This should be done after reads are checked and
sgpr_read_by_valu_as_lanemask_then_wr_by_salu is reset. The old version
also skipped checking the reads if the write check passed.
fossil-db (navi31):
Totals from 193 (0.24% of 79395) affected shaders:
Instrs: 3212435 -> 3212735 (+0.01%)
CodeSize: 16462868 -> 16463848 (+0.01%); split: -0.00%, +0.01%
Latency: 19492377 -> 19492462 (+0.00%); split: -0.00%, +0.00%
InvThroughput: 4419705 -> 4419718 (+0.00%); split: -0.00%, +0.00%
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Backport-to: 24.1
Backport-to: 24.2
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30818>
It's possible to have the linux/udmabuf.h header but not libdrm in some setups, like under Termux.
Fixes: 112063a060 ("llvmpipe: Only use udmabuf if header is found")
Acked-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Reviewed-by: Lucas Fryzek <lfryzek@igalia.com>
Signed-off-by: Valentine Burley <valentine.burley@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30853>
It is always implemented by vtn, and it is available to the shader
depending on the create_library option. Either way, no reason for the backend to
worry about this.
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Reviewed-by: Caio Oliveira <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30792>
F2FP.PACK_AB always uses src0 and src1, only F2FP.MERGE_C uses src1 and
src2.
This fixes "f32tof16" instruction test regression on vk3d.
Fixes: e19871bd6a ("nak: Use F2FP for nir_op_pack_half_2x16_split on SM86+")
Signed-off-by: Mary Guillemard <mary.guillemard@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30804>
In forks, every job exists, but no job is automatically run.
Should help with the problem of users clicking on that tempting "play"
button GitLab offers and that has been causing a lot of issues with
wasted resources.
Downside is, users now have to use the `bin/ci/ci_run_n_monitor.sh`
script to be able to run a bunch of jobs at once.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25252>
The default state doesn't seem well defined (or kernel driver bug
maybe?). Let's just set it to disabled on platforms where we're not
using it.
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Cc: mesa-stable
Found-by: Chuansheng Liu <chuansheng.liu@intel.com>
Reviewed-by: Tapani Pälli <tapani.palli@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30841>
To avoid breaking this because it's not enabled by default.
There is a couple of failures because MSAA is still broken with SDMA.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30778>
When encode isn't enabled, refuse the image usage, also use
the correct error on the decode check.
Fixes: 05cd42417f ("radv/video: enable video encoding behind perftest flag")
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30838>
There is two different alignment requirements:
a) IB VA must be aligned to ib_alignment
b) IB size must be aligned to ib_pad_dw_mask
Though RADV was aligning DGC cmdbuf to ib_alignment always, but this is
unnecessary. Using the optimal padding size for DGC cmdbuf removes a
bunch of useless NOPs.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30768>
Previously we determine wave_size of merged shader stages separately,
and ignore the condition which may cause them to be different.
Now we determine the wave_size of the TCS/GS part first, then use the
wave_size for VS/TES part. So that we can condider the previous shader
stage's information when determine the wave_size of TCS/GS, and two
stages in the merged shader can affect each other's wave_size.
This requires si_shader_selector to have two kinds of main part for
wave32 and wave64 when part mode, to be combined with other shader
part with various wave size.
This also enables merged shader stages with different
si_shader_info->has_divergent_loop to use wave32. We'll add another
condition for KHR_shader_subgroup latter.
Signed-off-by: Qiang Yu <yuq825@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30610>
ClustedAnd with bool argument and cluster_size==4 will be lowered
to quad_vote_all. So does ALU nir_iand/ior op with bool src.
OpenGL and Vulkan subgroup clustered_and tests with bool argument
fail when using LLVM. It seems LLVM has bug when quad vote bool
is in complex control flow. So stop using it for now.
Signed-off-by: Qiang Yu <yuq825@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30610>
