| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
|
|
|
|
|
|
| |
Without this initialization the temp registers used in tgsi_declaration
may used random indices, and this may result in failing translation from TGSI
with an error message "GPR limit exceeded", because the random index is greater
then the allowed limit implying that the shader uses more temporary registers then
available.
Signed-off-by: Gert Wollny <[email protected]>
Cc: <[email protected]>
Signed-off-by: Dave Airlie <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
First try with a "soft" depth, to try to schedule sfu instructions
further from their consumers, but fall back to hard depth (which might
result in stalling) if nothing else is avail to schedule.
Previously the consumer of a sfu instruction could end up scheduled
immediately after (since "hard" depth from sfu to consumer would be 0).
This works because legalize pass would insert a (ss) sync bit, but it
is sub-optimal since it would cause a stall.
Instead prioritize other instructions for 4 cycles if they would no
cause a nop to be inserted. This minimizes the stalling. There is a
slight penalty in general to overall # of instructions in shader (since
we could end up needing nop's later due to scheduling the "deeper" sfu
consumer later), but ends up being a wash on register pressure.
Overall this seems to be worth a 10+% gain in fps. Increasing the
"soft" depth of sfu consumer beyond 4 helps a bit in some cases, but 4
seems to be a good trade-off between getting 99% of the gain and not
increasing instruction count of shaders too much.
It's possible a similar approach could help for tex/mem instructions,
but the (sy) sync bit seems to trigger a switch to a different thread-
group to hide memory latency (possibly with some limits depending on
number of registers used?).
Signed-off-by: Rob Clark <[email protected]>
|
|
|
|
|
|
|
|
| |
If the blit isn't changing format, but is changing tiling, just lie and
call things ARGB (since the exact component order doesn't matter for a
tiling blit).
Signed-off-by: Rob Clark <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Overall a nice 5-10% gain for most games. And more for things like
glmark2 texture benchmark.
There are some rough edges. In particular, the hardware seems to only
support tiling or component swap. (Ie. from hw PoV, ARGB/ABGR/RGBA/
BGRA are all the same format but with different component swap.) For
tiled formats, only ARGB is possible. This isn't a big problem for
*sampling* since we also have swizzle state there (and since
util_format_compose_swizzles() already takes into account the component
order, we didn't use COLOR_SWAP for sampling). But it is a problem if
you try to render to a tiled BGRA (for example) surface.
The next patch introduces a workaround for blitter, so we can generate
tiled textures in ABGR/RGBA/BGRA, but that doesn't help the render-
target case. To handle that, I think we'd need to keep track that the
tiled format is different from the linear format, which seems like it
would get extra fun with sampler views/etc.
So for now, disabled by default, enable with FD_MESA_DEBUG=ttile. In
practice it works fine for all the games I've tried, but makes piglit
grumpy.
Signed-off-by: Rob Clark <[email protected]>
|
|
|
|
| |
Signed-off-by: Rob Clark <[email protected]>
|
|
|
|
|
|
|
|
| |
The rules are sufficiently different for a5xx with tiled textures, so
split this out into something that can be implemented per-generation.
The a5xx specific implementation will come in a later patch.
Signed-off-by: Rob Clark <[email protected]>
|
|
|
|
| |
Trivial. Found by Coccinelle.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
assert() is replaced by unreachable(), to avoid following building error:
external/mesa/src/gallium/drivers/radeonsi/si_shader.c:1967:1:
error: control may reach end of non-void function [-Werror,-Wreturn-type]
}
^
1 error generated.
Fixes: c797cd6 ("ac: add load_patch_vertices_in() to the abi")
Reviewed-by: Bas Nieuwenhuizen <[email protected]>
Reviewed-by: Timothy Arceri <[email protected]>
|
|
|
|
|
|
|
|
| |
Fixes a bunch of arb_gpu_shader_fp64 piglit tests for example:
generated_tests/spec/arb_gpu_shader_fp64/execution/built-in-functions/fs-mix-double-double-double.shader_test
Reviewed-by: Bas Nieuwenhuizen <[email protected]>
|
|
|
|
|
|
|
|
| |
I had 3.x putting swizzling in the texture state only for 16-bit texture
returns, and in the shader for 32-bit. This may be due to having mixed up
the return channel setup on 3.x back before I had moved it into the
compiler. On 4.x, the non-border-color texwrap tests are passing nicely
with both 16 and 32-bit returns with swizzling in the texture state.
|
| |
|
|
|
|
|
| |
Now that the actions are reused for centroid and nonperspective, give them
a more generic name.
|
|
|
|
|
| |
The fxcd/fycd instructions now return half-integer pixel centers when not
doing sample-rate shading.
|
|
|
|
| |
Revealed that I was writing past the TSDA, not the Z buffer as I expected.
|
| |
|
|
|
|
|
| |
We no longer have the small depth-specific output format enum, and instead
depth is just at the end of the output image format enum.
|
|
|
|
|
| |
The RGBX8 formats were dropped from V3D 4.x, but we don't really need them
anyway (we already handle other non-alpha formats by forcing A to 1).
|
| |
|
|
|
|
|
| |
This is a major performance boost on all of V3D, but is required on V3D
4.x where shaders are always either 2- or 4-threaded.
|
|
|
|
|
| |
This required extending the CL submit ioctl, because the tile alloc/state
buffer setup has moved from the BCL to register writes.
|
| |
|
|
|
|
|
|
|
| |
This required moving the register accesses to a separate v3dx file, since
the register definitions for each V3D version collide. It seems that
initializing the v3d_hw from a file dictating 3.3
(v3d_simulator_wrapper.cpp) is safe, though.
|
|
|
|
|
|
|
| |
The TLB load/store path is rebuilt in this version. There is no longer a
single-byte resolved store or the 3-byte extended store. Instead, you get
to always use general loads/stores (which, honestly, was tempting even in
previous versions).
|
|
|
|
|
| |
I accidentally emitted this into the RCL instead of the per-tile generic
list, so we wouldn't get tiles after the first cleared.
|
|
|
|
|
| |
This is going to get more complicated with V3D 4.1 support, which redoes
all the TLB packets.
|
|
|
|
|
|
| |
To conditionally compile cl_emit() macros per V3D version, we need it to
expand to whatever V3D we're building for. This required emitting #define
V3D_VERSION 33 in all our currently 3.3-only code.
|
|
|
|
|
|
|
|
|
| |
This simplifies the build by removing the need to link targets against
libsensors.
Suggested-by: Emil Velikov <[email protected]>
Signed-off-by: Dylan Baker <[email protected]>
Acked-by: Eric Engestrom <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This creates two new internal dependencies, idep_nir_headers and
idep_nir. The former encapsulates the generation of nir_opcodes.h and
nir_builder_opcodes.h and adding src/compiler/nir as an include path.
This ensures that any target that needs nir headers will have the
includes and that the generated headers will be generated before the
target is build. The second, idep_nir, includes the first and
additionally links to libnir.
This is intended to make it easier to avoid race conditions in the build
when using nir, since the number of consumers for libnir and it's
headers are quite high.
Acked-by: Eric Engestrom <[email protected]>
Signed-off-by: Dylan Baker <[email protected]>
|
|
|
|
|
|
|
| |
Don't use intermediate variables, use consistent whitespace.
Acked-by: Eric Engestrom <[email protected]>
Signed-off-by: Dylan Baker <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Currently the meosn build has a mix of two styles:
arg : [foo, ...
bar],
and
arg : [
foo, ...,
bar,
]
For consistency let's pick one. I've picked the later style, which I
think is more readable, and is more common in the mesa code base.
v2: - fix commit message
Acked-by: Eric Engestrom <[email protected]>
Signed-off-by: Dylan Baker <[email protected]>
|
|
|
|
|
|
|
| |
No need for a goto.
Reviewed-by: Neha Bhende <[email protected]>
Reviewed-by: Charmaine Lee <[email protected]>
|
|
|
|
|
| |
Reviewed-by: Neha Bhende <[email protected]>
Reviewed-by: Charmaine Lee <[email protected]>
|
|
|
|
|
| |
Reviewed-by: Neha Bhende <[email protected]>
Reviewed-by: Charmaine Lee <[email protected]>
|
|
|
|
|
| |
Reviewed-by: Neha Bhende <[email protected]>
Reviewed-by: Charmaine Lee <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Fixes the follow test for radeonsi nir:
tests/spec/arb_tessellation_shader/execution/quads.shader_test
Also stops 8 other tests from crashing, they now just fail e.g.
tcs-output-array-float-index-rd-after-barrier.shader_test
Reviewed-by: Marek Olšák <[email protected]>
Reviewed-by: Samuel Pitoiset <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Many of the functions declared in tgsi_build.h return structs (not struct
pointers). Therefore the full struct definitions are needed to avoid
warnings or errors:
In file included from src/gallium/drivers/nouveau/codegen/nv50_ir_from_tgsi.cpp:23:
external/mesa3d/src/gallium/auxiliary/tgsi/tgsi_build.h:47:1: error: 'tgsi_build_header' has C-linkage specified, but returns incomplete type 'struct tgsi_header' which could be incompatible with C [-Werror,-Wreturn-type-c-linkage]
This error shows up on Android builds using clang and -Werror.
Cc: Ilia Mirkin <[email protected]>
Signed-off-by: Rob Herring <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
BuilderSWR::swr_gs_llvm_fetch_input() (and consequently
swr_gs_llvm_fetch_input()), did not handle the case where
is_vindex_indirect or is_aindex_direct is set.
Implement it, using the code in draw_llvm.c as a guideline.
Fixes the following piglit tests:
dynamic_input_array_index (crash)
gs-input-array-vec4-index-rd
vs-output-array-vec4-index-wr-before-gs
Reviewed-by: Bruce Cherniak <[email protected]>
|
|
|
|
|
|
|
| |
v2: move those helpers to the header and use static inline
Signed-off-by: Samuel Pitoiset <[email protected]>
Reviewed-by: Bas Nieuwenhuizen <[email protected]> (v1)
|
|
|
|
|
|
| |
Allows for call-stack and exception handling for jitted functions.
Reviewed-by: Bruce Cherniak <[email protected]>
|
|
|
|
|
|
|
|
| |
Using a gather for elements less than 32-bits in size can cause
pagefaults when loading the last elements in a page-aligned-sized
buffer.
Reviewed-by: Bruce Cherniak <[email protected]>
|
|
|
|
|
|
| |
Results in far smaller and useful IR output.
Reviewed-by: Bruce Cherniak <[email protected]>
|
|
|
|
|
|
|
|
| |
Bake in USE_SIMD16_BUILDER code paths (for USE_SIMD16_SHADER defined),
remove USE_SIMD16_BUILDER define, remove deprecated psuedo-SIMD16 code
paths.
Reviewed-by: Bruce Cherniak <[email protected]>
|
|
|
|
| |
Reviewed-by: Bruce Cherniak <[email protected]>
|
|
|
|
| |
Reviewed-by: Bruce Cherniak <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Similar to const buffers. The driver must not emit any tes-related state if tes
is disabled, since the hw slots are all shared by VS, therefore it would
overwrite them (the mesa state tracker might not do this, but it would be
perfectly legal to do so).
Nevertheless I think the dirty state tracking logic in the driver is
fundamentally flawed when tes is disabled/enabled, since it looks to me like
the VS (and TES) state would not get reemitted to the correct slots (if it's
not dirty anyway). Unless I'm missing something...
Theoretically, the overwrite problem could be solved by using non-overlapping
resource slots for TES and VS (since we're not even close to using half the
resource slots), but it wouldn't work for constant buffers nor samplers, and
for VS would still need to propagate changes to both LS and VS, so probably
not a useful idea.
Unfortunately there's zero coverage of this with piglit, since all tessellation
shader tests are just shader_runner tests, which are unsuitable for testing
any kind of state dependency tracking issues (so I can't even quickly hack
something up to proove it and fix it...).
TCS otoh is just fine - like GS it has its own hw slots.
Tested-by: Konstantin Kharlamov <[email protected]>
Reviewed-by: Dave Airlie <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
With the exception of the default tess levels only ever accessed
by the default tcs shader, the LDS_INFO const buffer was only accessed by vtx
instructions, and not through kcache. No idea why really, but use this to our
advantage by not using a constant buffer slot for it. This just requires us to
throw the default tess levels into the "normal" driver const buffer instead.
Alternatively, could acesss those constants via vtx instructions too, but then
we couldn't use a ordinary ureg prog accessing them as constants and would have
to generate that directly when compiling the default tcs shader. (Another
alternative would be to put all lds info into the ordinary driver const
buffer, albeit we'd maybe need to increase the fixed size as it can't fit
alongside the ucp since vs needs access to the lds info too.)
Tested-by: Konstantin Kharlamov <[email protected]>
Dave Airlie <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Contrary to what the comment said, this appears to work just fine on my rv770
(tested with piglit textureSize 140 fs/vs samplerBuffer).
Dave Airlie confirmed it working on cayman too.
I have no clue though if it's actually preferrable to use it (unfortunately
we cannot get rid of the tex constants completely, as we still require them
for cube map txq).
Albeit filling in the format (1 channels or 4?) and the stuff related to mega-
or mini-fetch (what the hell is this...) is just a guess based on other usage
of vtx fetch instructions...
v2: it really needs to be done through texture cache (I botched the
testing because sb optimizations turned it automatically into tc, but
can't rely on it and isn't happening on tes).
Tested-by: Konstantin Kharlamov <[email protected]>
Reviewed-by: Dave Airlie <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Ideally we'd support 16 (d3d11 requires 15, and mesa subtracts one for non-ubo
constants), but that's kind of impossible (it would be only doable if either
we'd somehow merge the mesa non-ubo constants with the driver constants, or
only use the driver constants with vtx fetch instead of through the kcache
mechanism - the latter probably wouldn't be too bad).
For now just do as the comment already said, place the gs ring (not really
a const buffer in any case) which is only ever referred to through vc fetch
clauses at index 16. Throw in a couple asserts for good measure to make sure
the hw limit isn't exceeded.
Tested-by: Konstantin Kharlamov <[email protected]>
Reviewed-by: Dave Airlie <[email protected]>
|
|
|
|
|
|
|
|
|
| |
We only did this for the other stages, but obviously tess eval/ctrl need it
too.
This fixes the (newly modified) piglit texturing/textureSize test when run
with tes stage and bufferSampler.
Reviewed-by: Dave Airlie <[email protected]>
|
|
|
|
|
|
|
| |
It looks like this reloc belongs to setting the constant reg, which is skipped
for gs ring.
Reviewed-by: Dave Airlie <[email protected]>
|