[Intel-gfx] [PATCH v4] drm/i915/slpc: Optmize waitboost for SLPC

[Belgaumkar, Vinay]

On 10/21/2022 7:11 PM, Dixit, Ashutosh wrote:
> On Fri, 21 Oct 2022 17:24:52 -0700, Vinay Belgaumkar wrote:
> Hi Vinay,
>
>> Waitboost (when SLPC is enabled) results in a H2G message. This can result
>> in thousands of messages during a stress test and fill up an already full
>> CTB. There is no need to request for RP0 if boost_freq and the min softlimit
>> are the same.
>>
>> v2: Add the tracing back, and check requested freq
>> in the worker thread (Tvrtko)
>> v3: Check requested freq in dec_waiters as well
>> v4: Only check min_softlimit against boost_freq. Limit this
>> optimization for server parts for now.
> Sorry I didn't follow. Why are we saying limit this only to server? This:
>
> 		if (slpc->min_freq_softlimit == slpc->boost_freq)
> 			return;
>
> The condition above should work for client too if it is true? But yes it is
> typically true automatically for server but not for client. Is that what
> you mean?
yes. For client, min_freq_softlimit would typically be RPn.
>
>> Signed-off-by: Vinay Belgaumkar <vinay.belgaumkar at intel.com>
>> ---
>>   drivers/gpu/drm/i915/gt/intel_rps.c | 8 +++++++-
>>   1 file changed, 7 insertions(+), 1 deletion(-)
>>
>> diff --git a/drivers/gpu/drm/i915/gt/intel_rps.c b/drivers/gpu/drm/i915/gt/intel_rps.c
>> index fc23c562d9b2..32e1f5dde5bb 100644
>> --- a/drivers/gpu/drm/i915/gt/intel_rps.c
>> +++ b/drivers/gpu/drm/i915/gt/intel_rps.c
>> @@ -1016,9 +1016,15 @@ void intel_rps_boost(struct i915_request *rq)
>> 		if (rps_uses_slpc(rps)) {
>> 			slpc = rps_to_slpc(rps);
>>
>> +			if (slpc->min_freq_softlimit == slpc->boost_freq)
>> +				return;
> nit but is it possible that 'slpc->min_freq_softlimit > slpc->boost_freq'
> (looks possible to me from the code though we might not have intended it)?
> Then we can change this to:
>
> 			if (slpc->min_freq_softlimit >= slpc->boost_freq)
> 				return;
>
>
>> +
>> 			/* Return if old value is non zero */
>> -			if (!atomic_fetch_inc(&slpc->num_waiters))
>> +			if (!atomic_fetch_inc(&slpc->num_waiters)) {
>> +				GT_TRACE(rps_to_gt(rps), "boost fence:%llx:%llx\n",
>> +					 rq->fence.context, rq->fence.seqno);
> Another possibility would have been to add the trace to slpc_boost_work but
> this is matches host turbo so I think it is fine here.
>
>> 				schedule_work(&slpc->boost_work);
>> +			}
>>
>> 			return;
>> 		}
> Thanks.
> --
> Ashutosh