hard DriveOptical DriveOSMonitor
C300 / WD RE4 Dual / BluRay RW Windows 7 Pro Dual - Dell 2709W 
KeyboardPowerCaseMouse
MS 6000 V3 Silverstone ST1500 Mountain Mods Tower MS Sidewinder 
  hide details  
Reply
SR-2
(13 items)
 
  
CPUMotherboardGraphicsRAM
Dual - Xeon x5650 EVGA SR-2 Dual - EVGA GTX 470 / GT 240 Corsair Dominator GT 12GB (6 x 2GB) 
Hard DriveOptical DriveOSMonitor
C300 / WD RE4 Dual / BluRay RW Windows 7 Pro Dual - Dell 2709W 
KeyboardPowerCaseMouse
MS 6000 V3 Silverstone ST1500 Mountain Mods Tower MS Sidewinder 
  hide details  
Reply
post #306 of 353
This way is a little more efficient:



It would be better if you could use the G1/4" that are a straight shot, but if you're sending both tubes up or down it may be to close together; if it's going to get routed farther down the chamber then it'd be better to not use the 90 elbows.

I'm pretty sure I've seen people with push/pull and a D5 between them; you also have some room at the very front and back of the radiator mounts.

I've just remembered that these rads are 4-pass and go left-to-right or right-to-left. That is important to remember. Figure out which side the "In" and "Out" go to because you want the incoming water coming in at the side down-wind (from the fans) and outgoing up-wind. That will optimize the water/air transfer.
Edited by Electrocutor - 8/2/12 at 9:53pm
post #307 of 353
I dug this up from skinnee labs. I realize that you aren't using the tested CPU water blocks here, but the graphs should be pretty obvious that you want at least 1gpm through the CPU blocks because that is where heat transfer starts leveling out; about 1.5 gpm is optimal for where you start getting dimishing returns on higher flow.

post #308 of 353
Quote:
Originally Posted by Electrocutor View Post

I dug this up from skinnee labs. I realize that you aren't using the tested CPU water blocks here, but the graphs should be pretty obvious that you want at least 1gpm through the CPU blocks because that is where heat transfer starts leveling out; about 1.5 gpm is optimal for where you start getting dimishing returns on higher flow.

These dynamic systems become quite complex and difficult to predict as the level of complexity increases. just threw some more money at my SR-2 build in an effort to get it quieter while overclocked (this I definitely achieved). I was running a single loop with MCP35x2 pumps and 2 x SR-1 480s. In that scenario, i found that increasing the flow rates to the 1.5-2gpm range was where I hit the wall on diminishing returns. But up to 2 GPM, increasing flow rate reliably improved cooling---so much so that I initially used delta-T as a set point controller with the Aquaero driving the pump speed. Unfortunately, with 4x 480s and my fittings there was just too much restriction to get those types of flow rates even with the MCP35x2 pumping wide open. Sooo hundreds of dollars later, I opted for a single loop (shared reservoir) two pumping loop system that wound up like this:
Code:
   .---------------------------------------------------------.
   v                                                         |
 .---.                                             
 |Res|->(MCP35x2)->(480 XT45)->(480 XT45)->(CPU)->(CPU)------|
 |   |                            
 |   |->(XT)->(480 SR-1)->(480 ST30)->(MB)->(GPU)-----
 '---'                                               |
   ^                                                 |
   ---------------------------------------------------

CPU= Intel X5690 o.c. to 4.5GHz
GPU=EVGA GTX 680

Originally load was fairly well balanced in the pumping loops, but I changed out the GPU from 580 GTX Classfiied Ultra to a GTX 680 SC, saving about 75W on power. Now in this case, the GPU loop has less power demand than the CPU loop...at least until I add another card.

For simplicity sake I let the XT auto control it's flow rate--it seems to like 2.8 lpm no matter which combination I throw in front of it. That pretty much stays constant. As a matter of fact, the Aquastream XT is pretty stubborn, but that's another subject. Somewhat surprisingly, I found that with this new configuration, higher flow rates in the CPU loop were less efficient at cooling, and at a point much lower than the vaunted 4-6 lpm goal. The sweet spot for flow rate in the CPU loop seems to be somewhere between 1.8 and 2.4 lpm! Flow rates beyond about 3 lpm actually have worse performance with higher delta Ts (as in several degrees Centigrade by 6 lpm).

I'm still trying to understand this data...I need to graph it, add a few more data points, make it a bit more rigorous (OCD with a PhD in chemistry, bad combination).
post #309 of 353
Have you decided which CPUs you will be getting?
post #310 of 353
Thread Starter 
I have not picked out the rest of the components yet. Want to get all the water cooling parts and most of the wiring done.
SR-2
(13 items)
 
  
CPUMotherboardGraphicsRAM
Dual - Xeon x5650 EVGA SR-2 Dual - EVGA GTX 470 / GT 240 Corsair Dominator GT 12GB (6 x 2GB) 
Hard DriveOptical DriveOSMonitor
C300 / WD RE4 Dual / BluRay RW Windows 7 Pro Dual - Dell 2709W 
KeyboardPowerCaseMouse
MS 6000 V3 Silverstone ST1500 Mountain Mods Tower MS Sidewinder 
  hide details  
Reply
SR-2
(13 items)
 
  
CPUMotherboardGraphicsRAM
Dual - Xeon x5650 EVGA SR-2 Dual - EVGA GTX 470 / GT 240 Corsair Dominator GT 12GB (6 x 2GB) 
Hard DriveOptical DriveOSMonitor
C300 / WD RE4 Dual / BluRay RW Windows 7 Pro Dual - Dell 2709W 
KeyboardPowerCaseMouse
MS 6000 V3 Silverstone ST1500 Mountain Mods Tower MS Sidewinder 
  hide details  
Reply
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Build Logs
Overclock.net › Forums › Case Mods & Cases › Builds & Case Mods › Build Logs › [Build Log] CaseLabs STH10 - Tri EVGA - SR-X - Dual LGA-2011 - H20