[igt-dev] [Intel-gfx] [PATCH i-g-t] tests/i915/perf_pmu: Subtest to measure sampling error for 100% busy

Chris Wilson chris at chris-wilson.co.uk
Tue Feb 16 11:44:59 UTC 2021


Quoting Tvrtko Ursulin (2021-02-16 10:50:50)
> From: Tvrtko Ursulin <tvrtko.ursulin at intel.com>
> 
> Test that periodic reads of engine busyness against a constant 100% load
> are within the 5000ppm tolerance when comparing perf timestamp versus
> counter values.

We've previously only included the accuracy tests on platforms that
claim to be accurate. Can we rephrase the goal of the test such that it
sounds more like a universal truth that should also be valid for the
sampling backends? "The busyness should never exceed 100% (within a
margin of error)..." And since accuracy is currently protected by
has_busy_stats, let's try to find a different name.
-Chris


More information about the igt-dev mailing list