On Fri, May 03, 2019 at 03:12:01PM +0100, Chris Wilson wrote: > Quoting Ville Syrjälä (2019-05-03 15:04:57) > > On Mon, Apr 29, 2019 at 07:00:19PM +0100, Chris Wilson wrote: > > > Asking the GPU to busywait on a memory address, perhaps not unexpectedly > > > in hindsight for a shared system, leads to bus contention that affects > > > CPU programs trying to concurrently access memory. This can manifest as > > > a drop in transcode throughput on highly over-saturated workloads. > > > > We can't use the singalling semaphore variant? > > That requires us to broadcast a signal to each engine (for which I can > hear the cries of cross-powerwell wakes), and currently does not work > with execlists + context-id==0. Or at least it failed in my testing. Ah. If only we had MI_MWAIT. -- Ville Syrjälä Intel _______________________________________________ Intel-gfx mailing list Intel-gfx@xxxxxxxxxxxxxxxxxxxxx https://lists.freedesktop.org/mailman/listinfo/intel-gfx
