These test so called complex loop unrolling in Mesa which just means
unrolling a loop with two exits where the trip count is only known
for one of the exits.
---
.../execution/vs-loop-complex-unroll.shader_test | 66 ++
.../vs-loop-zero-iterations-two-exits.shader_test | 41
On Tue, Sep 13, 2016 at 11:12 PM, Dylan Baker wrote:
> Quoting Dylan Baker (2016-09-13 13:46:11)
>> Quoting Marek Olšák (2016-09-13 13:27:21)
>> > On Tue, Sep 13, 2016 at 8:30 PM, Dylan Baker wrote:
>> > > So, there were a few comments I've addressed,
Quoting Dylan Baker (2016-09-13 13:46:11)
> Quoting Marek Olšák (2016-09-13 13:27:21)
> > On Tue, Sep 13, 2016 at 8:30 PM, Dylan Baker wrote:
> > > So, there were a few comments I've addressed, and Marek acked one patch.
> > > Do y'all want to see a v2, or should I just push
Quoting Marek Olšák (2016-09-13 13:27:21)
> On Tue, Sep 13, 2016 at 8:30 PM, Dylan Baker wrote:
> > So, there were a few comments I've addressed, and Marek acked one patch.
> > Do y'all want to see a v2, or should I just push this?
>
> There are at least 2 piglit regressions
On Tue, Sep 13, 2016 at 8:30 PM, Dylan Baker wrote:
> So, there were a few comments I've addressed, and Marek acked one patch.
> Do y'all want to see a v2, or should I just push this?
There are at least 2 piglit regressions with --process-isolation 0:
On Mon, 2016-09-12 at 23:21 -0700, Eric Anholt wrote:
> Is one supposed to be used somewhere else? Is it supposed to have the
> value "1" at some point?
An earlier iteration of this test tried to verify that the label did in
fact change when you changed it. But it did this by trying to
Adam Jackson writes:
> First test does some basic API error checking, then provokes an error
> against the (implicit) thread object. There is clearly much more to
> cover here, but this at least touches all of the new entrypoints and
> verifies that it errors reasonably when
On 12 September 2016 at 14:55, Marek Olšák wrote:
> On Mon, Sep 12, 2016 at 3:33 PM, Emil Velikov
> wrote:
>> On 9 September 2016 at 14:10, Marek Olšák wrote:
>>> From: Marek Olšák
>>>
>>> Older versions don't
On 12 September 2016 at 16:31, Nicolai Hähnle wrote:
> From: Nicolai Hähnle
>
> This reduces the chance of false positives caused by system load.
From my experience the {arb,ext}-timer-query tests have always been
inconsistent. Even on a consecutive,