Released as part of their slightly less tested New Feature Branch, NVIDIA driver 615.71.09 is out now for Linux.
For those who want things nice and stable, you should stick to the 595 series with the most recent there being driver version 595.99.02. As for this new release it includes some Vulkan / Proton improvements, along with various bug fixes. How to get it will depend on what Linux distribution you use - they all have their own different update systems.
All the changes are listed below.
NVIDIA 615.71.09 Key Changes
- Added support for VK_NV_low_latency revision 2, to enable out-of-the-box Proton support for NVIDIA Reflex in Vulkan-native games which use NvLowLatencyVk.dll.
- Fixed a bug which prevented VK_NV_low_latency2 from reducing latency on swapchains created from VK_KHR_wayland_surface and VK_KHR_display surfaces.
- Fixed multiple startup and alt-tab hangs that occurred with NVIDIA Smooth Motion enabled.
- Fixed a bug that caused a file descriptor leak when the Vulkan loader unloads libGLX_nvidia.so.0 after a call to vkDestroyInstance.
- Added support for the VK_EXT_cluster_acceleration_structure Vulkan extension.
- Fixed an Xid 109 error when loading PRAGMATA on Wayland.
- Added support for cgroups-based memory partitioning.
- Added a new token, RmDisableDisplayGlitchPerfLimit, to the NVreg_RegistryDwords nvidia.ko kernel module parameter. This may reduce idle power consumption in some multi-monitor configurations, at the risk of changing memory clocks while display is using the memory, potentially resulting in momentary display glitches.
Example usage:
modprobe nvidia "NVreg_RegistryDwords=RmDisableDisplayGlitchPerfLimit=1". - Fixed a bug that could cause corruption when display output scaling is used on Blackwell GPUs.
- Fixed a bug that caused resume from suspend to fail if the system was previously hibernated:
https://github.com/NVIDIA/open-gpu-kernel-modules/pull/1199. - Fixed a rare correctness issue on Blackwell GPUs affecting kernels with two or more nested levels of thread divergence, where reconvergence could fail. Kernels with only a single level of divergence are unaffected, as are applications already built with NVCC 13.2.2 or newer. This fix guarantees correct execution, though it may come with a modest performance cost. Recompiling with NVCC 13.2.2 or newer will resolve the possible slowdown. If you are unable to recompile your application for any reason, the runtime behavior of this fix is controllable on a per-process-launch basis through the environment variable CUDA_SELECTIVE_DEVICE_CODE_RECOMPILE as follows:
- CUDA_SELECTIVE_DEVICE_CODE_RECOMPILE=DEFAULT
- Default behavior: attempt JIT recompilation for 15 seconds, runtime patch remaining code.
- CUDA_SELECTIVE_DEVICE_CODE_RECOMPILE=FORCED
- Re-finalize every eligible module, ignoring the time budget.
- CUDA_SELECTIVE_DEVICE_CODE_RECOMPILE=DISABLED
- Driver patching only, never re-finalize.
- CUDA_SELECTIVE_DEVICE_CODE_RECOMPILE=DEFAULT
🌐 External Sources: nvidia.com
Article taken from GamingOnLinux.com.
About the author - Liam Squires-Hand
I am the owner of GamingOnLinux. After discovering Linux back in the days of Mandrake in 2003, I constantly checked on the progress of Linux until Ubuntu appeared on the scene and it helped me to really love it. You can reach me easily by emailing GamingOnLinux directly.
See more from me
See more from me
Some you may have missed, popular articles from the last month:
You can also find comments for this article on social media: All posts need to follow our rules. Please hit the Report Flag icon on any post that breaks the rules or contains illegal / harmful content. Readers can also email us for any issues or concerns.
Patreon
PayPal