Scores declining steadily in 3Dmark performance
Scores declining steadily in 3Dmark performance
Hardware Details: Central processing unit AMD Ryzen 7 9800X3D, Graphics card MSI RTX 5090, System built on ASUS ROG Strix B650E-F motherboard, RAM capacity 32GB DDR5 with EXPO enabled, Storage solution includes a 2TB NVMe Gen4 SSD, Power supply rated at 1200W, cooling system features a high-end AIO cooler, Operating system is Windows 11 with all updates applied.
Issue Overview: Problems emerged following a PBO undervolt adjustment (all-core -10, +150 MHz boost) together with a UV 2800MHz clock at 0.91mV and VRAM overclocking to +2000 MHz using MSI Afterburner. A crash occurred in RDR2 displaying a black screen with flickering warnings, no WHEA errors. Performance degradation became evident in 3DMark tests, particularly in Time Spy and Port Royal. While Steel Nomad and CPU-focused benchmarks remain stable, overall results have declined. Some DX12 titles showed minimal GPU usage despite normal frame rates elsewhere (RDR2, KCD2, Cyberpunk 2077), except after restoring UV settings.
Actions Taken: Reinitialized CMOS several times, reset BIOS to default settings (EXPO off, PBO disabled), updated BIOS to latest version, performed a clean GPU driver install, reinstalled MSI Afterburner and all profiles prior to DDU, disabled G-Sync, turned off background apps including RivaTuner before uninstalling, checked NVIDIA control panel preferences, ensured optimal power plan in Windows 11, verified GPU/CPU load via HWInfo64, ran system file checks and DISM scans, reset VRAM settings, confirmed fan curves and thermal connections, disabled iGPU in BIOS.
Observed Symptoms: Performance dropped up to 2000 points in 3DMark Port Royal, Steel Nomad, and Time Spy; GPU utilization remained low in certain DX12 games (e.g., Hogwarts Legacy) despite normal FPS elsewhere; CPU stability issues under load; no thermal throttling observed.
Potential Causes: Possible BIOS corruption that cannot be recovered (latest version irreversible), possible VRAM instability from the overclock, hidden driver conflict, or an unusual interaction between DX12 and Windows post-crash.
Additional Notes: Game compatibility has improved after reverting UV settings, but some titles still exhibit reduced performance. Monitoring showed stable temperatures (CPU 55-57°C, GPU 57-61°C), no shutdown events. Stress tests passed, though occasional yellow Q-LED appeared during testing. Overall system remains functional, but further investigation into firmware integrity and driver configurations is recommended.
Welcome to the forums! Are you sure about this? It seems like it's not working well and keeps getting worse each time. I’d probably just restart everything and start fresh. Think about reinstalling Windows, updating the GPU driver, going back through the BIOS settings, resetting the CMOS, and then try adding OCs and UVs once you’re clean.
During assembly, I purchased approximately 14800-14900 in steel nomad and around 36500-37200 in time spy. After applying UV for both GPU and CPU, values rose to about 15100-15200 in steel nomad. The problem then emerged as previously noted. Currently, I receive 14200 in SN, and repeated benchmarking yields progressively lower scores (including a 12800). I’ve thought about a clean Windows 11 reinstall but wanted to explore other options first. If proceeding with that, should I run a DDU and CMOS check before the uninstall? I’m unsure if I can revert the latest BIOS changes.
Does this drop in performance happen only during consecutive sessions? It might be worth examining the thermal setup first—perhaps an inconsistent batch of paste, improper application, or other factors could be causing heat buildup that matches the throttling you notice. Also consider if a fan was disconnected or connections were loose.
I’m thinking about restarting the system by going back in BIOS and making sure fans are at maximum speed and PWM enabled. After that, I’ll reboot and reapply all my settings through FanControl. Someone mentioned that CMOS resets fan controls in BIOS, which could conflict with the software settings. I’ll verify the connections for the fan, but the hardware info doesn’t indicate thermal throttling from the CPU or GPU.
Also considering reapplying all my UV profiles to check for improvement. It seems a bit redundant since the stock doesn’t appear to perform well. My idea is that some functionality might still be active quietly in both the BIOS and GPU even after a clean CMOS and DDU setup. Updating the settings could potentially reset those hidden or silent configurations.
Apologies for the confusion. It seems they’re releasing an exploit patch and aren’t allowing rollbacks, which is frustrating. It appears to be reaching thermal limits. Are all temperatures normal? Based on your specs, it looks like you have sufficient funds to replace the SSD—consider swapping it out and testing if that resolves the issue. CMOS will reset everything, while DDU is software-based and won’t survive a fresh install.
If you're missing some of the thermal data OddOod mentioned, I'd check monitoring tools like HWinfo to see if performance throttling is occurring and how severe it is. I'm not an expert on where to look, but things like the VRM on your motherboard, the boot SSD, or other heat-sensitive parts could be involved. It might help to identify what's causing the slowdown before focusing on cooling solutions. This is just some ideas for consideration.