Vega 64 - reset issue, 404 W power draw

Collapse
X
 
  • Time
  • Show
Clear All
new posts
  • Prott
    Senior Member
    • May 2022
    • 154
    • Czechia

    #1

    Vega 64 - reset issue, 404 W power draw

    Hey,
    I have a problem with Vega 64. Unfortunately there seems to be no schematic, so I am going by what I feel might be the case for those two issues:

    1. On restart the card doesn't seem to clear the VRAM and acts out (the picture stays on the screen followed by a black screen or artifacts till the drivers load). If I turn the computer off and on again the issue is solved and the card works flawlessly. PCI_RESET resistance is around 1.8 MOhm and the resistor between core and the reset pin is 0 ohm, as it should be. Is there anything else that could be causing this issue?

    2. Card reports stable power draw as 404 W, which is causing the GPU to throttle. Does anybody know the location of the resistors inbetween the IC that reports the powerdraw and shunts? I assume the main IC (that also acts as a vcore/vmem voltage controller) is doing this job, but I couldn't pinpoint the connection to any shunts from there.

    Thank you so much.
  • Aliencomputres
    Senior Member
    • Aug 2019
    • 181
    • România

    #2
    Re: Vega 64 - reset issue, 404 W power draw

    Heloo,

    black screen or artifacts its generated by the Vram issue.

    Comment

    • Prott
      Senior Member
      • May 2022
      • 154
      • Czechia

      #3
      Re: Vega 64 - reset issue, 404 W power draw

      Originally posted by Aliencomputres
      Heloo,

      black screen or artifacts its generated by the Vram issue.
      Thank you, I would think so too. But this happens only during restart and only when it restarts from graphic into "DOS mode". I think it's too consistent to be a VRAM issue.

      1. From dos mode to dos mode: During restart the picture is not cleared (if I physically restart the board the same picture remains frozen - even with blinking cursor till new picture overwrites it).
      2. If I restart from Windows it either displays artifacts till it returns to Windows or prevents the computer from booting.

      Except those two scenarios the card works 100 %. I would imagine if the VRAM was going out the card would display artifacts at random times during load.

      I will put it in my other computer and see if it's actually not a motherboard issue. The thing has been acting weirdly recently

      Comment

      • Prott
        Senior Member
        • May 2022
        • 154
        • Czechia

        #4
        Re: Vega 64 - reset issue, 404 W power draw

        UPDATE: The 404 W power draw got solved by itself... which doesn't make me that sure it's solved.

        However the PEX_RESET issue is still present. I changed the U100 and neighboring capacitor, but no change. As far as I can tell, except power rails (3.3 and 1.8) and ground, PEX_RESET only connects to the core. I will check the reset signal with a scope and update further for anybody else having this issue.
        Last edited by Prott; 09-29-2022, 01:32 PM.

        Comment

        • Prott
          Senior Member
          • May 2022
          • 154
          • Czechia

          #5
          Re: Vega 64 - reset issue, 404 W power draw

          Ok, finally got around to check the card with scope. The card's RESET signal is always high, no matter if I restart the computer or not. Normally it should go low upon restart for a moment, but for this card it doesn't.

          In Vega the reset signal goes only into an AND gate, so somehow it seems as if this gate is backfeeding and keeping the signal high even if motherboard pulls it low. I already tried to exchange it with a gate from a donor board, but it behaves the same.

          For some motherboards it means card works ok until restart, for most it doesn't even work at all.

          Anybody experienced this issue? Thanks

          Comment

          • Prott
            Senior Member
            • May 2022
            • 154
            • Czechia

            #6
            Re: Vega 64 - reset issue, 404 W power draw

            Ok, so this reset issue is killing me a bit.

            I replaced the AND gate (U100) with a known good one and it didn't fix anything. I meassured following:

            - resistance between PERST and PERST_BUF (signal after AND gate), it's around few kohm.
            - resistance between PERST and 3.3VBUS is also few kms

            What could prevent motherboard from pulling the reset low? This is, essentialy, the circuit (sorry about the mess).

            I have a schematic for the ROG flavour of the card, so the base variant might be a bit different - which is where the problem probably lies. I can't find the "SU101" anywhere on my card. I wasn't able to find continuity between PERST and any other IC on the board.

            Does anybody have any tips before I go completely mad?
            Attached Files

            Comment

            Related Topics

            Collapse

            • Document Archive
              AMD Radeon RX Vega 56 / RX Vega 64 Schematic & Board View
              by Document Archive
              AMD Radeon RX Vega 56 / RX Vega 64 Schematic & Board View

              The Radeon RX Vega 56 is a high-end graphics card by AMD, launched on August 14th, 2017. Built on the 14 nm process, and based on the Vega 10 graphics processor, in its Vega 10 XL variant, the card supports DirectX 12. This ensures that all modern games will run on Radeon RX Vega 56. The Vega 10 graphics processor is a large chip with a die area of 495 mm² and 12,500 million transistors. Unlike the fully unlocked Radeon RX Vega 64, which uses the same GPU but has all 4096 shaders enabled, AMD has disabled some shading...
              10-01-2024, 01:07 PM
            • x16
              Lenovo P50 - nvidia issue - how to reset bios?
              by x16
              My P50 has a faulty nvidia graphics card. It only works with intel. I installed Windows and it works fine withe the intel graphic.
              Then i switched in bios under config->graphics device from "hybrid graphics" to "discrete graphics". Now i have no picture.

              I reset the bios over shorting cmos battery, but still no picture. So i need to change the grapics setting back to hybrid graphics again.

              I need a way to make bios changes over bios dump. Is that possible?
              06-17-2025, 01:44 AM
            • Document Archive
              ASUS D515DA R7-3700U, 15.6" FHD vIPS, 512GB SSD, 8GB RAM, VEGA 10, W10H, 1YR Notebook Specification for Upgrade or Repair
              by Document Archive
              This specification for the ASUS D515DA R7-3700U, 15.6" FHD vIPS, 512GB SSD, 8GB RAM, VEGA 10, W10H, 1YR Notebook can be useful for upgrading or repairing a laptop that is not working. As a community we are working through our specifications to add valuable data like the D515DA R7-3700U, 15.6" FHD vIPS, 512GB SSD, 8GB RAM, VEGA 10, W10H, 1YR boardview and D515DA R7-3700U, 15.6" FHD vIPS, 512GB SSD, 8GB RAM, VEGA 10, W10H, 1YR schematic. Our users have donated over 1 million documents which are being added to the site. This page will be updated soon with additional information. Alternatively...
              09-06-2024, 10:50 PM
            • rounin
              MacBook Air A2337 (820-02016) USB-C Port Issue – Only Works in One Orientation (CC Line Problem?)
              by rounin
              Hi everyone,

              I'm working on a MacBook Air 2020 (A2337, board 820-02016) and running into a strange USB-C issue. I've gone through quite a bit of testing and part replacement, but I'm stuck — hoping someone here can shed some light.

              🛠 Original Issue
              The machine initially wouldn't negotiate 20V over USB-C — stuck at 5V.
              Diagnosis showed PP1V5_UPC0_LDO_CORE shorted to ground.

              I replaced U3100 (UF400) using a donor board (A2179), and the machine powers on fine now.

              ⚠️ New Problem: One USB-C Port Only Works in One Orientation...
              04-19-2025, 01:18 AM
            • ugamazing
              Three 820-01700 32GB Logic Boards Same Issue: Won't Wake From Hibernate
              by ugamazing
              I have three identical-spec 820-01700 boards (2.6/32/512), and ALL THREE boards came to us with the same fault, from different sources: The boards don't wake from hibernation with the keyboard or lid, they ONLY wake when you press the power button. I understand this is a very minor issue, but the boards must be fixed!

              We will of course begin with the obvious (checking lid signals), but I couldn't help but notice these three boards were all 32GB variants. Has anyone noticed an issue relating to these 32GB boards in particular? I of course don't think it's a RAM issue, but it's bizarre...
              12-17-2024, 07:05 PM
            • Loading...
            • No more items.
            Working...