On Wed, Oct 17, 2018 at 12:14 AM Keith Packard <keithp@keithp.com> wrote:

> Jason Ekstrand <jason@jlekstrand.net> writes:
>
> > Doing all of the CPU sampling on one side or the other of the GPU
> sampling
> > would probably reduce our window.
>
> True, although as I said, it's taking several µs to get through the
> loop, and the gpu clock tick is far smaller than that, so even adding
> the two values together to make it fit the current implementation won't
> make the deviation that much larger.
>
> > This leaves us with a delta of I + max(P(M), P(R), P(G)).  In
> > particular, any two real-number valued times are, instantaneously,
> > within that interval.
>
> That, at least, would be easy to compute, and scale nicely if we added
> more clocks in the future.
>
> > Personally, I'm completely content to have the delta just be a the first
> > one: a bound on the difference between any two real-valued times.  At
> this
> > point, I can guarantee you that far more thought has been put into this
> > mesa-dev discussion than was put into the spec and I think we're rapidly
> > getting to the point of diminishing returns. :-)
>
> It seems likely. How about we do the above computation for the current
> code and leave it at that?
>

Sounds like a plan.  Note that I should be computed as I = end - start +
monotonic_raw_tick_ns to ensure we get a big enough interval.  Given that
monotonic_raw_tick_ns is likely 1, this doesn't expand things much.

I think a comment is likely also in order.  Probably not containing the
entire e-mail thread but maybe some of my reasoning above?

--Jason