First remove agx_shader_key from asahi_shader_key. It's trivial. agx_shader_key
is going to go away soon now that we lower everything in NIR.
Then everything else is mutually exclusive between stages. That means much
less to hash. drawoverhead test 1 from 2331 to 3443.
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20123>
Now we support all the vertex formats! This means we don't hit u_vbuf for format
translation, which helps performance in lots of applications. By doing the
lowering in NIR, the vertex fetch code itself can be optimized by NIR (e.g.
nir_opt_algebraic) which can improve generated code quality.
In my first implementation of this, I had a big switch statement mapping format
enums to interchange formats and post-processing code. This ends up being really
unwieldly, the combinatorics of bit packing + conversion + swizzles is
enormous and for performance we want to support everything (no u_vbuf
fallbacks). To keep the combinatorics in check, we rely on parsing the
util_format_description to separate out the issues of bit packing, conversion,
and swizzling, allowing us to handle bizarro formats like B10G10R10A2_SNORM with
no special casing.
In an effort to support everything in one shot, this handles all the formats
needed for the extensions EXT_vertex_array_bgra, ARB_vertex_type_2_10_10_10_rev,
and ARB_vertex_type_10f_11f_11f_rev.
Passes dEQP-GLES3.functional.vertex_arrays.*
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19996>
We can implement glPolygonMode(GL_FRONT_AND_BACK, ...) natively. What we can't
implement natively are two-sided polygon modes. For that Apple has a nontrivial
lowering which I don't feel the need to implement unless someone actually hits a
workload other than Piglit that uses it.
Vulkan requires only one-sided polygon modes (so this is sufficient there), and
GLES doesn't have polygon modes at all. If an app hits the unimplemented case,
throw a warning like Zink does.
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20017>
bind_sampler_states doesn't zero [nr_samplers, PIPE_MAX_SAMPLERS) so can get
non-null garbage samplers leading to a use-after-free (segfault derefencing
sampler) or a buffer overflow (writing samplers[] out).
Fixes crashes in Xonotic.
Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Reported-by: dcow
Tested-by: dcow
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19994>
Better occupancy, which is especially important when the background shader
does memory access (for reloads). On my 4K monitor, glmark2 -bdesktop fullscreen
from 95fps to 133fps.
At default settings, glmark2 -bterrain from 63fps to 71fps.
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19997>
Don't waste 1MB per batch for descriptors that may be completely unused.
Instead, upload the scissor and depth bias arrays at submit time. This is a
simple solution to a silly problem: we can't grow the scissor/depth bias arrays,
and we don't know how big they will be at draw time. We could...
1. Statically allocate large buffers? Waste lots of memory.
2. Statically allocate small buffers? Forces too much flushing.
3. Dynamically allocate a growable GPU buffer? Requires either reading
back write-combined memory contents, or maintaining a CPU copy in
addition to extra GPU copies, or doing complicated MMU shenanigans.
The first two options are slow and the last is complicated.
Instead, we upload these descriptors to a dynamically allocated CPU-side which
gets copied just once to the GPU at submit-time when the exact size is known,
minimizing wasted memory and copies and avoiding any unnecessary flushing or WC
memory reads.
In addition, this patch makes sure we flush if we would overflow with more than
65535 scissor descriptors in a batch. This is a (minor) bug fix.
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19971>
Flag day change to replace the previous hardcoded background/end-of-tile shaders
and the API-style load/store_output in fragment shaders with the generated
shaders and lowered *_agx intrinsics. This gets us working non-UNORM8 render
targets and working MRT. It's also a step in the direction of working MSAA but
that needs a lot more work, since the multisampling programming model on AGX is
quite different from any of the APIs (including Metal).
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19871>
Multisampling uses different values of the dimension enum in tandem with a new
samples field. Handle this in agx_pack_texture (split off here) so we can use
the new functionality for texture descriptors in reloads too.
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19871>
This avoids exposing the ISA-internal agx_format to the driver, instead hiding
it behind a real PIPE_FORMAT. This lets us use real pipe formats in formatted
intrinsics in NIR, which is convenient; it will allow us to simplify the
compiler/driver ABI; and it lets us use common format helpers (e.g.
util_format_get_blocksize) for the internal formats in driver lowering.
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19871>
We already have the notion of an agx_batch, which encapsulates a render
pass. Extend the logic to allow multiple in-flight batches per context, avoiding
a flush in set_framebuffer_state and improving performance for certain
applications designed for IMRs that ping-pong unnecessarily between FBOs. I
don't have such an application immediately in mind, but I wanted to get this
flag-day out of the way while the driver is still small and flexible.
The driver was written from day 1 with batch tracking in mind, so this is a
relatively small change to actually wire it up, but there are lots of little
details to get right.
The code itself is mostly a copy/paste of panfrost, which in turn draws
inspiration from freedreno and v3d.
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19865>
Stencil texturing is easy: S8_UINT is textured like R8_UINT (with a
little swizzle fixup), and stencil is always S8_UINT thanks to
u_transfer_helper. So we just need to do some fixups to make
u_transfer_helper's seperate_stencil work and everything will work out.
Passes dEQP-GLES31.functional.stencil_texturing.*
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19811>
We need to flush the appropriate batch(es in the future) before a synchronized
transfer for correct results. To do so without major performance regressions, we
need to do extra bookkeeping about which batches write which resources. We
already know about reads via the BO list.
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19606>
The big discovery is the "number of uniform registers" field. I learned
about this one accidentally when my preamble shaders weren't working
right, because we had inadvertently hardcoded "at most 32 registers" :-)
In the course of identifying that field, I found that the pipeline
address is used as a tagged pointer, with some unknown field in the
bottom bits and alignment demanded. The XML is updated to account for
this.
I later found that there's also a "number of general purpose registers
used by the preamble shader" field. I missed this one first, because the
encoding is slightly different from the usual "number of general purpose
registers in the main shader" field. The specification is slightly
coarser. I don't know why the hardware needs that
information anyway -- occupancy of the preamble shader should be
irrelevant -- but it's not a big deal.
Finally I found that the "more than 4 textures?" bit is... not that. I
do not yet know what it is, but it is... not that.
These all use the new groups() modifier for GenXML
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18813>
This gets shader-db's runner working, in conjunction with a shader-db ./run
modified to set ASAHI_MESA_DEBUG=precompile. This flag triggers precompiles of
all shaders witha default key so we can exercise the compiler.
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18813>
A lot of code zero-initializes pipe_sampler_state, and sets the states
the non-zero fields manually. This means that normalized_coords is the
"default" setting.
However, setting normalized_coords to true isn't allways allowed, and
we'd need to check PIPE_CAP_TEXRECT first. So it's not really the ideal
default here. There's recently been found quite a bit of bugs in this
area, where the state-tracker didn't properly lower texrects.
Let's switch this around to avoid more bugs like this in the future.
Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18971>
Alyssa Rosenzweig