>>> On Thu, Aug 7, 2014 at 3:59 PM, Alex Deucher <alexdeucher at gmail.com> >>> wrote: >>>> We should be using PFP as much as possible. Does the attached patch help? > Unfortunately not.
Maybe add a readback of the VM base addr pointer to make sure that the write has really reached the SBRM? Otherwise I'm out of ideas as well, Christian. Am 08.08.2014 um 04:38 schrieb Michel D?nzer: > On 08.08.2014 00:55, Alex Deucher wrote: >> Note that there is no PFP (or CE) on the compute queues so we can't >> use PFP (or CE) for compute. > AFAICT cik_hdp_flush_cp_ring_emit() always uses the PFP though. > > >> Note also that the engine bit is not always consistent (for some packets 0 >> = ME, 1 = PFP and for others 1= ME and 0 = PFP). > Ugh. Then we should probably use explicit *_ENGINE_PFP/ME macros instead > of *_ENGINE(lucky_number). :) > > >>> On Thu, Aug 7, 2014 at 3:59 PM, Alex Deucher <alexdeucher at gmail.com> >>> wrote: >>>> We should be using PFP as much as possible. Does the attached patch help? > Unfortunately not. > >
