R9 290 Vapor X OC Issues

[george1976]

Hey guys
I was wondering if someone with more graphics card experience than me can help please.
My issue is:
One of the fans stopped working, replaced it with a very close match from China that is working as intended.
Replaced the thermal pads everywhere on the board, and paste on gpu, or so I thought.
Booted into windows, all looked fine, proceed to Furmark where I had problems:
Temp was jumping right away from 41>58>64>black screen.
Reboot, temp in idle fine, Furmark same issues.
Now I dismantled the card and found out that gpu contact with heatsink was about 35% with my amount of paste, even if thermal pads were the same thickness as Sapphire sent me.
I made that good, but now I can see the bios screen, goes to Windows and then video output dies on both monitors.
I removed the drivers with DDU, and in that low resolution I can get into W10 normally and all is fine.
As soon as I start installing the drivers from AMD, when the installation reaches the point where the screen will flicker to change resolution to 1920x1080, screen goes black there is no more video output from the card.
Now I am using the onboard video from the 2500K.
My question is: Because I didn't reboot very quickly when I got the black screen from Furmark did I burn a part of the gpu? Is there anything else I can try?

System is: Z68XP-UD3---I5 2500K---R9 290 VAPOR X OC---1TB HDD---EVGA SUPERNOVA G2 850W

Thank you

 

[Pulseammo]

On your second dismantlement have you made a mistake somewhere? Perhaps something from the heatsink shorting the PCB? What thermal paste did you use? Nothing conductive I hope like liquid metal.

You shouldn't use furmark to test GPU temperature, the load that it puts on the card is unreasonable compared to even a high demand game. Personally I use Unigine Valley and run for about 15 minutes to heatsoak the card and that's as hot as it will get in any normal use scenario. I suspect the missing piece of your puzzle might be that the thermal pads had bad contact with the VRMs, they were getting too hot at the card was shutting down even though your core temperature wasn't high. You should use GPU-z to monitor your temps and pay close attention to VRM temps as well as core temp.

 

[george1976]

I have Gelid GP Extreme 1mm on vrm's and memory chips, and on the back between backplate and pcb some aag thermopads that compressed nicely from 3mm. Those are the only ones able to short something but they are not electrically conductive, or at least up to 4Kv.
I don't know if it's possible for the bios to be corrupted? Intelligent Fan control wouldn.t work even before I started taking the board apart no matter the setting of the switch, and the fan that was broken wasn't stuck in any way, it would only start after the board reched 85C, so maybe the board burnt it?

 

[Davedree]

Not a fix but try installing the driver with just one monitor, and testing with just 1 monitor.
Hawaii has a problem in that it needs to run high memory speeds for dual display, or high refresh rate. When it does this it will run in a 2d clock mode but without enough voltage for the memory.

Another try would be use the 2500k for the display and install the amd driver, once installed, shutdown and try just 1 monitor in the amd card. If it black screens at bootup go back on the 2500k and look in device manager if the 290 is yellow flagged, then it may need atimdagpatcher to help it install on windows10.

 

[george1976]

I now have the card apart for a third look Nothing looks wrong, pads made contact everywhere and gpu has enough paste.
Before taking the card apart again windows was freezing forcing me to reboot even if nothing was connected to the Amd card, was just in the system, I was using onboard graphics.
I tried to reinstall windows just using the onboard video and the amd card just sitting in the system and Windows will fail to install at the last stage after getting ready etc.
Now I have taken the amd card out of the system completely and windows installed just fine and no more hiccups...….
Checked carefully the pcb, nothing.
Going to keep digging

 

[Pulseammo]

Seems like you've taken the right steps so far. In the past I've seen a couple of threads where there's a faulty GPU which works fine but doesn't allow the drivers to install otherwise it crashes the system. From memory they tend to be nVidia GPUs though and there never seems to be a solution, it's interesting that yours seems to have started after the second re-pad / paste... Would be good to know what causes the "can't install drivers" behaviour.

Not much help I know, but good luck finding a fix.

 

[hardkore]

I had this exact card and exactly the same problem. Bet you wont find this answer anywhere else!!

This problem started when AMD released the crimson drivers. What would happen is AMD lowered the base voltage in the drivers which caused the card at stock to black screen. Try installing Pre crimson drivers and see if that works. If it does you'll have to email Sapphire and ask for a revised BIOS.

The only way to fix that if it is indeed the drivers. This is why ill never buy AMD again, they just washed their hands of the fact the stupid drivers crippled my card.

 

[george1976]

I had this exact card and exactly the same problem. Bet you wont find this answer anywhere else!!

This problem started when AMD released the crimson drivers. What would happen is AMD lowered the base voltage in the drivers which caused the card at stock to black screen. Try installing Pre crimson drivers and see if that works. If it does you'll have to email Sapphire and ask for a revised BIOS.

The only way to fix that if it is indeed the drivers. This is why ill never buy AMD again, they just washed their hands of the fact the stupid drivers crippled my card.

Before I replaced the fan, card worked with the clocks and voltage lowered, I know it's an OC card but didnt liked the heat and fan noise, and was just fine like that with the games I played on 1080p.
It worked like that for a very long time without issues and with different drivers. So I don't think it's the drivers, same version, just me took the card out and back in.
The only variable was my mistake with the paste and trying to run furmark that crashed it and I believe burnt something, you would have thought it has protections in place....
At my last try, GPU-Z showed the card as generic windows display adapter, not even R9 290 lol, temps on gpuz sensors tab were ideal.......
I gave up for the moment, put old xfx 6950 back in, works OK so that rules out MB, memory, hdd etc wich I already tested from outside windows anyway.
Thank you for your advice.

 

[Shiari]

It doesn't need to burn to get destroyed (too much voltage can destroy without leaving a visible mark for example), and failure due to too much heat without actually burning may also be a cause of a component going bad. "Protections" can only do so much, such as measuring a temperature somewhere on a point on the GPU, whereas say another point on the same GPU die might've gotten a lot hotter because of no thermal contact there so any protection either didn't get the chance to kick in at all or it kicked in too late.

If you replaced the cooling correctly, used adequate paste etc and it won't work anymore with that my guess is you did overheat and destroy something in the process, probably by using Furmark. Furmark would be the kind of tool you'd use (and I never use it) once you know the cooling works well to test how well the cooling really works, and not to test whether a GPU works.

 

[george1976]

Furmark would be the kind of tool you'd use (and I never use it) once you know the cooling works well to test how well the cooling really works, and not to test whether a GPU works.

Hehe
That was my mistake, idle temps were excellent and instead of firing up a video, or a game, I went straight to Furmark.I guess this was an expensive learning lesson.
Now I can't play my games at the quality I want to and prices for a new card or a new build are not at the levels I would commit.
Was just wondering, if I replace only the graphics card would my 2500k overclocked to 4.1-4.2 be able to cope with games for another year?

thx

 

[Shiari]

Depends on the games and the resolution you're running them at, but mostly will be just fine yes.

 

[george1976]

Thanks for the quick reply mate
I have a 1080p monitor and looking to switch to a freesync 1440p one in the future.
Vega 64 was my idea, not because top of the line from Amd but because I don’t want to change the card for a few years.
Although not sure if the price will go down a lot more, Red Devil one is some £450-£460 at Overclockers

 

[Shiari]

Vega 64 drives a freesync monitor of pretty much any resolution very well, I'm running 3440x1440 up to 75Hz, and my housemate has a similar setup but up to 80Hz. We're both under water though.

Your 2500K could do with a bit of a boost, you should be able to push that to 4.5 with a decent cooler and that would help as it would no doubt not be able to get the most out of every single game, some games do like having lots of cores which you just don't have plus it's quite an old gen now. Mostly you'll be fine though, even if it'll bottleneck things a bit sometimes ... but you can't avoid having a bottleneck somewhere.

If you do get the Red Devil, beware that it's quite large and heavy