Lines Matching refs:compute
224 * Long-running compute command buffers, while still using traditional end of
226 fences which get reattached when the compute job is rescheduled.
273 userspace is allowed to use userspace fencing or long running compute
296 faults on GPUs are limited to pure compute workloads.
299 compute side, like compute units or command submission engines. If both a 3D
300 job with a DMA fence and a compute workload using recoverable page faults are
303 - The 3D workload might need to wait for the compute job to finish and release
306 - The compute workload might be stuck in a page fault, because the memory
317 achieved through e.g. through dedicated engines and minimal compute unit
328 fences must complete before a compute job with page fault handling can be
330 made visible anywhere in the system, all compute workloads must be preempted
347 Fences` discussions: Infinite fences from compute workloads are allowed to