Let's just say that this passes dEQP.
If you think the _mesa_float_to_half conversions are costly, you can
enable FP16 uniforms only if the CPU supports F16C, which is fast.
Drivers will control whether this is used, not common code.
ARM will need something that is equivalent to F16C.
Acked-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9050>
It is useful for the precisions of varyings to match across shader
stages at link-time to enable precision lowering optimizations, which
would otherwise require costly draw-time fixups.
The goal is to enable `producer->precision == consumer->precision` to be
an invariant drivers may rely on for linked shaders.
v2: keep transform feedback outputs at mediump - mareko
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com> (v1)
Reviewed-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9050>
Added:
* a pass that renumbers bases of IO intrinsics
* a pass that converts mediump IO to 16 bits, optionally using the new
packed varying slots
* a pass that sets (forces) mediump in IO intrinsics (for testing)
* a pass that remaps VARYING_SLOT_VAR[0..15]_16BIT to VARYING_SLOT_VAR[0..31]
(if some shader stages don't want packed varyings)
* a pass that folds type conversions around texture opcodes into those
opcodes (e.g. tex(f2f32(coord), ..) is changed into tex accepting f16)
* a pass that changes (legalizes) sampler src and dst types based on specified
hw constraints (e.g. derivatives must be the same type as coordinates)
Reviewed-by: Matt Turner <mattst88@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9050>
This allows mediump inputs and outputs to be trivially lowered into packed
16-bit varyings where 1 slot is occupied by 2 16-bit vec4s, without any
packing instructions in NIR and without any conflicts with 32-bit varyings.
The only thing that is changed is IO semantics in intrinsics to get packed
16-bit varyings.
This simplifies supporting 16-bit types for drivers that have 32-bit slots
everywhere except the fragment shader where they can do 16-bit interpolation
on either the low or high half of each slot.
Reviewed-by: Matt Turner <mattst88@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9050>
This breaks CLOn12's handling of CL CTS test_basic vector_creation for char3 (at least).
Removing this cast causes us to try to load from a deref with no alignment info.
Fixes: 99bb2a4d ("nir/opt_deref: Don't remove casts with alignment information")
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10165>
The current handling for SPIR-V memory semantics is very specific to
the wording in the SPIR-V spec, which breaks its handling of OpenCL
(compared to what we had working downstream before merging upstream).
Update/relax the logic here to support CL's barrier(CLK_GLOBAL_MEM_FENCE);
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10165>
Set the final value in si_texture_create_object, so that other places
don't have to derive it redundantly.
The only thing to remember is that HTILE stencil can be enabled when
stencil is not present, and it can be disabled when stencil is present
due to various workarounds.
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10003>
FMASK is usually pretty large. It's better to leave the cache to shaders.
FMASK stores are still cached, but they can be evicted sooner, which is
the same as other color stores. Only DCC, HTILE, and CMASK are cached.
I haven't benchmarked this, but it seems like the right thing to do.
This only affected APUs.
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10003>
bo_free is called on external BOs when there are no objects left which
reference them. The function unmaps the address range associated with
any maps which occured. However, if the BO is busy (not idle), it
doesn't mark the pointer to the start address as invalid. This can lead
to a segfault later on.
At the end of bo_free, these BOs are still present in the handle hash
table. If such a BO is reused (i.e., when a DMABUF with the same handle
is reimported) and the driver attempts to get another mapping, the
bufmgr will incorrectly assume that the map pointer is still valid and
reuse it. This leads to a segfault. Set the pointer to NULL to mark it
as invalid.
Enables iris to run and pass the piglit test,
ext_image_dma_buf_import-reimport-bug.
Cc: mesa-stable
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9230>
Marek Olšák