[Mesa-dev] [PATCH] anv/blorp: Emit depth flush and stall prior to HiZ clears
chadversary at chromium.org
Fri Aug 31 21:58:44 UTC 2018
On Fri 31 Aug 2018, Jason Ekstrand wrote:
> On Fri, Aug 31, 2018 at 4:35 PM Nanley Chery <nanleychery at gmail.com> wrote:
> If that doesn't fix it, I think it'd be good to comment that we've
> observed this pipecontrol be necessary for 3DSTATE_WM_HZ_OP.
> I'm happy to add some comments. We just need to decide how big of a hammer to
> use. Like I said above, in GL we pull out a pretty big one.
Speaking from prior experience, I support big flush hammers for hiz.
When a gpu hang happens, it can be incredibly difficult to diagnose it
as a hiz hang. It's better to prevent most of them with a large hammer
rather than diagnose them individually and try to prevent them with
several, single-focus tiny hammers.
Considering the impact of a gpu hang on Chrome (the entire OS may use
a single GL ctx, and for Vulkan it may do the same, using a small number
of VkDevices), we should err on the side of avoiding the gpu hang
rather than err on the side of "too many flushes are bad".
More information about the mesa-dev