Initial PRIME Support Lands In RADV Radeon Vulkan Driver
Back in November we saw patches for wiring in PRIME support to the RADV Vulkan driver and last week rewritten RADV PRIME code was released while this weekend it has landed in Mesa Git.
RADV now has PRIME support on Mesa 17.1-dev Git. PRIME support is about better dealing with multiple GPUs where one GPU that does the rendering may need to transfer it to another GPU for display on the same Linux system in a Vulkan world, generally activated via DRI_PRIME=1. This RADV PRIME support isn't anything to do about better Vulkan multi-GPU rendering support or the Vulkan multi-GPU support improvements slated for the next major update of the specification (Vulkan-Next).
Just over 200 lines of code get RADV PRIME'd. The work was done by Red Hat's David Airlie, one of the co-founders of this open-source Radeon Vulkan driver, "This adds the ability to present to a different GPU that rendering using a driver side operation that can copy from the tiled to linear shared image. This does prime support completely in the swapchain present code, and each queue has a precreated command buffer for each image and for the each queue family. This means presenting should work on graphics and compute queues and transfer in the future."
RADV now has PRIME support on Mesa 17.1-dev Git. PRIME support is about better dealing with multiple GPUs where one GPU that does the rendering may need to transfer it to another GPU for display on the same Linux system in a Vulkan world, generally activated via DRI_PRIME=1. This RADV PRIME support isn't anything to do about better Vulkan multi-GPU rendering support or the Vulkan multi-GPU support improvements slated for the next major update of the specification (Vulkan-Next).
Just over 200 lines of code get RADV PRIME'd. The work was done by Red Hat's David Airlie, one of the co-founders of this open-source Radeon Vulkan driver, "This adds the ability to present to a different GPU that rendering using a driver side operation that can copy from the tiled to linear shared image. This does prime support completely in the swapchain present code, and each queue has a precreated command buffer for each image and for the each queue family. This means presenting should work on graphics and compute queues and transfer in the future."
6 Comments