Overclock.net › Forums › Cooling › Water Cooling › /Facepalm - GPU1 is 10-12* hotter in parallel
New Posts  All Forums:Forum Nav:

/Facepalm - GPU1 is 10-12* hotter in parallel

post #1 of 6
Thread Starter 
I guess I got a air bubble, or some other kind of particulate stuck in one of my MCW82 blocks, I noticed the other day that GPU1 was hotter than 2. So I blew through a couple furmark bench runs, and I see that GPU1 is 10-12c hotter than GPU2 in a parallel setup. I guess I know what my project is going to be on Saturday or Sunday...

I probably did not properly flush everything as I am still quite the n00b to water cooling, so I may also be taking my blocks apart and doing a full flush of both rads. Really annoying though as all the water is clear in the whole loop, guess something slipped by.

Before I climb into that, I wanted to see if there was any rational explaination to the temperature difference. After load is removed they are within 3* of each other at idle (#1 37c, #2 34c)
Downfall
(14 items)
 
  
CPUMotherboardGraphicsRAM
Phenom II 1100T BE @ 4.12 (265 x 15.5) @ 1.525v Asus Sabertooth 990FX 2x XFX 6870 Black Edition 8GB Corsair XMS3 9-9-9-24 1766mhz 
Hard DriveHard DriveCoolingOS
Corsair Force 3 Western Digital Black Edition Custom water cooling loop, started as Rasa RS360 Win7 x64 Enterprise 
MonitorPowerCaseMouse
Sony Bravia KDL-40S4100 (40" 1080P) Antec HCG-900 Cooler Master HAF 932 Logitech MX518 
  hide details  
Reply
Downfall
(14 items)
 
  
CPUMotherboardGraphicsRAM
Phenom II 1100T BE @ 4.12 (265 x 15.5) @ 1.525v Asus Sabertooth 990FX 2x XFX 6870 Black Edition 8GB Corsair XMS3 9-9-9-24 1766mhz 
Hard DriveHard DriveCoolingOS
Corsair Force 3 Western Digital Black Edition Custom water cooling loop, started as Rasa RS360 Win7 x64 Enterprise 
MonitorPowerCaseMouse
Sony Bravia KDL-40S4100 (40" 1080P) Antec HCG-900 Cooler Master HAF 932 Logitech MX518 
  hide details  
Reply
post #2 of 6
There are a lot of variables with water cooling unfortunately. I would check for air bubbles or particles. If that's not the case, then re-seating the block would be the next move, which is not always fun.

I run my 3 cards in parallel, under load they are all within a couple degrees of each other. This is the same for idle (atm 37, 36, 38). But I've never noticed a big of a difference than yours. Normally 3-5 degree difference under load though at worse.
post #3 of 6
If its air you can just pinch the tube before the card and release it. The sudden pressure surge pops the bubbles out. Also make sure the tubes going to both cards are as close to equal in length (and bends) as possible after the split since headloss due to friction in the tube mathematically will reduce flow through that card. If its not air take it apart and clean it out. And dont forget the possibility of just a bad TIM seat.
New beginning
(13 items)
 
  
CPUMotherboardGraphicsRAM
Ryzen 1700 Msi x370 Titanium  Aorus 1080 TI waterforce wb Gskill tridentZ 3200 
Hard DriveHard DriveCoolingOS
WD blue m.2 SSD Intel 370 SSD custom loop W10 
MonitorPowerCaseMouse
asus 24" Corsair RMx 850 gold Custom Define S cube Logitech G502 
Mouse Pad
Corsair mm300 
  hide details  
Reply
New beginning
(13 items)
 
  
CPUMotherboardGraphicsRAM
Ryzen 1700 Msi x370 Titanium  Aorus 1080 TI waterforce wb Gskill tridentZ 3200 
Hard DriveHard DriveCoolingOS
WD blue m.2 SSD Intel 370 SSD custom loop W10 
MonitorPowerCaseMouse
asus 24" Corsair RMx 850 gold Custom Define S cube Logitech G502 
Mouse Pad
Corsair mm300 
  hide details  
Reply
post #4 of 6
Thread Starter 
no good. Just drained and refilled my loop to see if there was a simple air bubble, and I have the same result after refilling. I guess its time for a full teardown.

It looks like the flowrate through the top block (the hotter one) is lower, the lines going into it had small air bubbles left after the bottom one had none left. Also looked like air bubbles that were flowing through the top block were just moving slower. I am hoping I am just seeing things, because I dont really want to crack these blocks open, they are only about a month or so old.

I'm going to tear down and attempt to reseat the block in question, perhaps a cat hair got in between the block and the gpu. In case anyone is wondering, I have two MCW82 blocks in parallel with this as the bridge.
Downfall
(14 items)
 
  
CPUMotherboardGraphicsRAM
Phenom II 1100T BE @ 4.12 (265 x 15.5) @ 1.525v Asus Sabertooth 990FX 2x XFX 6870 Black Edition 8GB Corsair XMS3 9-9-9-24 1766mhz 
Hard DriveHard DriveCoolingOS
Corsair Force 3 Western Digital Black Edition Custom water cooling loop, started as Rasa RS360 Win7 x64 Enterprise 
MonitorPowerCaseMouse
Sony Bravia KDL-40S4100 (40" 1080P) Antec HCG-900 Cooler Master HAF 932 Logitech MX518 
  hide details  
Reply
Downfall
(14 items)
 
  
CPUMotherboardGraphicsRAM
Phenom II 1100T BE @ 4.12 (265 x 15.5) @ 1.525v Asus Sabertooth 990FX 2x XFX 6870 Black Edition 8GB Corsair XMS3 9-9-9-24 1766mhz 
Hard DriveHard DriveCoolingOS
Corsair Force 3 Western Digital Black Edition Custom water cooling loop, started as Rasa RS360 Win7 x64 Enterprise 
MonitorPowerCaseMouse
Sony Bravia KDL-40S4100 (40" 1080P) Antec HCG-900 Cooler Master HAF 932 Logitech MX518 
  hide details  
Reply
post #5 of 6
Thread Starter 
alright just took it apart and re-seated the blocks, and it looks like after limited testing it was a bad seat on GPU1. I have identical idle temps now (36* at this time), and load values are within 2* of eachother (43 and 45 under furmark)

I know this is probably a common problem and no one is getting anything out of my posts, but I will update again in case we have any new comers who use the search function...
Downfall
(14 items)
 
  
CPUMotherboardGraphicsRAM
Phenom II 1100T BE @ 4.12 (265 x 15.5) @ 1.525v Asus Sabertooth 990FX 2x XFX 6870 Black Edition 8GB Corsair XMS3 9-9-9-24 1766mhz 
Hard DriveHard DriveCoolingOS
Corsair Force 3 Western Digital Black Edition Custom water cooling loop, started as Rasa RS360 Win7 x64 Enterprise 
MonitorPowerCaseMouse
Sony Bravia KDL-40S4100 (40" 1080P) Antec HCG-900 Cooler Master HAF 932 Logitech MX518 
  hide details  
Reply
Downfall
(14 items)
 
  
CPUMotherboardGraphicsRAM
Phenom II 1100T BE @ 4.12 (265 x 15.5) @ 1.525v Asus Sabertooth 990FX 2x XFX 6870 Black Edition 8GB Corsair XMS3 9-9-9-24 1766mhz 
Hard DriveHard DriveCoolingOS
Corsair Force 3 Western Digital Black Edition Custom water cooling loop, started as Rasa RS360 Win7 x64 Enterprise 
MonitorPowerCaseMouse
Sony Bravia KDL-40S4100 (40" 1080P) Antec HCG-900 Cooler Master HAF 932 Logitech MX518 
  hide details  
Reply
post #6 of 6
Good to hear a simple reseat worked.
New beginning
(13 items)
 
  
CPUMotherboardGraphicsRAM
Ryzen 1700 Msi x370 Titanium  Aorus 1080 TI waterforce wb Gskill tridentZ 3200 
Hard DriveHard DriveCoolingOS
WD blue m.2 SSD Intel 370 SSD custom loop W10 
MonitorPowerCaseMouse
asus 24" Corsair RMx 850 gold Custom Define S cube Logitech G502 
Mouse Pad
Corsair mm300 
  hide details  
Reply
New beginning
(13 items)
 
  
CPUMotherboardGraphicsRAM
Ryzen 1700 Msi x370 Titanium  Aorus 1080 TI waterforce wb Gskill tridentZ 3200 
Hard DriveHard DriveCoolingOS
WD blue m.2 SSD Intel 370 SSD custom loop W10 
MonitorPowerCaseMouse
asus 24" Corsair RMx 850 gold Custom Define S cube Logitech G502 
Mouse Pad
Corsair mm300 
  hide details  
Reply
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Water Cooling
Overclock.net › Forums › Cooling › Water Cooling › /Facepalm - GPU1 is 10-12* hotter in parallel