Instability in GPU power supply?
Instability in GPU power supply?
GPU voltage fluctuates down by nearly a whole 0.3V whilst playing games. Does not happen as often in other games, moreso in a specific game, but it is apparent. These fluctuations cause frame stuttering and are only mitigated by restarting the computer. It is not being undervolted, I have no hand in the GPU behaviour, it simply happens on its own. I have tried upping the power limit to give it more breathing room and it still occurs. My temps are entirely fine. Edge temps barely reaching 56 degrees whilst the hot spot reaches no more than 79 (yes this is a high delta, this card has been known to get deltas as much as 60... but I fixed the issue (for now) by just lowering the graphics of the games I play) The little dips (ignore the massive spikes down- those are instances where the game window is not open) are the instances where the voltage will drop by 0.3V consistently for whatever reason. The point of where the voltage stabilizes is when I was entered into a match of War Thunder. Peaks of voltage hit 1.20V. Lows hit 0.93V. Specs: 6750XT GIGABYTE 7600X Rm 750X 32GB, Cl36, 6000Mhz (expo on) G.Skill, trident SN770 2TB SSD + S70 Blade (boot drive) GIGABYTE B650 AORUS ELITE AX
GPU voltage and performance vary with workload. Several factors could cause this problem. Check if your GPU and CPU usage are normal during low activity. Look at the utilization numbers for both components. Consider adjusting the GPU voltage through afterburner to maintain stable speed and power. This should help keep consistent performance. Next, identify which games are causing issues and when they occur. Note the resolution you’re using and how CPU usage changes under those conditions. Also, observe RAM consumption and background processes during these events. Have you already tried DDU, reinstalled drivers, and updated all software?
The GPU usage is being impacted by certain processes. As illustrated in the chart and additional notes. The CPU stays fairly stable. The most affected game is War Thunder. I’ve observed this in Rainbow 6 Siege as well. Recently, I played a lot of Ultrakill and never saw frame drops. All my games run at 1440p with maximum graphics settings. While watching, in War Thunder I used around 6.3GB to about 7.2GB of VRAM. Only a few background apps like HWinfo, GPUz, and Afterburner are active. I’m considering using DDU again, but it hasn’t resolved the problem yet. This issue appears only during explosions in War Thunder, though I don’t understand why the card struggles with animations and graphics. It should easily handle high-octane scenes like those seen in the game. The voltage seems to change unpredictably, disrupting GPU performance while the clock remains stable.
Voltage could be restricted below spike levels with adrenaline or afterburner. Additionally, certain games suffer from poor optimization or rely heavily on complex, slow-to-render particle effects.
This situation might be unusual, as it doesn’t happen in other games. It seems War Thunder usually leans toward Nividia cards, but I’m not sure. After further testing, the main issues appear to be voltage swings—ranging from 0.83 to 1.006 only—and inconsistent GPU usage. The MSI afterburner isn’t stabilizing the voltage even though I enabled it. This also matches periods of heavy GPU activity. I hope I can adjust those settings myself. No other factors, like CPU changes, are involved. Everything else stays consistent except for voltage and GPU fluctuations, which usually cause at least a 0.17 drop. I reset the AMD profile and turned off eco mode on my CPU, which helped a little. Could memory instability be contributing to this? I’ve suspected my RAM might be unreliable, leading to crashes.
I question this since GPUs typically use their own RAM (VRAM). I also noticed Afterburner can behave oddly with some AMD cards; I prefer Adrenalin because it’s straightforward. Voltage will differ, but shifting between two values seems unusual—maybe there’s a specific reason. Have you experimented with War Thunder settings to check if changing them affects the problem or voltage?
The main variation appears in MSI AB or AMD software settings. After further adjustments, the difference between the two fluctuation points increased. I believe I can improve the narrowing process. Stability in FPS has improved slightly, likely from GPU clock and usage rather than just voltage. It seems a combination of both factors. I’m finished tweaking now but will return tomorrow with an update if available. Edited April 3, 2024 by YusExa Spelling
Making a move... I assumed it was a blend of shifting and letting go... that’d make for a really funny GPU.