StutterPC experiencing lag when under stress (Tuf 4090 OC / 5950x)
StutterPC experiencing lag when under stress (Tuf 4090 OC / 5950x)
Hey there, I’m really struggling here and need some guidance. I’ve tried several solutions, and while I managed to fix things by completely wiping and reinstalling my GPU drivers, the problem still lingers. I’ve shared my experience online, but only received a handful of replies so far.
What’s happening: My PC suddenly stutters while gaming, even in games that aren’t super demanding—like Helldivers and Destiny 2. I upgraded with an ASUS TUF Gaming RTX 4090 OC Edition and a 1600W PSU early this year, and it’s been mostly stable. The stutter started only recently, though I don’t know exactly when.
What I’ve noticed: Monitoring tools reveal that when stuttering happens, GPU usage drops by 10-25%, CPU drops 5-10%, and CPU frequency falls 25%. Everything snaps back quickly. There’s also a strange effect on my RGB lighting—when the game stutters, it visibly lags. During stutters, FPS drops by about 7-14 FPS, and the game becomes unresponsive for a few seconds to a couple of seconds depending on how intense the load is. The issue is unpredictable—sometimes I can play for hours without it, other times it stops completely in a short time. It’s not game-breaking, but it’s really frustrating.
What I’ve tried and what worked: Restarting the PC multiple times had no effect. I updated all drivers to the latest versions and cleaned old ones. After that, the stuttering got slightly less severe, but it still happened over time. I updated Windows 11 and even tried a beta version. Still nothing changed.
The TUF Gaming RTX 4090 is pre-overclocked, so I adjusted the overclock settings. Locking frame rates and enabling V-sync didn’t help. Changing graphics quality or updating RGB firmware/firmware didn’t resolve it. I flashed the BIOS, re-seated RAM, updated NVMe SSDs, and even changed my graphics settings. Nothing seemed to stick.
What hasn’t I tried yet: Reinstalling Windows completely. I’d prefer this as a last resort because it takes a long time.
PC specs: OS – Windows 11, BIOS – A.70 CPU (Ryzen 5950x with Corsair 420mm), RAM – Trident Z RGB 64GB DDR4 3600 (4x16GB), GPU – ASUS TUF Gaming RTX 4090 OC Edition, PSU – EVGA 1600 P+ Supernova Platinum M.2 NVME, SSDs – Samsung 970 EVO 1TB, 860 EVO 1TB, HDD – 4TB external, thermal checks passed. All fans and filters are clean. Components are new or recently acquired (2021), and I haven’t noticed any obvious bottlenecks.
I’m hoping this gives you a good starting point and really appreciate any advice from someone more experienced. I’d love to get this back to normal before it becomes a bigger issue.
Check your system with a different GPU? I noticed several 4090 units acting oddly lately, possibly due to driver or Windows problems. You might try GPU-Z to track GPU activity and look at the Perfcap Reason setting in Sensors. This can help identify why usage drops. Success is guaranteed!
I could give it a shot. My old GTX Titan X from 2015 is quite outdated, so I wasn't confident it would give the best performance. It kept running my PC for about ten years until I swapped it out with a 4090 in early February because it was really taxing in demanding games. I also just downloaded GPU-Z to check things out later. Thanks!
I conducted several tests in Warhammer 40k: Darktide, which demands a lot of resources and runs at full capacity (including RTX and DLSS). I entered the hub world center and stayed still for a few minutes, rotating the camera every 30 seconds. Even while stationary, the stuttering happened roughly every 12 to 15 seconds, making the game unresponsive for about half a second. The PerfCap reported "VRel" consistently throughout the session. I haven’t used GPU-Z before, so I’m unsure what it means, though a quick search suggests it relates to power consumption. GPU Load stayed at 60% or lower, never exceeding that. I saved the test logs if you want to review them. Maybe you can analyze more from this data. I haven’t tested with my older GPU yet, but I plan to try next. GPU-Z Sensor Log.txt
I’m unable to examine your log right now, but I can discuss the VRel value for the Perfcap Reason setting. VRel stands for Voltage Reliability, indicating that at the operating voltage, the GPU can run at a higher clock speed without becoming unstable. The reason you’re experiencing VRel before reaching the GPU’s power limit is likely related to insufficient voltage. Possible causes include a failing PSU, faulty cables, or a defective 12+4 connector. Take a moment to inspect your connections and cables. If the issue persists, switching to a different PSU might help. While a Supernova Platinum is durable, even top hardware can fail over time. Another option is using ASUS GPU tweak or MSI afterburner to slightly boost the voltage, but avoid setting it too high to prevent damage.
Consider removing all RGB applications and any software managing USB ports, including mice, keyboards, and AIO pumps. Your pump may switch to silent mode for testing purposes, though this won’t affect performance. Disconnect unused USB devices. Try using alternative mice and keyboards if available. In BIOS, change PCIe settings from Gen4/Auto to Gen3. Turn off the Global C-State option. Restart and enter BIOS settings. Switch RAID mode to AHCI, disable Safe Boot, and save changes. Ensure Samsung Magician is installed with the latest firmware. If your motherboard uses RAID for NVMe storage, reset certain configurations before installing the driver. Verify the presence of a standard NVMe controller in Device Manager to confirm proper setup. Keep your Bitlocker disabled during this process or turn it off beforehand and back on afterward to prevent issues. You can also use command prompt commands for similar adjustments, though the GUI method is straightforward.
Based on your advice, I adjusted the power settings in Nvidia Control Panel, switching the power draw mode from "Normal" to "Prefer Maximum Performance." This change significantly improved performance for a day. However, the stuttering returned. I disconnected nearly<|pad|>’s power cables and reattached them, but the issue persisted. I’m really puzzled by this situation.