R9 290x and 390 Black screening

jamiec79jamiec79 Member Posts: 20
I would very much appreciate peoples assistance.

I'm having problems with the black screen crash whist running QTMiner for Ethereum.

I'd been running an MSI R9 290x on windows 7 64 bit with amd 14.12 drivers and installed open cl.

I added a second GPU (MSI R9 390) and initially the rig ran at 55-58M/hs which i was pleased with. A day or two later the computer black screened a short while after starting to mine and more recently it black screen once at Windows desktop.

I've reset the MSI Afterburner settings to default.

I've tried uninstalling and reinstalling AMD drivers from 14.12 to 15.12 and i can't get the rig to stop blackscreening.

I did wonder whether it was because there is no shared driver from AMD that will run both GPU's without crash.

Temps are all within range.

I've seen so many threads where people have suggested undervolting, overvolting, powertuning, adjust clock speeds, I've tried them all and in combination with each other

In terms of the bat file to run QT miner i've kept it very simple, qtminer.exe -s eu1.ethpool.org:3333 -u my wallet address.rig1 -G

I'm at a loss to see what else it could be, i've read almost every thread available and what works for others doesn't seem to work for me?

I'm running two 600W PSU's (one does the MOBO, SSD and R9 290x) and the other is running only the R9 390.

I purchased a dual psu 24 pin splitter which is used to kick the second PSU into life.

I apologise if this has already been solved in another thread, if so i haven't seen it.

Any help would be greatly appreciated!

The overall aim is to test the process and if works but 5 x R9 390's which may remove the driver challenge i currently face running a R9 290x and R9 390...


  • LP12LP12 Member Posts: 40
    try one card at a time...
  • jamiec79jamiec79 Member Posts: 20
    Thanks for replying. Tried that and both cards work on their own. When I then start then both I get black screen. I did look at the event viewer and there was a number of critical error 41, which I think is a kernel problem. All my temps are fine. I wonder whether the psu splitter I'm using is effecting the power of the second psu, effectively under powering it?

    Any thoughts would be greatly appreciated.
  • rdnkjdirdnkjdi Member Posts: 135 ✭✭
    edited April 2016
    What are you using to power your cards from the PSU? I've had similar problems and it had to do with the 6 pin to 8 pin converter cable I was using. Or the splitter from one six pin to two 8 pins having a bad pin or not being substantial enough.

    I've also had this issue when there was a problem with the romex cable supplying the riser.
  • jamiec79jamiec79 Member Posts: 20
    I'm using an 8 and 6 pin connector from the PSU but taking from two seperate cables from the psu rather than using the 8 and 6 pin on the same cable. I have a seperate power supply for each gpu.
    I wanted to check but even though the gpu is powered by the psu should I also be connecting the powered pci molex connection to the psu as well? I'm going to spend more time tonight trying not to establish what the critical error 41 means in the event viewer, it references kernel error, but looking into that it looks like a very general error and difficult to pin point the reason behind it....
  • LP12LP12 Member Posts: 40
    are you using powered risers? if so you need to plug the molex from the riser via a converter cable direct to your PSU, you can then "lose" the SLi/Xfire connectors to the mobo, thus taking strain off the mobo and obviously on top of that the 390's need a pair of pci powered cables to the PSU... if you've done all this maybe try measuring the rails out of the PSU? this maybe might not be very useful as they'll be quiescent and not pulling anything... BTW PLEASE DO NOT GO POKING YOUR MULTIMETER PROBES INTO THE PSU ITSELF OR THE MAINS CONNECTOR - THESE VOLTAGES ARE LETHAL
  • jamiec79jamiec79 Member Posts: 20
    edited April 2016
    Good point

    both cards are powered by the pcie 8 & seperate 6 pin direct to the psu... (completely seperate cables rather than using the same wire with both on). They are powered risers.... I'm not using the power via molex. so to clarify do the 390 and 290x both need an 8 & 6 pin connectors along with connecting the molex on the powered riser?? Worried I might fizzle something.... Admittedly when I mined for other coins with the 290x I only ever connected the pcie 6&8 pins directly to the psu. Even though the riser has the seperate molex I never used it and I didn't once have a black screen or if I did it was because I'd pushed the intensity/clock settings up slightly too far...

    I'm thinking I'll check the ram, swap the psu (although both fire up) check the 24 pin splitter (which is to fire the second psu/gpu). Reinstall the drivers I have tried everything from 14.12 to 16.3 (think that's the most recent Crimson one) from AMD and all drivers between. I use ddu in safe mode to remove previous driver... I could try flashing the gpu with the most recent vbios.
    I am at a bit of a loss really...

    Problems only started when I added the second new 390 8gb, brand new be quiet 600w psu to power it. I don't know whether the two cards (albeit) from the same family don't like being on the same driver set...

    Any comments/suggestion very much welcome!
  • LP12LP12 Member Posts: 40
    black screen smacks of power issues to me; BlueSOD maybe ram, cpu, drivers; but yes all these diagnoses are a process of elimination so go ahead and rule out the components you cited in the last post. you're using 2x pci power for each card to psu which is fine, but i would still try molex to sata power from the riser to the psu and not use the sli molex's on the mobo; oh and just one more word of warning; if you're doing all this fiddling make sure you're grounded with a good anti static bracelet as the cards, risers, mobo and ram are very sensitive to static - but please don't ground yourself whilst the rig is powered up... :-(
  • jamiec79jamiec79 Member Posts: 20
    I'm trying a fresh install of win7 sp1, then install one card at a time and see which if any causes the problem.

    I'm still a bit confused won't I short the gpus they each have an 8 & 6 pin connector from seperate psu's and then connected the powered riser molex to the psu (as you suggest) at the same time? Currently the powered riser is effectively unpowered via the molex. I just don't want to make a mistake and fry either/both my gpus.

    I appreciate your response it's good just to discuss this as it's very frustrating.
  • jamiec79jamiec79 Member Posts: 20
    Oh and I haven't used the sli/molex connector on the mobo, I never have... Just thought the two pci power from seperate cables on the psu would suffice? And two seperate psus running the two seperate cards.

    Maybe this is where I've been going wrong I should have also connected the powered riser molex to the psu via one of the specific cables? I just didn't want to kill the gpus lol.
  • LP12LP12 Member Posts: 40
    i think you need at least the sli molex's on the mobo connected to the psu at a very minimum with multiple cards...
  • jamiec79jamiec79 Member Posts: 20
    Ok I'll try it when I get home, so I still leave the 8 & 6 pin attached to each gpu directly to the psu. Then connect the sli molexs on the mobo to the psu aswell. Thanks again for responding
  • jamiec79jamiec79 Member Posts: 20
    Thank you I'll try tonight :)
  • LP12LP12 Member Posts: 40
    yes try that; at least it gives you the piece of mind of not connecting the riser molex to the psu; i can quite understand your reticence at not wanting to fry a expensive video card...
  • LP12LP12 Member Posts: 40
    oh and yes, you must leave the 6/8 pin pcie's connected directly to the PSU - these are what the card derives most of their power from on the 12v rail; the sli's are subsidiary power connections...
  • jamiec79jamiec79 Member Posts: 20
    No joy, I've tried installing win 10 to see if that helps... I'm trying to get one card at a time working to see if that helps... Using amd 15.11 drivers... Only thing now is when I open afterburner I can't adjust the over clock settings like I used to... 3 steps forward 4 steps back :(
  • LP12LP12 Member Posts: 40
    try eliminating the risers; how many x16 lanes you got on the mobo?
  • jamiec79jamiec79 Member Posts: 20
    Is it worth flashing the vbios on the gpu? I'm running out of ideas...
  • jamiec79jamiec79 Member Posts: 20
    Just one x16 and five x1
  • LP12LP12 Member Posts: 40
    single x16 riser in x1 slot with one gpu, power up, test; other card in same slot with same riser, power up, test; then repeat with same slot with other riser, both cards - this eliminates the risers. then take one card and one riser and try in all lanes - this eliminates the lanes. keep 2xpcie to psu from cards and both sli moxlex's on the mobo connected throughout the tests.
  • coolrcoolr Member Posts: 7
    Guys you gotta try this bitcoin-mining site there giving away 15hz just for signing up its so insane, there practically giving away free cash. Heres the link https://hashocean.com/signup/?rid=435264. P.S you can thank me later :)
  • Adrian234Adrian234 Member Posts: 15
    Do you use msi afterburner ? If yes, uncheck "syncronize settings for similar processors" , that may force them to move to same voltage / frequency and may not be stable.
  • jamiec79jamiec79 Member Posts: 20
    I do have msi and I'd already unchecked it. Managed to get the r9 390 card working and Windows 10. I have dared add the second card as that is where is all went wrong before.

    I'll try again today to see if I can get the 290x running. Thank you for your thoughts! I appreciate the support.
  • jamiec79jamiec79 Member Posts: 20
    It's official I can't get the r9 290x and 390 to work off same mobo... The mobo sli and pcie 6&8 pin are connected. I run each independently no problem, same time black screen. Could it be ram? The virtual memory size? I currently on have 4gb ram installed, but thought that was enough? It could be the 24pin splitter cable, which I use to jump the 2nd psu and gpu is faulty...
    But I'd rather replace ram without knowing for sure or bypass the 24 pin to jump the second psu and replace them both with 1000-1200w psu... But for a good one its around £150 ish... The final option is to buy another r9 390 hoping that will work as it's exactly the same card rather than mixing the 290x and 390 msi cards.... I'm running out of options, other than stop wasting money. Thoughts?
  • LP12LP12 Member Posts: 40
    i've re-read the whole thread carefully; you said originally the rig ran briefly with both cards though? did anything else happen, or would precipitate when it started black-screening? i can't for the life of me think it's lack of ram on the mobo; but faulty ram? maybe? hell, can you force win 7 into 2GB? if you can you can eliminate each DIMM... 24-pin splitter problem? go back to my earlier post and try measuring the rails on the psu running the 390 - if you're not seeing anything on the 12v rails here it could be the splitter; but duh me, can't you just see the 2nd psu working (or not), if the fan spins up? i'll think on it some more tomorrow; it's late here (1amET)...
  • jamiec79jamiec79 Member Posts: 20
    When both connected both start up fine, fans spinning. Then almost immediately after I begin mining I get black screen.. I have noticed that one card tries to use 16. ... And the other 15.12. If I try to change them to both be 15.12 it black screens... Everything else continues running and not screen output... I've got a add2psu jumper circuit thingie arriving next day or two to eliminate the splitter, then I could try buying a brand new pair of riser cables.. Might get powered and unpowered... I'm not using the molex from the riser cable so seems pointless buying with... But I know it works, just want to
    eliminate the cheaper elements before working up to a bigger chunk of ram and or a 1200w psu... Thank you for taking the time to read my elongated thread. It's greatly appreciated !

    The 390 by the way is purring at a steady 33M/hs :)
  • LP12LP12 Member Posts: 40
    really sorry i can't figure it for you jamie; i'm normally good at this sort of troubleshooting - lol - hmmmm, interesting the driver versions are different on each card; is that normal? i don't know too much about multiple cards and the software side of things... sounds a bit fishy to me - maybe might it be a good idea to ask in the MSI/Sapphire/ATI forums to see if someone a bit more knowledgeable than me knows if that's ok?
  • jamiec79jamiec79 Member Posts: 20
Sign In or Register to comment.