video tdr failure on rx480

RiggeRigge Member Posts: 3
edited February 2017 in Mining
Hi, I need some tips how to get my card going again.
I flashed my card with atiflash and modded my bios with Polariseditor. I changed the timings to 1750/2000 and memory clock to 2235 and it was stable for some hours and then crashed. after it crashed i didn't have time to reflash it again. so I shutdown the rig and flipped the Vbios switch to "gaming mode" a bios that i haven't modded. And it worked great for 2 days but after that it crashed again. then when i set it as a display card i only came to windows logging then it crashes and says it's a error "video tdr failure (atikmpag.sys). so I reinstalled windows and it worked good until I installed the drivers and then gets the same message. but if i install another card (with modded bios) as a display card its works again.

Things i tried:
changed risers
installed newer and older drivers
install the card direct on the mobo.
run linux (pimpOS)
flash original bios with vbios switch in "silent mode" (bios i originally flashed)

My setup:
Msi z97 gaming 5
Intel celeron g1840
4 gb of ram
Evga gq 1000w
sapphire nitro+ RX480 8gb (Samsung memory) x3
win 10/PimpOs
sg-miner/claymore

I hope you guys have some tips o:) or I think I need to RMA it

Comments

  • yodiyodi Member Posts: 257 ✭✭
    edited February 2017
    did you install driver pack for mobo ?
    what bios are you using if modded ?
    try lower to 1110 the core clock to test BSOD generaly result of high core
    how many gpu ?
    try an other ram if you have one

    sgminer is really stable on pimp
  • workwork Member Posts: 2,084 ✭✭✭✭
    If it isn't working on stock BIOS as thr primary and ONLY card, plugged directly into the mobo, then you have a faulty GPU. RMA time in that case.
  • RiggeRigge Member Posts: 3
    yodi said:

    did you install driver pack for mobo ?
    what bios are you using if modded ?
    try lower to 1110 the core clock to test BSOD generaly result of high core
    how many gpu ?
    try an other ram if you have one

    sgminer is really stable on pimp

    Yes I installed Mobo drivers. I running core at 1200 (standard). I'm running 3 GPUs but when I tested the faulty card I only plugged in that card. I do not have any more ram to test with but I can buy more.
  • cidmocidmo Member Posts: 378 ✭✭✭
    edited February 2017
    yea i have got stuff like that from time to time on my main machine
    partly i blame the MSI mobos, partly i blame having so many HDD, SSD, USB HDD, etc
    but mostly i blame windows 10
    from what i have gathered at this point some MSI mobo will not run 1x powered risers in the 2nd pcie x16 slot (sli/CS slot)
    on gigabyte mobos this can actually screw up the bios chips somehow when all pcie are filled
    but for the most part that error "video tdr failure (atikmpag.sys) has to do only with the driver not the actual card
    u could even try a different driver version - saw u tried this did u fully uninstall things like msi afterburner too?
    also idk how ur uninstalling but if u havent yet when super stuck try DDU display driver uninstaller
    im assuming it doesnt happen in linux?

    i should add what fixed it for me is i dont mine on my main machine anymore
    could be just windows 10 just doesnt work well when mining certain algos and doing other things idk
    problem only appeared after i switched to windows 10, windows 7 was totally fine
  • workwork Member Posts: 2,084 ✭✭✭✭
    @cidmo windows 10 is perfectly stable for mining. And I think it is unlikely a different driver version will help the OP.
  • cidmocidmo Member Posts: 378 ✭✭✭
    edited February 2017
    work said:

    @cidmo windows 10 is perfectly stable for mining. And I think it is unlikely a different driver version will help the OP.

    lol thats a pretty bold statement considering all the different types of hardware out there
    and wrong considering i havent blue screened once since i stopped mining on my main machine as stated above
    and atikmpag.sys is the driver so a BSOD would point to THAT
    even if that isnt that actual cause its part of the cause and trying something like a catalyst driver might shed some light
    a friend had the same thing happen to him when a new version of intel chipset drivers came out for his machine, the 2 together did not work so he tried a catalyst driver and it worked
    and in my case that happened to me over and over again it WAS the driver that was having problems in win 10 but not win 7 on multiple cards but not all
    same exact error as him, same time frame, same OS, same MOBO manufacturer, all i did to fix it was stop mining nothing else

    really we have no idea until he says if the card worked in linux or not
    all he said is he tried linux pimpOS
    if the card doesnt mine in linux then ofc it should be RMA but im just trying to save someone $25 shipping
    i would lean more to some program like afterburner with some sort of profile is loading on startup
    thats why i recommended DDU and uninstalling afterburner if he hasnt
    Post edited by cidmo on
  • workwork Member Posts: 2,084 ✭✭✭✭
    edited February 2017
    Chipset drivers really aren't comparable to display drivers. If you can't mine on your main machine then it isn't stable - period. If it doesn't crash when not under load that is because it's not under load! There is a reason people test overclocks with intense benchmarks - stable at idle is not neccesarily stable overall.

    "All I did to fix it was stop mining" - you didn't fix anything... you just removed the thing triggering the symptoms.

    AMD driver crashes rarely have anything to do with a faulty driver. It is almost always faulty hardware or bad clock/volt mods.
  • RiggeRigge Member Posts: 3
    Now I have tired win 7 and I can mine with claymore miner for sometimes 5 min and sometimes one hour but I get the same BSOD "video tdr failure (atikmpag.sys)". I did also a fresh install of win 10 and i get same BSOD when i tried to log in. also tried a fresh install of linux and it crashes after 2 min of mining. So now I am gonna RMA it... while I wait for the shop to replay to my RMA I'm going to try it on of my old computer to see if it going to help or if I get the same error.

    Anyway, thx for the help ! :)
  • cidmocidmo Member Posts: 378 ✭✭✭
    edited February 2017
    work said:

    Chipset drivers really aren't comparable to display drivers. If you can't mine on your main machine then it isn't stable - period. If it doesn't crash when not under load that is because it's not under load! There is a reason people test overclocks with intense benchmarks - stable at idle is not neccesarily stable overall.

    "All I did to fix it was stop mining" - you didn't fix anything... you just removed the thing triggering the symptoms.

    AMD driver crashes rarely have anything to do with a faulty driver. It is almost always faulty hardware or bad clock/volt mods.

    for him it wasnt the chipset drivers per say as i said
    it was the combination of the 2 the display driver and new chipset driver all he had to do was use a catalyst driver thats it, that card is still running today no problems in fact has one of the highest asics ive ever seen
    obviously something with the new chipset driver was effecting the interaction with the card itself thru the display driver
    mining and stable mining are 2 different things benchmark tests and stress tests are 2 different things
    mining for 2 hours is mining, crashing 2 hours after load signals a conflict not a hardware issue
    i ran a IT help desk for 2 years lol seen pretty much everything in that not including building their network from scratch
    ur really gonna argue over the semantics of the word fix?

    "AMD driver crashes rarely have anything to do with a faulty driver. It is almost always faulty hardware or bad clock/volt mods."

    wrong its almost always a software conflict
    drivers can have conflicts with other hardware drivers, hence the need to rollback to previous drivers sometimes
    not to mention sometimes those conflicts can be brand specific so in essence the driver is only faulty for specific hardware but not faulty for all
    rarely is it ever hardware u should know this as u probably have more than 100 GPU like me
    i have returned 2 GPU ever and RMA 1
    and 1 PSU
    and the 2nd part of ur comment is exactly what i said above

    i guess the main thing is cant say windows 10 is stable in general let alone for mining if u understand kernels
    unless u have custom kernels for every piece of hardware thats impossible, even then


    @Rigge glad u figured it out sry it turned out to be a bad card but its gotta happen to someone
    i remember when the 480s came out and we were all test modding them a few guys had to return theirs after modding and mining for a similar reason to urs
    they figured they somehow messed up the memory controller with too tight of timings and even flashing back to stock bios didnt do anything at that point ofc
    i have a nano thats like that i think the mem controller is jacked from setting the freq too low and not running it with enough volts
    Post edited by cidmo on
or to comment.