Home | History | Annotate | Download | only in i915

Lines Matching refs:semaphores

391 	 * Are we using semaphores when the gpu is already saturated?
393 * Using semaphores incurs a cost in having the GPU poll a
401 * system is overloaded and using semaphores at this time only
403 * further use of semaphores until we are idle again, whence we
406 if (request->sched.semaphores &&
408 engine->saturated |= request->sched.semaphores;
476 if (request->sched.semaphores && i915_request_started(request)) {
478 request->sched.semaphores = 0;
854 return rq->sched.semaphores | rq->engine->saturated;
926 to->sched.semaphores |= from->engine->mask;
1071 /* Ensure both start together [after all semaphores in signal] */
1247 * precludes optimising to use semaphores serialisation of a single
1352 * Boost actual workloads past semaphores!
1354 * With semaphores we spin on one engine waiting for another,