Samuel Pitoiset 7015e22cb6 ac/nir: cull triangles/lines when all W positions are zero/NaN ...

It looks like the fixed-func hardware is very slow to cull primitives
with zero pos.w but shader based culling helps a lot.

This fixes a massive performance gap with the FSR2 demo compared to
AMDGPU-PRO, +228% on RDNA2.

Based on my investigation, AMDGPU-PRO seems to always cull these
primitives. Note that disabling NGG culling with AMDGPU-PRO reports the
same performance as RADV without that fix. Also note that the FSR2
sample doesn't specify any cull mode (ie. VK_CULL_MODE_NONE is used),
so this is the only reason PRO was culling more than RADV.

Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/7260
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31891>

2024-10-30 17:09:37 +00:00

..

ac_binary.c

…

ac_binary.h

…

ac_cmdbuf.c

amd: do not emit PA_SU_PRIM_FILTER_CNTL in the common GFX preamble

2024-10-25 07:41:22 +00:00

ac_cmdbuf.h

ac,radeonsi,radv: add common GFX preambles

2024-08-27 14:14:57 +00:00

ac_debug.c

…

ac_debug.h

ac: Add VCN IB parser

2024-09-23 19:25:08 +00:00

ac_descriptors.c

amd/descriptors: set fmask_tile_swizzle for TC-compat CMASK images on GFX8

2024-10-28 08:21:12 +01:00

ac_descriptors.h

…

ac_drm_fourcc.h

…

ac_formats.c

…

ac_formats.h

…

ac_gather_context_rolls.c

…

ac_gpu_info.c

ac/gpu_info: print 32bpp modifiers

2024-09-27 19:21:55 +00:00

ac_gpu_info.h

ac: add ac_gfx103_get_cu_mask_ps()

2024-08-27 14:14:57 +00:00

ac_hw_stage.h

…

ac_ib_parser.c

…

ac_msgpack.c

…

ac_msgpack.h

…

ac_nir_cull.c

ac/nir: cull triangles/lines when all W positions are zero/NaN

2024-10-30 17:09:37 +00:00

ac_nir_helpers.h

ac/nir: Shorten the name of ac_nir_calc_io_offset_mapped.

2024-08-08 16:55:02 +00:00

ac_nir_lower_esgs_io_to_mem.c

ac/nir: Shorten the name of ac_nir_calc_io_offset_mapped.

2024-08-08 16:55:02 +00:00

ac_nir_lower_global_access.c

…

ac_nir_lower_image_opcodes_cdna.c

…

ac_nir_lower_ngg.c

treewide: don't lower to LCSSA before calling nir_divergence_analysis()

2024-10-24 10:06:17 +00:00

ac_nir_lower_ps.c

ac/nir: rewrite ac_nir_lower_ps epilog to fix dual src blending with mono PS

2024-10-21 12:06:14 +00:00

ac_nir_lower_resinfo.c

ac/nir: set .image_dim and .image_array for all opcodes

2024-09-27 19:21:55 +00:00

ac_nir_lower_subdword_loads.c

…

ac_nir_lower_taskmesh_io_to_mem.c

…

ac_nir_lower_tess_io_to_mem.c

ac/nir/tess: Remove no_inputs_in_lds.

2024-09-05 19:54:29 +00:00

ac_nir_lower_tex.c

nir: change signature of nir_src_is_divergent()

2024-10-24 10:06:17 +00:00

ac_nir_meta_cs_blit.c

ac/nir: set .image_dim and .image_array for all opcodes

2024-09-27 19:21:55 +00:00

ac_nir_meta_cs_clear_copy_buffer.c

ac/nir/meta: tune clear/copy_buffer performance for gfx6-10.3

2024-09-17 15:19:32 -04:00

ac_nir_meta_ps_resolve.c

…

ac_nir_meta.h

ac/nir/meta: tune clear/copy_buffer performance for gfx6-10.3

2024-09-17 15:19:32 -04:00

ac_nir_opt_outputs.c

…

ac_nir_opt_shared_append.c

amd/nir: add ac_nir_opt_shared_append

2024-09-19 16:21:47 +00:00

ac_nir.c

ac/nir: Move varying cost functions from radeonsi to common code.

2024-09-06 09:16:21 +00:00

ac_nir.h

amd/nir: add ac_nir_opt_shared_append

2024-09-19 16:21:47 +00:00

ac_parse_ib.c

ac: Add VCN IB parser

2024-09-23 19:25:08 +00:00

ac_perfcounter.c

…

ac_perfcounter.h

…

ac_pm4.c

amd: use a valid size for ac_pm4_state allocation

2024-07-22 10:09:34 +00:00

ac_pm4.h

…

ac_rgp_elf_object_pack.c

…

ac_rgp.c

ac/rgp: assume GFX11_5 use the same SQTT/RGP versions as GFX11

2024-07-17 16:25:19 +00:00

ac_rgp.h

…

ac_rtld.c

…

ac_rtld.h

…

ac_shader_args.c

…

ac_shader_args.h

…

ac_shader_util.c

nir: always emit ddx intrinsics

2024-10-17 09:50:19 +00:00

ac_shader_util.h

nir: add hole_size parameter into the vectorize callback

2024-10-15 05:50:24 +00:00

ac_shadowed_regs.c

…

ac_shadowed_regs.h

…

ac_spm.c

ac/spm: do not abort when the SPM BO is too small

2024-10-29 18:33:17 +00:00

ac_spm.h

ac/spm: do not abort when the SPM BO is too small

2024-10-29 18:33:17 +00:00

ac_sqtt.c

…

ac_sqtt.h

…

ac_surface_meta_address_test.c

…

ac_surface_modifier_test.c

…

ac_surface_test_common.h

…

ac_surface.c

ac/radeonsi: compute htile for tile mode RADEON_SURF_MODE_1D on GFX6-8

2024-10-29 16:23:51 +00:00

ac_surface.h

Revert "ac/surface: add RADEON_SURF_VIEW_3D_AS_2D_ARRAY for GFX9+"

2024-10-28 12:47:38 +00:00

ac_uvd_dec.h

…

ac_vcn_av1_default.h

…

ac_vcn_dec.c

…

ac_vcn_dec.h

…

ac_vcn_enc_av1_default_cdf.h

…

ac_vcn_enc.c

ac: Add ac_vcn_init_enc_cmds

2024-09-20 06:58:29 +00:00

ac_vcn_enc.h

radeonsi/vcn: Add support for VCN5 AV1 compound

2024-10-10 09:02:21 +00:00

ac_vcn.h

radv/video: add event support for VCN4

2024-08-26 22:19:09 +00:00

amd_family.c

ac/llvm: use LLVM processor gfx942 for GFX940 when it's available

2024-09-27 19:21:55 +00:00

amd_family.h

…

amd_kernel_code_t.h

…

gfx10_format_table.h

…

gfx10_format_table.py

…

meson.build

amd: Pass addrlib cpp args to the tests

2024-10-14 16:52:31 +00:00

sid_tables.py

…

sid.h

…