The following VulkanCTS 1.1.4 tests will not complete on Braswell hardware:
The tests are crashing due to allocation failure after running for some time which is caused by enormous size of the resulting native code of the shaders.
Initially it was bisected to:
8fdee457a4cb20f4587b5ec817aa1f9325bd5f1c is the first bad commit
Author: Jason Ekstrand <firstname.lastname@example.org>
Date: Thu Jan 10 13:39:05 2019 -0600
anv/pipeline: Move lower_explicit_io much later
Now that nir_opt_copy_prop_vars can properly handle array derefs on
vectors, it's safe to move UBO and SSBO lowering to late in the
pipeline. This should allow NIR to actually start optimizing SSBO
Reviewed-by: Caio Marcelo de Oliveira Filho <email@example.com>
However this commit just triggered the underlying issue - force loop unrolling without any check for resulting unrolled size.
I've sent a merge request to fix this: https://gitlab.freedesktop.org/mesa/mesa/merge_requests/1365