RPCS3 Workaround Boosts Performance on NVIDIA GPUs

The RPCS3 team says contributor Yahfz found a workaround for an NVIDIA driver bug that was bottlenecking graphics performance in the PlayStation 3 emulator. The change is notable because the team is reporting meaningful gains across both older GeForce GTX hardware and newer RTX GPUs rather than a narrow improvement on one generation.

RPCS3 says AMD GPU drivers do not exhibit this particular bug, so Radeon users should not expect the same uplift from the workaround. The project also cautions that the size of the gain depends on the game, hardware and exact scene being tested. These are RPCS3 project measurements, not independent cross-platform benchmarks or a guaranteed percentage for the emulator as a whole.

The largest example published so far is an open-road Gran Turismo 5 test on a Ryzen 7 9850X3D and RTX 3070, where RPCS3 reports an increase from 107 FPS to 147 FPS. In that same scenario, reported VRAM use fell from 7.1GB to 4.8GB. A separate Gran Turismo 5 test on a Core i9-12900K and RTX 5070 moved from 60 FPS to 67 FPS, illustrating how strongly the result can vary.

That variation matters more than the headline maximum. Emulation can shift bottlenecks between CPU emulation, graphics translation, synchronization and the host driver depending on the title and scene. Recovering time lost specifically inside an NVIDIA driver path can therefore produce a large gain where that path was limiting performance and a much smaller gain elsewhere.

The workaround is not confined to Blackwell or recent RTX cards. RPCS3 tested a laptop with a Core i7-7700HQ and GTX 1060 and reported roughly 20% higher performance in Saints Row IV and about 10% in Saints Row: The Third. That makes the change particularly interesting for emulator users on older PCs, where recovering driver-side overhead can matter more than on a system with abundant GPU headroom.

RPCS3 already maintains a public graphics-driver issues page documenting vendor-specific OpenGL and Vulkan behavior. That history is useful context: mature emulators routinely encounter driver behavior that needs application-side handling even when the emulator itself is not the original source of the problem.

Alongside the performance results, the RPCS3 team said it does not have a concrete route for submitting independent-software-vendor driver bugs to NVIDIA in a way that reliably gets them fixed. The project says previous attempts through NVIDIA developer forums did not produce results and has asked NVIDIA to establish direct communication for future reports.

For users, the immediate story is simpler: the emulator now avoids a costly NVIDIA-driver behavior itself. A later driver-side fix could still be preferable because it would address the underlying behavior rather than requiring software to route around it, but RPCS3 has not announced such an NVIDIA fix. The defensible claim today is therefore about the emulator workaround and the project’s measured examples, not a new GeForce driver release.