Possible Items On The OpenCL Roadmap: Improved Vulkan Interop, Arbitrary Precision
There was an OpenCL BoF during the Khronos Group day at SIGGRAPH 2016. At this session yesterday they did comment on OpenCL roadmap discussions taking place.
Among the items mentioned during the OpenCL BoF @ SIGGRAPH 2016 were possible features on the roadmap of:
- Desktop: Vulkan interoperability, arbitrary precision for increased performance, pre-emption, collective programming and improved execution model.
- Vulkan Compute leveraging OpenCL: types, precision and accuracy, pointers and address spaces, execution model.
- Mobile: arbitrary precision for inference engine and pixel processing efficiency, pre-emption and QoS scheduling for power efficiency.
- HPC / Data Center: enhanced streaming processing, enhanced library support.
- FPGAs: enhanced execution model, self-synchronized and self-scheduled graphics, fine-grained synchronization between kernels.
- Embedded: arbitrary precision for power efficiency, hard real-time scheduling, async DMA.
Want to learn more about the latest state of OpenCL 2.2, SYCL 2.2, and SPIR-V 1.1 along with other OpenCL compute matters? See the PDF slides.
Sadly there was no word at the OpenCL BoF on when NVIDIA's proprietary driver may finally support OpenCL 2...
Among the items mentioned during the OpenCL BoF @ SIGGRAPH 2016 were possible features on the roadmap of:
- Desktop: Vulkan interoperability, arbitrary precision for increased performance, pre-emption, collective programming and improved execution model.
- Vulkan Compute leveraging OpenCL: types, precision and accuracy, pointers and address spaces, execution model.
- Mobile: arbitrary precision for inference engine and pixel processing efficiency, pre-emption and QoS scheduling for power efficiency.
- HPC / Data Center: enhanced streaming processing, enhanced library support.
- FPGAs: enhanced execution model, self-synchronized and self-scheduled graphics, fine-grained synchronization between kernels.
- Embedded: arbitrary precision for power efficiency, hard real-time scheduling, async DMA.
Want to learn more about the latest state of OpenCL 2.2, SYCL 2.2, and SPIR-V 1.1 along with other OpenCL compute matters? See the PDF slides.
Sadly there was no word at the OpenCL BoF on when NVIDIA's proprietary driver may finally support OpenCL 2...
1 Comment