Version 0.5.3

• Add entry points needed for NVIDIA Reflex by forwarding the relevant calls into LatencyFlex. LatencyFlex is now an optional runtime dependency, thus Reflex is available when LatencyFlex is properly set up. The recommended version is LatencyFlex 0.1.0. See also this note about Reflex and Usage about setup and titles known to work.
• Fix a crash when querying NVIDIA PhysX capabilities in certain conditions. This affected e.g. the NVIDIA Supersonic Sled Demo.
• Stub some methods for NVIDIA driver settings. This should make no difference in practice, but slightly improves logging statements.

Version 0.5.2

• Update NVAPI headers to R510.
• Add entry points needed for querying NVIDIA PhysX capability. PhysX requires nvcuda.dll from Wine-Staging and a recent PhysX runtime. See also this note about PhysX
• Add several methods for GPU system information. Those additional methods help to correctly correlate CUDA devices to NVAPI adapter.
• Return fixed “DXVK-NVAPI” string when querying NVAPI interface version. This helps to identify DXVK-NVAPI.

Version 0.5.1

• Only report NVIDIA GPUs running by NVIDIA's proprietary driver. The old behavior where all GPUs are reported, is now guarded behind the DXVK_NVAPI_ALLOW_OTHER_DRIVERS environment variable.
• Report SM (Streaming Multiprocessor) version based on GPU architecture when querying D3D12 capabilities.
• Report external device ID when querying PCI device identifiers.

Version 0.5

• Add entry points needed for NVIDIA DLSS 2.x for D3D12 by forwarding the relevant calls into VKD3D-Proton, this requires VKD3D-Proton 2.5. See also this note about DLSS 1.x.
• Add entry points needed for NVIDIA DLSS 2.x for D3D11 by forwarding the relevant calls into DXVK, this requires DXVK 1.9.3. See also this note about DLSS 1.x.
• Add support for SetDepthBoundsTest for D3D12. This feature is directly supported by newer versions of D3D12, but e.g. certain versions of Frostbite Engine still use NVAPI for this feature (#45).
• Logging level defaults to none, logging must be explicitly enabled now by setting DXVK_NVAPI_LOG_LEVEL to info.
• Fetch subsystem ID from NVML when NVML is available. This might help identifying the current GPU (#50).
• Use NVML to query current performance state when NVML is available.

Building DXVK-NVAPI requires MinGW 9 and Python 3 from this version on.

Version 0.4

• Update NVAPI headers to R470.
• Use NVML (NVIDIA Management Library) when available to query temperature, utilization, clock info and VBIOS version for NVIDIA GPUs. This requires wine-nvml (https://github.com/Saancreed/wine-nvml) on Linux. See this project how to add NVML support to Wine/Proton. Several dedicated and in-game benchmarks visualize those GPU info.
• Introduce DXVK_NVAPI_DRIVER_VERSION environment variable to override the reported driver version. This can be useful to prevent annoying alerts due to falsely outdated driver version.
• Introduce DXVK_NVAPI_LOG_LEVEL=none environment setting to prevent any log output.
• Fix SetDepthBoundsTest/BeginUAVOverlap/EndUAVOverlap usage with a D3D11 context instead of a D3D11 device.
• Add test suite (built optionally). Run nvapi64-tests.exe without arguments to validate Wine setup and query the local system. Run the internal unit-tests with nvapi64-tests.exe [@unit-tests], useful for contributors only. The tests can also be run on Windows against NVIDIA's nvapi64.dll to compare results.

Version 0.3

• Update NVAPI headers to R465.
• Add several methods for GPU system information.

Version 0.2.2

• Add file logging (requires DXVK_NVAPI_LOG_PATH to be set to a location).

Version 0.2.1

• Improve internal implementation.

Version 0.2

• Forward BeginUAVOverlap/EndUAVOverlap to DXVK.
• Forward MultiDrawInstancedIndirect/MultiDrawIndexedInstancedIndirect to DXVK.
• Add methods for basic system information.

Version 0.1.2

• Produce binaries using a more modern built environment.