Sign in with
Sign up | Sign in
Your question
Solved

R9 290 Crossfire: Works with stock cooler, but doesn't with waterblock.

Tags:
  • Water Cooling
  • Crossfire
  • Gigabyte
  • Sapphire
  • Graphics
  • Graphics Cards
Last response: in Graphics & Displays
Share
June 18, 2014 7:57:20 PM

Hello readers,

I will try and make this as concise as possible to help make troubleshooting easier :) 

Motherboard: Maximus Hero VII
CPU: i7 4770k @ 4.3Ghz, 1.25v
RAM: G-skill 16Gb (2x8) @ 1600
PSU: Seasonic Platinum 1000watt
GPU: R9 290 (Sapphire), R9 290 (Gigabyte Windforce), stock clocks.
Watercooling: 2 x GPU block (XSPC Razor), 1 x CPU block (XSPC Raystorm)

I was initially running the Sapphire 290, recently decided to add a second. Before adding the Gigabyte card to the loop, I made sure it worked. It did. Removed Gigabyte card, re-installed Sapphire, booted. Re-installed CCC 14.6 after drive sweep. Rebooted with Sapphire card.

At this point I prepared the Gigabyte card for the loop, and attached it into the second red slot on the MB. Booted with no success in CCC, GPU-z and device manager (All only showing the primary card). I thought it might be a seating issue, re-seated card, tried again. No success.

Removed Sapphire card from loop, only ran Gigabyte card in primary. Black screen. Cleared CMOS, tried again. Black screen.

Thought I damaged the card during the block install. So I went back to square one. Removed the waterblock, installed stock cooler (Windforce). Running short on patience, I persisted to stick the card in the second red slot on the MB with the Sapphire card still in place. Booted.

And to my somewhat joyous pain, it recognised the card and Crossfire was working. I could liken this feeling to touching yourself whilst crying.

So now I'm back to running around in circles. Any and all ideas surrounding troubleshooting this mindf*ck are welcome.

Some thoughts: Gigabyte card worked (and continues to work) with provided heatsink. It's doing so as I type this, in Crossfire. With waterblock attached, doesn't even post. It's a poetic injustice.

Help please :|

More about : 290 crossfire works stock cooler waterblock

a c 171 U Graphics card
June 18, 2014 8:26:42 PM

Is it checking for the current draw from the fan(s) spinning? Try (just for testing) having the old cooler still plugged in but hanging loose.
m
0
l
June 18, 2014 8:54:16 PM

That wasn't it. Still posts in Crossfire without header.

Edit: I will remove the cooler from the pcb and have it hanging loose with header unplugged, then boot. Might help narrow it down further. I'll get around to trying that later as I've exhausted my patience and need to walk away before I mess it up completely.

If it doesn't work then, would it be a contact issue between pcb components and heatsink? which is why waterblock attached also failed...
m
0
l
Related resources
a c 171 U Graphics card
June 18, 2014 9:01:12 PM

Damn. That's really weird.

Any other connectors that get changed?

Only thing I can think of is that the change in mounting pressures and torsion is causing a bad contact to work/not work...
m
0
l
June 18, 2014 9:07:42 PM

Funnily enough, the Windforce heatsink/fans are attached to the pcb with only 6 screws and 1 fan header - on the simple side when it comes to assembly/dis-assembly.

I too thought it was seating issue (it's always the simplest thing). Thought it might have been the flow bridge flexing the contacts. But it didn't post by itself without the second gpu :/ 
m
0
l
June 18, 2014 9:11:32 PM

Someone Somewhere said:

Only thing I can think of is that the change in mounting pressures and torsion is causing a bad contact to work/not work...


Waterblock too tight against components? My experience is one would generally kill the GPU in that instance.

m
0
l

Best solution

June 19, 2014 12:22:19 AM

Tried it all over again. Attached the waterblock. Then stared it for a while. I realised that a row of capacitors on the pcb are actually soldered on a few mm higher than they probably should be. As a result, the waterblock, once tightened on applies pressure to them. So I applied adequate tension to the 4 screws surrounding the processor, and virtually no tension to the right hand side of the block where the capacitors are raised. I inserted a second washer to all holes for the back-plate as a precaution.

Plugged it all in, attached flow bridge after seating both cards, ran it dry to test. It worked. I have no idea what the exact issue was, might or might not have been the components, or over-tension, possibly even seating. After going through the process more times than I would have liked. Success.

Cheers for the brainstorm SomeoneSomewhere.
Share
!