Hello everybody! I am having a strange issue with my graphics card. Yesterday i did service maintenance: clening from dust, removing oxide from copper pipes, repasting.
The graphics card was working fine before the service maintenance.
While remounting the heatsink, the card fell from my hands and run 2 inches before hitting the table i was working on.
Once reinstalled the heatsink, the card on pc, i turned on: everything worked as usual for the first six minutes; then screen flickering, and black screen. i was forced to restart the pc with the power button. Situation was the same.
Running OS is Windows 10 64 18362.
The device manager says: error 43, the device is not functioning properly so it was disabled. Afraid of causing further damage, i turned off the pc. After some time, i tried restarting, but now screen is totally black.
I removed the graphics card, removed the heatsink, and did a reflow procedure without flux at controlled temperature with max value of 250 celsius degrees.
After the cooling down, i reassembled the gpu, remounted on pc: now there is picture, no flickering, but screen resolution stays at 640 x 480. The device manager still says there is error 43.
Driver reinstallation fails.
Running GPU Z i found a strange thing: VRAM is 0 MB.
Here is data, copied by hand:
GPU: GP106 REVISION: A1
TECHNOLOGY: 16NM DIE SIZE: 200 MM^2
RELEASE DATE: JUL 19 2016 TRANSISTORS: 4400M
BIOS VERSION: 86.06.59.00.2A
SUBVENDOR: MSI DEVICE ID: 10DE 1C02 - 1462 3287
ROPS/TMUS 48/80 BUS INTERFACE: PCIEX16 3.0 @ X16 1.1
SHADERS: 1280 UNIFIED DIRECTX SUPPORT: 12
PIXEL FILLRATE: 84.4 GPixel/s TEXTURE FILLRATE: 140.7 GTEXEL/s
MEMORY TYPE: GDDR5 (SAMSUNG) BUS WIDTH: UNKNOWN
MEMORY SIZE: 0 MB BANDWIDTH: UNKNOWN
DRIVER VERSION: 10.0.18362.1 / WIN10 64
DRIVER DATE JUN 21 2006 DIGITAL SIGNATURE: UNKNOWN
GPU CLOCK: 1544 MHZ MEMORY: 2002 MHZ BOOST: 1759 MHZ
DEFAULT CLOCK: 1544 MHZ MEMORY: 2002 MHZ BOOST: 1759 MHZ
NVIDIA SLI: UNKNOWN RESIZABLE BAR: DISABLED
The system is using the default screen driver.
So it seems that the GPU chip is unable to connect to VRAM modules, so can it be a broken ball below the die that did not resolder?
Should i try another reflowing session?
Or maybe a reballing?
I am open to your experience.
I would recommend VMTCE (Video Memory Stress Test)
http://mikelab.kiev.ua/PROGRAMS/vmt.zip
It is also included in UBCD:
https://www.ultimatebootcd.com/index.html
0
Do not expect correct data in GPU-Z without driver. Check the corners of the GPU any are broken there is nothing to do.
0
It could be, but in this case isn't, because GPU die is perfect!
0