R9 380 hitting 99C, Lowering memory or clock by 1hz will decrease performance by 99%, help needed

RVEGARVEGA Member Posts: 19
edited February 13 in Mining
Hello! i just bought the 6 R9 380 CHEAP, i've only got 4 out of the 6 cards to work (computer won't find the other 2)...
when i first start claymore they all mine at 18.6 MH/S (nice, great!) however GPU 1,2,3 are hitting 90C!! Card 4 is stable at 65C @ 18.3 MH/S

if i lower the Core clock from 1040 to 1036 my temps get down to around 48C (??) and are mining at 6 MH/S.....

anyone got any clue on how to fix this ?
Post edited by RVEGA on

Comments

  • RVEGARVEGA Member Posts: 19
    Okey, i found a problem, if i change my core clock to ANYTHING below 1040, it will automatically be set to 300.....
  • LippyCKLippyCK Member Posts: 24
    that overheat is probably because of bad thermal paste or radiator is clogged with dust. If they are not under warranty take them apart clean them and change paste, if possible some high quality paste. i personaly use Thermal Grizzly hydronaut for most of my stuff and its great.

    Use Crimison beta blokchain drivers, your hasrate should go up for 2 or more Mh/s

    with two cards that are not working, they might be dead or it might be your risers, try them in different risers.
    you can try ms dos version of atiflash on ms dos bootable usb and check if you can find those cards there.
    if you find them there it should be bios problem and simply flashing bios from one of working cards could help.
    You can also probably find bios for your card on TechPowerup search for "vga bios collection"

    or if none of the above works on those 2 cards you can try to bake them in oven, watch few videos on youtube and try it. that sometimes works if cracked soldering is problem, but its rarely long term solution, even if it works its for couple of weeks to couple of months maybe
  • RVEGARVEGA Member Posts: 19
    edited February 13
    Hey, Thanks for your answer!
    i do think it's a "software" problem, check this out.




    by just changing the core clock by 2Hz my MH/S drops by 90%, and as you can see it get's stuck on 300 Hz instead of 1038.

  • LippyCKLippyCK Member Posts: 24
    yeah, that is bios issue i think, if your bios is moded then they behave like that,once you change core clock or core voltage in bios any change on core in afterburner under value set in bios they will throttle to 300mhz, at least in my experience,or if your bios dont have other states other than 1040 and 300 mhz. but your overheating shouldnt be a software problem, since you say those 3 are overheating and 4th is ok, only software problem could be too high voltage in bios.
  • RVEGARVEGA Member Posts: 19
    ah okey, so i could just take the BIOS from card 4 (the one that works great) and put it on the other 3 cards ?
  • LippyCKLippyCK Member Posts: 24
    Yeah, that should work if too high voltage is problem, but im still quite sure problem is bad thermal paste or bad heatsink contact(maybe screws are too loose), or maybe thing that just came to mind, maybe your cards are too close to eachother, if gpu4 is card with fans drawing free air and other cards are too close they draw very hot air from back of the card in front of them(at 65C on core back side of card is around 50C) i keep my card around 7.5cm(3inch) apart.
Sign In or Register to comment.