Vega 64 not running at correct clocks and other problems.

Discussion in 'Videocards - AMD Radeon' started by MaxThroughput, Feb 23, 2018.

  1. MaxThroughput

    MaxThroughput Guest

    Messages:
    2
    Likes Received:
    0
    GPU:
    XFX Radeon Vega 64
    Hey All,

    I am having some weird issues with my XFX Vega 64. Firstly, it has an EKWB water block on it, so temperatures haven't been an issue. I was using the original bios and was about to find a max gpu speed of 1622mhz as listed in the global wattman. This equates to an actual clock speed of 1580, occasionally it would hit 1600. When running it full bore for a while, I noticed that graphically in wattman, the gpu had been varying speed erraticly, and it would occasionally crash. Flash forward to today, I installed the XFX watercooled bios and have a 1.25v gpu now. This required me to reinstall windows because the start button stopped working and would not return to a working state after numerous changes. This isnt the first time I've flashed to watercool bios and had that happen. Upping the voltage seems to have made the gpu clock more stable.

    Anyway, the issue I'm having with the watercool bios is that its currently set to 1622mhz gpu, with max power limit, and the highest clock speed I can seem to run according to hiwinfo64, is 1560mhz while using 365watt. It won't go above this value, and if i raise the wattman speed above 1630mhz, my machine will crash.
    What's going on here? I have an RMA open because the card wouldn't seem to run reliably at it's advertised speed of 1560mhz, without needing a watercool bios. Should I just return it? I just didn't want to have to deal with the hassle of going without a graphics card for weeks, but I feel like I have received an inferior product. At these speeds I would have been much better off going with an air cooled 1080ti. The only thing I have going for me is that I bought the card under MSRP in December before everything went crazy in the mining world.

    Thanks,
    Max
     
  2. MaxThroughput

    MaxThroughput Guest

    Messages:
    2
    Likes Received:
    0
    GPU:
    XFX Radeon Vega 64
    Update: I ran a quick battlefield 4 match and noticed the GPU remaining at 1590 for the duration. Again it is set to 1622 in wattman, so i think that's within reason. For some reason Skyrim SE, and startcraft II with max driver settings will only draw 1500mhz from the gpu. Same goes furmark and unigen heaven.

    Ran a second match and could only achieve 1540 max gpu. Gpu core was at 61C, I have my fans set to basically maintain quiet and temperature at 61-64C. After running the fans and radiatiors full speed the temp is at 50C and I can maintain 1610mhz.

    There is very obviously some internal throttling occuring when this card hits 60C. The target temp is set to 70C/70C but that doesn't appear to matter.
     
    Last edited: Feb 23, 2018
  3. OnnA

    OnnA Ancient Guru

    Messages:
    17,851
    Likes Received:
    6,741
    GPU:
    TiTan RTX Ampere UV
    It's Max allowed ( those 1600+)
    If can do it will ;)
     
  4. GREGIX

    GREGIX Master Guru

    Messages:
    855
    Likes Received:
    222
    GPU:
    Inno3d 4090 X3
    1st. What is ur hot spot tmpure from gpu-z?
    2nd. If something help those cards from my and others exp this is UnderVolting not increading gpuVoltage.
    3. This could be HBM temp issue too. Is it 60 too? I know 70 is not issue with my TUL card but...is air.
    Anyway...had best results with undervolt hbm and gpu...powerdraw was like 220-270w while gaming and was buttery smooth although gpu core speed varied greatly.
     

  5. Glasya

    Glasya Guest

    Messages:
    2
    Likes Received:
    0
    GPU:
    ASUS RX Vega 64 8GB
    Hello.
    Does anybody know what is GPU VR VDDC temp in HW Info?
    It goes to 110C° (once it went to 118C°) while gaming while GPU and HBM temp stays under 70°.
    GPU VR MVDD and GPU Hotspot temps are around 95°
    I have ASUS Strix Vega 64, Power Target is 30%, GPU Voltage is 1180 (20 below ASUS default) and HBM is @ default.
    Fan goes to 2921RPM because temp target is set to 65° It is only way to keep this VDDC temp under 110C°
    Max W=323.

    Thank you.
     
  6. GREGIX

    GREGIX Master Guru

    Messages:
    855
    Likes Received:
    222
    GPU:
    Inno3d 4090 X3
    Voltage regulator. The square thingies around gpu. Probably u did not secured their cooling...so this is real issue and cause of ur problem. One of the complex(gpu, vrm,hotspot,hbm) reaches dangerous tem-card start throttling.
    Buy decent thermal pads or replace/correct position that u r using now(if any...)
    That should fix it.
    Damn. Srry OP, was thinking is ur vrm problem;)

    Anyway...Glasya...if u r on stock cooling then rma card or change cooler. Or try to fix by urself but u will loose warranty;)
     
    Last edited: Feb 23, 2018
  7. Glasya

    Glasya Guest

    Messages:
    2
    Likes Received:
    0
    GPU:
    ASUS RX Vega 64 8GB
    Thank you
    I will test it for sometime and see how it goes. Mayby i will return it or play with powertarget a bit more.
    I don`t want to take off the cooler because of warranty.
    This is first ASUS card with that kind of temp. Last ones were cool and quiet though.

    I let you know.

    EDIT:
    I set the GPU Voltage from 1200mv to 1100mv, Memory voltage from 1100mv to 1080mv and raised the GPU Clock from 1630mhz to 1650mhz.
    Now GPU Clock is almost steady (1590mhz - 1605mhz, before it didnt reach 1500mhz very often) and maximum power usage is 280W.
    GPU fan max speed is now 2430 rpm (down from 2921rpm) and GPU VR VDDC max temp is 101C°. Maximum Hotspot temp is 93C°
    HBM Memory is @ default speed (950mhz).
     
    Last edited: Feb 26, 2018
  8. wdpir32k3

    wdpir32k3 Guest

    Messages:
    8
    Likes Received:
    0
    GPU:
    Rx Vega 64 LC
    I have a Vega 64 LC and my wattman settings are clocks 1742/1752 for my voltage 1150/1150 HBM 1110 mV 1100 power limit at 20% I'm able to stay in the 1600 to 1700 also my watt usage is under 300 i hope these settings help you if you need me to run more tests to help let me know
     
  9. GREGIX

    GREGIX Master Guru

    Messages:
    855
    Likes Received:
    222
    GPU:
    Inno3d 4090 X3
    As I played with that today on my sons PC, I did almost all as u, but mem was 1040 at 1049mV, core speed not touchd, but increased power limit to 50%. So state 7 and 6 was at 1100mV, and as I remember power usage while rainbow six siege was like 220W max. BUT it has custom cooler, morpheus 2 with 2x 120mm fans on it so...
    Temp was (max):
    GPU - 52
    HotSpot - 70
    HBM - 54
    VRM- 67

    On balanced temps were higher by some margin and powerdraw hit 280-330W sometimes. Funny thing he still sees 250-290FPS in game in his settings(preferably lows, he needs more FPS=less latency=better reaction time), same score as in balanced...
    And there is less warmer in his room now :)
     
  10. stevin48

    stevin48 Guest

    Messages:
    11
    Likes Received:
    0
    GPU:
    asus ares 5870x2
    my sapphire nitro+ vega 64 bought new may 2019 ran like a champ above 1600 mhz core but started noticing lower clock speeds each driver updategot to update 19.12.1 clocks was only 1200 to 1300 mhz and losing 15 fps on valley bench updated to 19.12.2 the 2020 new driver and clocks again slower 1000 to 1200 mhz but mem stated bouncing from 800 to 945 which has never done temps fine edge temp only reaching 64 c . was afraid card was going bad . but uninstalled driver started going back to older drivers each time clocks would improve a little now back to driver 18.5.2 and clocks are back to 1580 to 1603 not the 1630 it was running but i know its drivers now not the card . i just bought this card 8 months ago new WHY is AMD doing this . if anyone having this problem with VEGA cards try this .. it stinks but it worked on my card .
     

  11. OnnA

    OnnA Ancient Guru

    Messages:
    17,851
    Likes Received:
    6,741
    GPU:
    TiTan RTX Ampere UV
    Guys sometimes HWinfo64 reports incorect values for Vega GPU clock.
    I don't see any downclocks on my GPU, if i set 1650MHz (Forza) it stays in those bracket (P6-P7)
    In menus it downclock even more 1200-1300MHz.

    The 19.12.2 Driver 2020 it is similar in performance as 19.11.x etc. but it's most stable driver to date.
    Not a single restart was needed from the begining.

    Please try 2020 ;)
     
    Last edited: Dec 17, 2019
    The_Amazing_X likes this.
  12. stevin48

    stevin48 Guest

    Messages:
    11
    Likes Received:
    0
    GPU:
    asus ares 5870x2
    sapphire nitro+ vega 64 does not like newer driver or mine dont the 19 .12.2 my core clocks dropped as have been each newer update the 19.12.2 my clocks were down avg 1100 mhz to 1300 mhz and mem dropping while temp only 63 c . and could not get to undervolt . uninstalled went back few drivers and got it up to 1500 mhz turbo . went back to amd driver 18.5.2 and on turbo my clocks tun above 1600 mhz again ITS THE DRIVERS amd wants us to upgrade our cards. sapphire is core clock of 1630 mhz . the 18.5.2 driver you wont see mem bouncing up and down or the gpu core clock lowered . its rock solid . watch memory clock while bench using msi afterburner on the newer driver it dives as low as 500 mhz causing frame drops and studder on newer drivers
     

Share This Page