[Intel-gfx] [PATCH i-g-t] tests: Add gem_exec_nop_concurrent test

Morton, Derek J derek.j.morton at intel.com
Fri Oct 16 07:17:27 PDT 2015


>
>On Fri, Oct 16, 2015 at 02:00:42PM +0100, Thomas Wood wrote:
>> On 15 October 2015 at 09:05, Derek Morton <derek.j.morton at intel.com> wrote:
>> > This test is based on gem_exec_nop but submits nop batch buffers
>> > concurrently from different threads to check for ring hangs and
>> > other issues during concurrent submissions.
>>
>> Is there any reason not to include this as extra subtests in
>> gem_exec_nop so that related tests are grouped together?
>

gem_exec_nop takes 25 minutes to run on android (with the speed improvement patch I submitted yesterday) gem_exec_nop_concurrent takes a similar time. I would strongly resist combining the two into a test that takes nearly an hour for us to run.

>It's a better fit for ringfill. It already does concurrent clients to the same ring in order to catch exactly the same bugs as described here, so extending it to fill all rings concurrently should be easy.

I will take a look at gem_ringfill

//Derek

>-Chris
>
>--
>Chris Wilson, Intel Open Source Technology Centre
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.freedesktop.org/archives/intel-gfx/attachments/20151016/e834bf68/attachment.html>


More information about the Intel-gfx mailing list