1. OCAU Merchandise now available! Check out our 20th Anniversary Mugs, Classic Logo Shirts and much more! Discussion here.
    Dismiss Notice

RX 570 issue - drivers or hardware?

Discussion in 'Troubleshooting Help' started by pwlm1975, Jul 15, 2020.

?

What is my next step for this issue?

  1. Bite the bullet at buy a new card?

    2 vote(s)
    40.0%
  2. Try to get replacement motherboard (Skylake/Kaby Lake range)?

    1 vote(s)
    20.0%
  3. Appeal to the Gods by sacrificing a goat?

    2 vote(s)
    40.0%
  1. pwlm1975

    pwlm1975 Member

    Joined:
    Apr 18, 2016
    Messages:
    579
    Location:
    COVID Central, Southside Brisbane
    Hi everyone,

    I hope you're all staying safe at the moment. I have a Sapphire RX 570 Nitro+ 4GB, and have started to experience weird frame rate issues which are inconsistent.

    About a month ago I noticed FPS fluctuating in many games, including NFS, Sims 4, Far Cry 4 (this one was very obvious), and not only was the frame rate being hit, but the FPS would fluctuate severely, moving up and down by up to 30fps in about 1/2 second increments. I also noticed the frames were lagging frequently, it looked like someone was speeding up and slowing down the game. I used FRAPS, AMD Overlay and the myriad game launcher's FPS reporting software and kept getting the same result. Using Unigine Valley I find the same result - wildly fluctuating FPS and very obvious stuttering. I start to deduce my card may be dying (ex-mining card that I paid next to nothing for 3 years ago).

    Using DDU I uninstall the drivers and reinstall them, and voila! The problem is fixed. FPS go back to where they were expected. Then I run Valley and get good results and consistent FPS there as well. However, if I turn the machine off and back on later, the FPS and frame lag issues start again.

    What I have done system specs are in sig):

    1. Updated Windows;
    2. Tried newest Windows certified AMD drivers, both with and without the fancy software;
    3. Updated all chipset drivers;
    4. Tried 2 old driver suites (2019 and 2017 post-RX570 launch) both with and without the fancy software;
    5. 2nd PCIe slot;
    6. Different HDMI cable;
    7. BIOS settings - removed all overclocks (including RAM now running at 2133MHz), made sure graphics are set to PCIe output, and set the PCIe version to 3.0 (was Auto when the issue started)
    8. Different PCIe power cable (PSU is an Antec 650W Earth Pro something something, around 12 months old);
    9. Scanned for viruses using both AVG and Defender;
    10. Reinstalled Windows from scratch.

    Really at wits end why a fresh install of drivers would work, then not work within minutes. Has anyone else got any suggestions? I don't have a second power supply to check with (not one with the requisite 8 + 6 pin plugs, or with enough juice) so I can't test it out on a second machine without removing the PSU. Results below from Valley at High graphics/DX11/1080p/2 x AA which matches results prior to this issue.

    https://imgur.com/uwD2vAH

    Please if anyone can give me some pointers I'd be very grateful. TIA.
     
  2. Myne_h

    Myne_h Member

    Joined:
    Feb 27, 2002
    Messages:
    10,888
    Download GPUZ

    Monitor the temperatures.
    My money is on it hitting ~100C and slowing down.

    IF so: replace the thermal grease, and clean the dust out the heatsink.

    Another possibility is that it is getting stuck at 300mhz which is a bug I came across - but that wasn't in gaming, so I don't know if it's possible.
     
  3. demiurge3141

    demiurge3141 Member

    Joined:
    Aug 19, 2005
    Messages:
    2,158
    Location:
    Melbourne 3073
    Sounds like overheating.
     
  4. OP
    OP
    pwlm1975

    pwlm1975 Member

    Joined:
    Apr 18, 2016
    Messages:
    579
    Location:
    COVID Central, Southside Brisbane
    Hey thanks all. I just ran Valley after putting the card onto the 2nd PCIe slot again and first run from cold boot the FPS was consistent and much more like what I'm used to seeing. Temps in HWInfo64 and GPUZ showed the card hit 69 degrees. What is odd is that the fans didn't report spinning at all but I am running the test again, and watching it spin when it gets to 60 degrees as per the fan curve.
     
  5. OP
    OP
    pwlm1975

    pwlm1975 Member

    Joined:
    Apr 18, 2016
    Messages:
    579
    Location:
    COVID Central, Southside Brisbane
    Second run after 30 second shutdown, visually it's smooth as it should be.

    Valley score (1080p high detail/2 x AA/DX11) 3389.

    GPU usage 100%
    GPU chip power draw 167.7W
    Max temp 72 degrees (29% fan speed)

    So possibly a PCIe slot power delivery issue?
     
  6. Myne_h

    Myne_h Member

    Joined:
    Feb 27, 2002
    Messages:
    10,888
    Could have been a shitty contact between the card and board - which might make it run at <16x lanes, or pcie1/2 speed.

    Pulling it out and reseating it might have fixed it.

    Mobo model?
     
  7. OP
    OP
    pwlm1975

    pwlm1975 Member

    Joined:
    Apr 18, 2016
    Messages:
    579
    Location:
    COVID Central, Southside Brisbane
    It's an Asus Z170-P of 2016 vintage. I can run it in the lower slot no issue, it does not seem to have affected 1080p performance which is what the monitor is anyway. That card is a heavy mother so I guess it's possible it's caused issues through weight. I can use some LEGO to prop the card up and support it in the second slot.
     
  8. Myne_h

    Myne_h Member

    Joined:
    Feb 27, 2002
    Messages:
    10,888
  9. OP
    OP
    pwlm1975

    pwlm1975 Member

    Joined:
    Apr 18, 2016
    Messages:
    579
    Location:
    COVID Central, Southside Brisbane
    Hmmm, I'm confuzing myself now. The slot it's currently is the black one (near the bottom). GPUZ reports running at x4. That slot reported the 3389 result from the Valley benchmark.

    The top slot (grey) had a lot of flex on the outer edge nearest to the release mechanism when I last reseated it.
     
  10. Myne_h

    Myne_h Member

    Joined:
    Feb 27, 2002
    Messages:
    10,888
    I'd try it in the top slot again and see if you can tighten the backplate better. You can usually use 2 screws.
     
  11. OP
    OP
    pwlm1975

    pwlm1975 Member

    Joined:
    Apr 18, 2016
    Messages:
    579
    Location:
    COVID Central, Southside Brisbane
    Now running on the top (grey) slot, GPUZ reporting x16 and running Valley benchmark gives result of 3357 (margin of error) and 68 degrees/32% fan speed.

    Edit: just saw your reply Myne_h. Screws are about 1/2 turn after initial contact.
     
    Last edited: Jul 15, 2020
  12. Myne_h

    Myne_h Member

    Joined:
    Feb 27, 2002
    Messages:
    10,888
    *shrug*

    Chalk it up to loose contacts and hope for the best
     
    pwlm1975 likes this.
  13. OP
    OP
    pwlm1975

    pwlm1975 Member

    Joined:
    Apr 18, 2016
    Messages:
    579
    Location:
    COVID Central, Southside Brisbane
    So after a couple of hours I come back, and... it's stuffed again. This truly has me utterly bewildered. I'm going to look at it again tomorrow.
     
  14. baby_ute

    baby_ute Member

    Joined:
    Oct 25, 2008
    Messages:
    530
    Location:
    Sydney, Eels Territory
    Try resetting the cpu and check the pins, the top pcie slot run via the cpu and the x4 slot would run via the chipset
     
    nCrypt, Myne_h and pwlm1975 like this.
  15. OP
    OP
    pwlm1975

    pwlm1975 Member

    Joined:
    Apr 18, 2016
    Messages:
    579
    Location:
    COVID Central, Southside Brisbane
    Thanks baby_ute I'll give that a shot tomorrow and see what gives.
     
  16. baby_ute

    baby_ute Member

    Joined:
    Oct 25, 2008
    Messages:
    530
    Location:
    Sydney, Eels Territory
    you still getting problems?
     
  17. OP
    OP
    pwlm1975

    pwlm1975 Member

    Joined:
    Apr 18, 2016
    Messages:
    579
    Location:
    COVID Central, Southside Brisbane
    Heya, I haven't had a chance to reseat the CPU, probably over the weekend. I'll definitely come back and let you know.
     
  18. OP
    OP
    pwlm1975

    pwlm1975 Member

    Joined:
    Apr 18, 2016
    Messages:
    579
    Location:
    COVID Central, Southside Brisbane
    So I have pulled the cooler off and taken the following photos of the LGA and the CPU.

    https://imgur.com/a/X2YlkDb

    I couldn't see anything discoloured or that would indicate bent pins but others are no doubt more knowledgeable than I. Its the first time the CPU has been out of its socket since I put it together.
     
  19. de_overfiend

    de_overfiend Member

    Joined:
    Jul 12, 2001
    Messages:
    2,417
    Location:
    Gold Coast
    cant see any problems with the socket or the cpu. i doubt its cpu related

    reapply thermal paste and install the cooler

    out of curiosity what hard drive are you using? ssd or hdd. is the game installed on the ssd if you have a ssd?

    also the lego block trick is a good idea if the card is heavy. Top slot is faster as others have mentioned.

    also have you installed the correct chipset driver from the motherboard manufacturers website. I see from your previous posts that you did a fresh install

    also have you tried removing the overclock on the cpu and resetting the bios back to default?

    also have you tried extra cooling fans on the card... even a desk fan pointed into the open side of the case

    also try OCCT to test everything https://www.ocbase.com/
     
    pwlm1975 likes this.
  20. OP
    OP
    pwlm1975

    pwlm1975 Member

    Joined:
    Apr 18, 2016
    Messages:
    579
    Location:
    COVID Central, Southside Brisbane
    Done. I turned it all on, ran Valley with the card in the x16 slot, restarted, turn off and back on, then reinstalled the PCI WLAN card. Ran Valley each time and could not make it hiccup once. Results where I expect them and no juddery frames.

    SSD for OS, drivers etc. and 7200RPM Toshiba for games. Valley is running off the SSD (2.5" Crucial BX 240GB 65% free).

    This card is around 700g from what I can find. I have to get more Lego bricks now the card is a few slots higher.

    Yes first thing I do when I reinstall is do the full driver suite for the mobi, then I do a restore point. Definitely for the correct board.

    Yes, and yes.

    Yes although temps weren't an issue in the first instance, and the open case + desk fan made about 2 degrees difference overall according to HWMonitor.

    Doing that now. I appreciate all the help too :)
     

Share This Page

Advertisement: