MSI GTX 770 4Gb

February 18, 2022
 |  No Comments

Seller Note “After installing drivers screen goes white, sometimes blue and then black. Displays fine once plugged in before installing drivers.”

Summary

  • The unusual 4Gb version, very nice condition, has a backplate unlike my two other MSI GTX 770 2Gb cards (MSI GTX 770 2GB (2048 MB) Twin Frozr (A) MSI GTX 770 2GB (2048 MB) Twin Frozr (B))
  • Resistances:
    • VCore – 11.8Ω ohms
    • Vmem – 77.5Ω vmem (Hynix H5GQ2H24AFR-R2C)
    • PEX – 625Ω
    • 5V – 382.4Ω
    • 3.3V – 752Ω
    • 12Ω – 6KΩ+ (Slot), 50KΩ+ (8 pins)
  • Actually very capable under load, Kombuster, Heaven and Subnautica all seem to run fine at about 60 fps
  • The GPU-Z graph is very unusual: The GPU is flat out at near 100%, but power consumption is around 70%! Vrel limiting all the time, and a strange choppy VRM graph with VDCC phasing in and out…
  • Once it crashed to a grey screen once after exiting heaven, it doesn’t always crash though (only once for me in two sessions)
  • After benchmarking / load, a slight clicking about a second between clicks, might be heat-related or a VRM fault?
  • MATS was fine
MSI GTX 770 4Gb

Need to inspect the VRM area.

Thoughts:

  • Maybe a faulty VRM capacitor would result in a choppy/unstable VRM output voltage?
  • 67% TDP @ 100% GPU load might mean 1 or more power stages are compromised?

Update – 30/06/2022

I have tried observing each phase using an oscilloscope. The card seems to sustain load and crash after the load is removed. I checked each phase looked the same at rest (the oscilloscope shape had 3 identical waves for each phase). Then under load, the oscilloscope showed identical patterns with 5 waves in each phase. Then when I exited the game, the pc crashed and only 1 phase had any waves (looked like the standby pattern). Maybe this is a minimal pattern for the VRM e.g. only a single main phase running or perhaps this is the PWM misbehaving? A clicking noise could be heard during the load.

Actually, having discussed it on LER, it seems plausible that given:

  • VRM is stable under load and each phase shows the same waveform -> Switching components are probably OK
  • When the load drops, the crash occurs -> Possibly the capacitors are worth checking and the coils

Capacitors are E356 820 2.5v,

Coils are 0.22UH/35A/S