[Piglit] [PATCH 6/7] ARB_pipeline_statistics_query (frag): basic test
benjamin.widawsky at intel.com
Mon Feb 16 13:15:13 PST 2015
On Mon, Feb 16, 2015 at 03:42:45PM -0500, Ilia Mirkin wrote:
> On Mon, Feb 16, 2015 at 2:59 PM, Ben Widawsky
> <benjamin.widawsky at intel.com> wrote:
> > I think we all agree there is no point in holding up the series for this, right?
> Yeah, just a question of what actually makes it in :) Given that the
> counter is returning BS values for unknown reasons, I think the test
> should just fail on the affected platforms until you figure out a
> better upper bound. There's no shame in having failing tests, esp due
> to hardware idiocy.
I don't mind failing tests, though I suspect until a developer actually wants
this, it won't ever get fixed by us. Although, as I said on IRC, I think I'd
like to do 2 tests for this then - one proving PS invocations is non-zero if
QUERY_COUNTER_BITS is non-zero (since I believe that should be impossible no
matter the HW implementation), the other expects a reasonable value, for some
definition of reasonable which I've yet to define.
In all, this extension is a bit of a trainwreck to write good tests for. I had
secretly hoped one of the authors would chime in with their input.
Anyhoo... will be a while before I have time to work on this again. Let's hope I
make 10.6 :P
Ben Widawsky, Intel Open Source Technology Center
More information about the Piglit