mirror of
https://github.com/KhronosGroup/OpenCL-CTS.git
synced 2026-03-19 06:09:01 +00:00
While the cl_khr_semaphore extension spec does state that are no implicit dependencies between already enqueued commands and the clEnqueueSignalSemaphoresKHR, it's nothing special as this is already true for any other event that's not a barrier or marker. Also, the CTS can't reliably assume implementations to reorder events even in an out of order queue as this is highly implementation defined behavior and implementations may or may not choose to reorder events in a specific order. I don't see a reason why this should be tested for semaphores, but not for any other commands, especially as it imposes a restriction on how to implement out of order queues that wasn't enforced before. Closes: https://github.com/KhronosGroup/OpenCL-CTS/issues/2439