Vulkan 1.4.336 is out today as the first spec update in two weeks for this high performance graphics and compute API.
With two weeks having passed compared to the common weekly-ish release, there are more issues resolved than usual. The issues/fixes are the usual assortment of small corrections and clarifications to the specification text.
Most notable are typically the new Vulkan API extensions introduced. With Vulkan 1.4.336 there is one new extension and it’s a NVIDIA vendor extension: VK_NV_compute_occupancy_priority.
The VK_NV_compute_occupancy_priority extension provides applications with …
Vulkan 1.4.336 is out today as the first spec update in two weeks for this high performance graphics and compute API.
With two weeks having passed compared to the common weekly-ish release, there are more issues resolved than usual. The issues/fixes are the usual assortment of small corrections and clarifications to the specification text.
Most notable are typically the new Vulkan API extensions introduced. With Vulkan 1.4.336 there is one new extension and it’s a NVIDIA vendor extension: VK_NV_compute_occupancy_priority.
The VK_NV_compute_occupancy_priority extension provides applications with control over how their compute workloads make use of GPU compute resources. VK_NV_compute_occupancy_priority can allow for prioritization relative to other ongoing GPU workloads by specifying their priority with values of low, normal, or high. The VK_NV_compute_occupancy_priority was worked on by a handful of NVIDIA engineers while we’ll see in time if this compute occupancy priority work ends up seeing interest or adoption from other Vulkan driver vendors.
All the details on the Vulkan 1.4.336 changes can be found via Vulkan-Docs on GitHub.