panfrost: correct tile-buffer size for some v7 GPUs

These GPUs had their tilebuffer sizes listed at twice their actual
values. While that still works, it ends up disabling pipelining in some
cases. This gives a significant performance hit, compared to using the
correct values.

But, it turns out to be hard or impossible to trigger at the moment, due
to the limited number of MSAA samples we support. Once that changes,
this is a lot easier to trigger, so let's fix it up.

Reviewed-by: Lars-Ivar Hesselberg Simonsen <lars-ivar.simonsen@arm.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/33925>
This commit is contained in:
Erik Faye-Lund 2025-03-10 21:53:39 +01:00 committed by Marge Bot
parent b4ebffa1aa
commit c99a3298d9

View file

@ -60,11 +60,11 @@ const struct panfrost_model panfrost_model_list[] = {
MODEL(0x6000, 0, "G71", "TMIx", NO_ANISO, 8192, {}),
MODEL(0x6221, 0, "G72", "THEx", 0x0030 /* r0p3 */, 16384, {}),
MODEL(0x7090, 0, "G51", "TSIx", 0x1010 /* r1p1 */, 16384, {}),
MODEL(0x7093, 0, "G31", "TDVx", HAS_ANISO, 16384, {}),
MODEL(0x7090, 0, "G51", "TSIx", 0x1010 /* r1p1 */, 8192, {}),
MODEL(0x7093, 0, "G31", "TDVx", HAS_ANISO, 8192, {}),
MODEL(0x7211, 0, "G76", "TNOx", HAS_ANISO, 16384, {}),
MODEL(0x7212, 0, "G52", "TGOx", HAS_ANISO, 16384, {}),
MODEL(0x7402, 0, "G52 r1", "TGOx", HAS_ANISO, 16384, {}),
MODEL(0x7402, 0, "G52 r1", "TGOx", HAS_ANISO, 8192, {}),
MODEL(0x9091, 0, "G57", "TNAx", HAS_ANISO, 16384, {}),
MODEL(0x9093, 0, "G57", "TNAx", HAS_ANISO, 16384, {}),