| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
| |
And vice versa.
Reviewed-by: Elie Tournier <[email protected]>
|
|
|
|
| |
Reviewed-by: Elie Tournier <[email protected]>
|
|
|
|
| |
Reviewed-by: Elie Tournier <[email protected]>
|
|
|
|
| |
Reviewed-by: Elie Tournier <[email protected]>
|
|
|
|
|
|
| |
Definitely not actually a fused-multiply add.
Reviewed-by: Elie Tournier <[email protected]>
|
|
|
|
| |
Signed-off-by: Elie Tournier <[email protected]>
|
|
|
|
|
|
| |
v2: use mix.
Signed-off-by: Elie Tournier <[email protected]>
|
|
|
|
| |
Signed-off-by: Elie Tournier <[email protected]>
|
|
|
|
| |
Signed-off-by: Elie Tournier <[email protected]>
|
|
|
|
| |
Signed-off-by: Elie Tournier <[email protected]>
|
|
|
|
|
| |
v2: use mix
Signed-off-by: Elie Tournier <[email protected]>
|
|
|
|
|
|
| |
v2: use mix
Signed-off-by: Elie Tournier <[email protected]>
|
|
|
|
| |
Signed-off-by: Elie Tournier <[email protected]>
|
|
|
|
| |
Signed-off-by: Elie Tournier <[email protected]>
|
|
|
|
|
| |
v2: use mix
Signed-off-by: Elie Tournier <[email protected]>
|
|
|
|
|
|
|
| |
v2: use mix and findMSB to optimise.
v3: [Sagar] Fix zFrac0 == 0u case in __normalizeRoundAndPackFloat64
Signed-off-by: Elie Tournier <[email protected]>
|
|
|
|
| |
Signed-off-by: Elie Tournier <[email protected]>
|
|
|
|
| |
Reviewed-by: Kenneth Graunke <[email protected]>
|
| |
|
|
|
|
|
|
| |
v2: use mix.
Signed-off-by: Elie Tournier <[email protected]>
|
|
|
|
|
|
| |
v2: use mix.
Signed-off-by: Elie Tournier <[email protected]>
|
|
|
|
| |
Signed-off-by: Elie Tournier <[email protected]>
|
|
|
|
|
|
|
| |
The following patches will add implementations of various
double-precision operations to this file.
Reviewed-by: Kenneth Graunke <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Will be used to convert the .glsl source file containing software fp64
routines to a .h file that can be included while building the compiler.
This commit contains two squashed together: the first from Ian adding
the utility (with the existing title), and the second from Dylan making
the code both python2 and python3 compatible.
This is somewhat modeled after the xxd utility that comes with Vim.
Signed-off-by: Ian Romanick <[email protected]>
xxd.py: Make python2 and 3 compatible
This makes use of unicode_literals, so that undecorated strings are
considered text (python2 unicode, python3 str) and not bytes in python2
and text in python3. It makes use of io.open, which provides python2
with python3's open behavior (it's an alias in python3), in particular
support for the 't' and 'b' option. Finally, it decorates all of the
string literals with the 'b' prefix, so that python interprets them as
bytes.
I've removed the stdin and stdout options, as python2 always requires
these to be bytes, but python3 always treats them as text (there is a
way to get at the underlying bytes buffer, but that's even more
complexity), and makes the input files required arguments.
In the meson we use the '@INPUT@' shorthand instead of listing each
input, as meson will expand that to [prog_python, '@INPUT0@', @INPUT1@,
..., @OUTPUT@, ...]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Otherwise we can end up with IR that looks like this:
(
(declare (temporary ) vec4 f@8)
(assign (xyzw) (var_ref f@8) (var_ref f) )
(call f16 ((swiz y (var_ref f@8) )))
(assign (xyzw) (var_ref f) (var_ref f@8) )
))
When we really need:
(declare (temporary ) float inout_tmp)
(assign (x) (var_ref inout_tmp) (swiz y (var_ref f) ))
(call f16 ((var_ref inout_tmp) ))
(assign (y) (var_ref f) (swiz y (swiz xxxx (var_ref inout_tmp) )))
(declare (temporary ) void void_var)
The GLSL IR function inlining code seemed to produce correct code
even without this but we need the correct IR for GLSL IR -> NIR to
be able to understand whats going on.
Reviewed-by: Kenneth Graunke <[email protected]>
|
|
|
|
|
|
|
| |
Based on a patch from Tim Arceri, but I had to substantially rewrite it
as a result of the NIR derefs rework.
Reviewed-by: Kenneth Graunke <[email protected]>
|
|
|
|
|
|
|
|
| |
These are broken on a future platform, but it turns out we don't need
to fix them, since they're just type-converting moves with strided
source. Kill them.
Reviewed-by: Iago Toral Quiroga <[email protected]>
|
|
|
|
| |
Reviewed-by: Iago Toral Quiroga <[email protected]>
|
|
|
|
|
|
| |
It's redundant with the functionality provided by lower_regioning now.
Reviewed-by: Iago Toral Quiroga <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This legalization pass is meant to handle situations where the source
or destination regioning controls of an instruction are unsupported by
the hardware and need to be lowered away into separate instructions.
This should be more reliable and future-proof than the current
approach of handling CHV/BXT restrictions manually all over the
visitor. The same mechanism is leveraged to lower unsupported type
conversions easily, which obsoletes the lower_conversions pass.
v2: Give conditional modifiers the same treatment as predicates for
SEL instructions in lower_dst_modifiers() (Iago). Special-case a
couple of other instructions with inconsistent conditional mod
semantics in lower_dst_modifiers() (Curro).
Reviewed-by: Iago Toral Quiroga <[email protected]>
|
|
|
|
| |
Reviewed-by: Iago Toral Quiroga <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Currently the visitor attempts to enforce the regioning restrictions
that apply to double-precision instructions on CHV/BXT at NIR-to-i965
translation time. It is possible though for the copy propagation pass
to violate this restriction if a strided move is propagated into one
of the affected instructions. I've only reproduced this issue on a
future platform but it could affect CHV/BXT too under the right
conditions.
Cc: [email protected]
Reviewed-by: Iago Toral Quiroga <[email protected]>
|
|
|
|
|
|
|
|
|
| |
I triggered this bug while prototyping code for a future platform on
IVB. Could be a problem today though if a strided move is
copy-propagated into a type-converting move with DF destination.
Cc: [email protected]
Reviewed-by: Iago Toral Quiroga <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
was already split.
This seems to be a problem in combination with the lower_regioning
pass introduced by a future commit, which can modify a SIMD-split
instruction causing its execution size to become illegal again. A
subsequent call to lower_simd_width() would hit this bug on a future
platform.
Cc: [email protected]
Reviewed-by: Iago Toral Quiroga <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Align16 is no longer a thing, so a new implementation is provided
using Align1 instead. Not all possible swizzles can be represented as
a single Align1 region, but some fast paths are provided for
frequently used swizzles that can be represented efficiently in Align1
mode.
Fixes ~90 subgroup quad swap Vulkan CTS tests.
Cc: [email protected]
Reviewed-by: Iago Toral Quiroga <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
|
| |
lower_integer_multiplication() implements 32x32-bit multiplication on
some platforms by bit-casting one of the 32-bit sources into two
16-bit unsigned integer portions. This can give incorrect results if
the original instruction specified a source modifier. Fix it by
emitting an additional MOV instruction implementing the source
modifiers where necessary.
Cc: [email protected]
Reviewed-by: Iago Toral Quiroga <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
|
| |
Looks like it is impossible that 'last' variable is a null
because at least the get_vs_prog_data shouldn't return a null pointer.
So this check is unnecessary starts from commit:
99d497c5b691 "anv/pipeline: Replace get_fs_input_map with ..."
This small issue is found by cppcheck.
Signed-off-by: Andrii Simiklit <[email protected]>
Reviewed-by: Jason Ekstrand <[email protected]>
|
|
|
|
|
|
|
|
|
| |
This ensures that during encoding, applications can get
the correct status of the surface before submitting
more operations on the same.
Reviewed-by: Leo Liu <[email protected]>
Signed-off-by: Indrajit Das <[email protected]>
|
|
|
|
|
|
|
|
|
| |
This reverts commit f6a6da8131383d8eeee07cd59326a70f4b15866b.
With this commit we see massive amounts of asserts triggering
in lp_fence_wait(), assert(f->issued), for instance with libgl_xlib
state tracker and piglit. Not entirely sure if the assert could
just be removed.
|
|
|
|
|
|
|
|
|
|
| |
We have found some pipe_surface leaks internally.
This is the same code as surface_destroy in radeonsi.
Ideally, surface_destroy would be in pipe_screen.
Cc: 18.3 <[email protected]>
Reviewed-by: Brian Paul <[email protected]>
|
|
|
|
| |
Reviewed-by: Brian Paul <[email protected]>
|
|
|
|
| |
Reviewed-by: Brian Paul <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
With Mesa 18.1, commit be973ed21f6e, si_llvm_load_input_vs()
changed the number of source 32-bit wide dword components
used for fetching vertex attributes into the vertex shader
from a constant 4 to a variable num_channels number, depending
on input data format, with some special case handling for
input data formats like 64-Bit doubles.
In the case of a GL_DOUBLE input data format with one
or two components though, e.g, submitted via ...
a) glTexCoordPointer(1, GL_DOUBLE, 0, buffer);
b) glTexCoordPointer(2, GL_DOUBLE, 0, buffer);
... the input format would be SI_FIX_FETCH_RG_64_FLOAT,
but no special case handling was implemented for that
case, so in the default path the number of 32-bit
dwords would be set to the number of float input components
derived from info->input_usage_mask. This ends with corrupted
input to the vertex shader, because fetching a 64-bit double
from the vbo requires fetching two 32-bit dwords instead of 1,
and fetching a two double input requires 4 dword fetches
instead of 2, so in these cases the vertex shader receives
incomplete/truncated input data:
a) float v = gl_MultiTexCoord0.x; -> v.x is corrupted.
b) vec2 v = gl_MultiTexCoord0.xy; -> v.x is assigned
correctly, but v.y is corrupted.
This happens with the standard TGSI IR compiled shaders.
Under NIR with R600_DEBUG=nir, we got correct behavior
because the current radeonsi nir code always assigns
info->input_usage_mask = TGSI_WRITEMASK_XYZW, thereby
always fetches 4 dwords regardless of what the shader
actually needs.
Fix this by properly assigning 2 or 4 dword fetches for
one or two component GL_DOUBLE input.
Fixes: be973ed21f6e ("radeonsi: load the right number of
components for VS inputs and TBOs")
Signed-off-by: Mario Kleiner <[email protected]>
Cc: [email protected]
Cc: Marek Olšák <[email protected]>
Signed-off-by: Marek Olšák <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Fixes artifacts in World of Warcraft when Multi-sample Alpha-Test is
enabled with DXVK.
It also fixes artifacts with Fallout 4's god rays with DXVK.
Various piglit interpolateAt*() tests under NIR are also fixed.
v2: formatting fix
update commit message to include Fallout 4 and the Fixes tag
Fixes: f4e499ec791 ('radv: add initial non-conformant radv vulkan driver')
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=106595
Signed-off-by: Rhys Perry <[email protected]>
|
|
|
|
|
|
|
| |
Loosely based on RadeonSI.
Signed-off-by: Samuel Pitoiset <[email protected]>
Reviewed-by: Bas Nieuwenhuizen <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The Vulkan spec 1.1.97 says:
"variablePointers specifies whether the implementation supports
the SPIR-V VariablePointers capability. When this feature is
not enabled, shader modules must not declare the
VariablePointers capability."
As the SPIR-V feature is enabled, we should turn on the
extension feature as well.
All dEQP-VK.spirv_assembly.instruction.compute.variable_pointers.*
pass with the khronos internal repo. Note that a bunch of them
fails with the public repo, but it's expected as they violate the
specification.
Signed-off-by: Samuel Pitoiset <[email protected]>
Reviewed-by: Bas Nieuwenhuizen <[email protected]>
|
|
|
|
|
|
| |
Reviewed-by: Bas Nieuwenhuizen <[email protected]>
Acked-by: Eric Engestrom <[email protected]>
Signed-off-by: Samuel Pitoiset <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
|
| |
From glibc printf(3):
Z A nonstandard synonym for z that predates the appearance of z.
Do not use in new code.
Z may not exist on non-glibc systems. Prefer the standard symbol.
Reviewed-by: Eric Engestrom <[email protected]>
Reviewed-by: Timothy Arceri <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
If there is no last fence, due to no rendering happening yet, just
create a new signaled fence and return it, to match the expectations of
the EGL sync fence API.
Fixes random "Could not create sync fence 0x3003" assertion failures from
Skia on Android, coming from the following code:
https://android.googlesource.com/platform/frameworks/base/+/master/libs/hwui/pipeline/skia/SkiaOpenGLPipeline.cpp#427
Reproducible especially with thread count >= 4.
One could make the driver always keep the reference to the last fence,
but:
- the driver seems to explicitly destroy the fence whenever a rendering
pass completes and changing that would require a significant functional
change to the code. (Specifically, in lp_scene_end_rasterization().)
- it still wouldn't solve the problem of an EGL sync fence being created
and waited on without any rendering happening at all, which is
also likely to happen with Android code pointed to in the commit.
Therefore, the simple approach of always creating a fence is taken,
similarly to other drivers, such as radeonsi.
Tested with piglit llvmpipe suite with no regressions and following
tests fixed:
egl_khr_fence_sync
conformance
eglclientwaitsynckhr_flag_sync_flush
eglclientwaitsynckhr_nonzero_timeout
eglclientwaitsynckhr_zero_timeout
eglcreatesynckhr_default_attributes
eglgetsyncattribkhr_invalid_attrib
eglgetsyncattribkhr_sync_status
v2:
- remove the useless lp_fence_reference() dance (Nicolai),
- explain why creating the dummy fence is the right approach.
Signed-off-by: Tomasz Figa <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The binding is checked against the limits later in the function, so we
need to make sure we don't overflow before the check here.
Fixes this valgrind warning (and sometimes segfault):
==1460== Invalid write of size 4
==1460== at 0x74C98DD: ast_declarator_list::hir(exec_list*, _mesa_glsl_parse_state*) (ast_to_hir.cpp:4943)
==1460== by 0x74C054F: _mesa_ast_to_hir(exec_list*, _mesa_glsl_parse_state*) (ast_to_hir.cpp:159)
==1460== by 0x7435C12: _mesa_glsl_compile_shader (glsl_parser_extras.cpp:2130)
in
dEQP-GLES31.functional.debug.negative_coverage.get_error.compute.
exceed_atomic_counters_limit
Reviewed-by: Timothy Arceri <[email protected]>
|