NVIDIA Puts Out 290 Linux Driver Series Beta
It was just in August that NVIDIA was pushing out driver betas for their Linux/Solaris/FreeBSD 285.xx series, but now that the series is stable, they have moved onto the 290.xx series. On Friday NVIDIA released the 290.03 Linux driver beta.
Changes for the NVIDIA 290.03 Beta is support for the GeForce 510 GPU, fixes a bug for loading the driver on some integrated graphics systems, VDPAU fixes, support for limiting heap allocations in the OpenGL driver via the __GL_HEAP_ALLOC_LIMIT environmental variable, an "Accel" option for the X.Org driver to disable GPU hardware acceleration, modified how the OpenGL driver allocates executable memory so it can function properly if /tmp is mounted noexec, GLSL fixes, improved performance by caching compiled OpenGL shaders to the disk, and a bug fix for slow rendering on older GPUs with X.Org Server 1.11.
The new "Accel" option for turning off hardware acceleration is if you wish to use your NVIDIA GPU solely for OpenCL/CUDA compute processing and wish to provide such tasks exclusive access to the GPU. The "GLShaderDiskCache" option for increasing performance by caching compiled OpenGL shaders to the disk is also interesting and will be looked into and potentially benchmarked by Phoronix.
Read more about the driver or download the x86/x86_64 Linux 290.03 Beta release from this forum thread.
Changes for the NVIDIA 290.03 Beta is support for the GeForce 510 GPU, fixes a bug for loading the driver on some integrated graphics systems, VDPAU fixes, support for limiting heap allocations in the OpenGL driver via the __GL_HEAP_ALLOC_LIMIT environmental variable, an "Accel" option for the X.Org driver to disable GPU hardware acceleration, modified how the OpenGL driver allocates executable memory so it can function properly if /tmp is mounted noexec, GLSL fixes, improved performance by caching compiled OpenGL shaders to the disk, and a bug fix for slow rendering on older GPUs with X.Org Server 1.11.
The new "Accel" option for turning off hardware acceleration is if you wish to use your NVIDIA GPU solely for OpenCL/CUDA compute processing and wish to provide such tasks exclusive access to the GPU. The "GLShaderDiskCache" option for increasing performance by caching compiled OpenGL shaders to the disk is also interesting and will be looked into and potentially benchmarked by Phoronix.
Read more about the driver or download the x86/x86_64 Linux 290.03 Beta release from this forum thread.
4 Comments