This is the first commit of the several required to support
subprocess-based parallel compilation on OpenGL.
This provides the base API and implementation, and exposes the max
subprocesses setting on the UI, but it's not used by any code yet.
More information and the rest of the code can be found in #121925.
This one includes:
- A new `GPU_shader_batch` API that allows requesting the compilation
of multiple shaders at once, allowing GPU backed to compile them in
parallel and asynchronously without blocking the Blender UI.
- A virtual `ShaderCompiler` class that backends can use to add their
own implementation.
- A `ShaderCompilerGeneric` class that implements synchronous/blocking
compilation of batches for backends that don't have their own
implementation yet.
- A `GLShaderCompiler` that supports parallel compilation using
subprocesses.
- A new `BLI_subprocess` API, including IPC (required for the
`GLShaderCompiler` implementation).
- The implementation of the subprocess program in
`GPU_compilation_subprocess`.
- A new `Max Shader Compilation Subprocesses` option in
`Preferences > System > Memory & Limits` to enable parallel shader
compilation and the max number of subprocesses to allocate (each
subprocess has a relatively high memory footprint).
Implementation Overview:
There's a single `GLShaderCompiler` shared by all OpenGL contexts.
This class stores a pool of up to `GCaps.max_parallel_compilations`
subprocesses that can be used for compilation.
Each subprocess has a shared memory pool used for sending the shader
source code from the main Blender process and for receiving the already
compiled shader binary from the subprocess. This is synchronized using
a series of shared semaphores.
The subprocesses maintain a shader cache on disk inside a
`BLENDER_SHADER_CACHE` folder at the OS temporary folder.
Shaders that fail to compile are tried to be compiled again locally for
proper error reports.
Hanged subprocesses are currently detected using a timeout of 30s.
Pull Request: https://projects.blender.org/blender/blender/pulls/122232
The mesh triangulation data is stored in CPU memory with the same format
as the triangles GPU index buffer. Because of that we can skip creating a
temporary copied owned by the GPU API. One way to do that is to just
upload the data directly and avoid keeping a reference to it. However, we
can only upload GPU data from the main thread with OpenGL, so instead
reference the data and keep track of whether to free it.
When drawing a mesh with a single material and 1.8 million faces, this
change gives a 12-15% improvement in framerate, from about 32 to 37 FPS.
Part of #116901.
Pull Request: https://projects.blender.org/blender/blender/pulls/122175
Multiple generations of Intel GPU have the same issue where multi
texture binding results in invalid operations where the driver
reports that the internal texture format isn't supported.
Previously this was only enabled for UHD devices, but this PR
enables it for any Intel GPU. It was detected to be faulty on
UHD600 and Iris.
Pull Request: https://projects.blender.org/blender/blender/pulls/121479
On windows the OpenGL backend of the UHD630 driver (but could also be other
GPUs that use the same driver) reports of supporting `GL_ARB_multi_bind`.
But when enabling it can result in incorrect bindings and report errors
about unsupported internal texture formats. These are internal driver issues.
Might also fix#107642 as it shows the same error message. EEVEE-Next
relies more on using the same binding slot for the same texture in order
to reduce actual bindings which makes this more prominent.
Pull Request: https://projects.blender.org/blender/blender/pulls/121062
Compute shaders are required since 4.0. There was one occasion where
an older AMD driver failed and support was turned off. This driver
is now marked unsupported.
This PR includes:
- removing the check in viewport compositing
- remove properties from system info
- always construct draw manager.
- remove unused pass logic in draw hair/curves
- add deprecation warning when accessed from python
Pull Request: https://projects.blender.org/blender/blender/pulls/120909
The specialization constant default hash was implemented in gl_shader.hh
But the same implementation is needed for vulkan. This PR moves the
default hash to a common place where both backends can use it.
Pull Request: https://projects.blender.org/blender/blender/pulls/120889
This happened when the buffer was bound as a SSBO as the
slot was not marked as used.
We do not tag the unbind as we never call `unbind` for
SSBOs. Also we don't track what target type the UBO is
bound to. This could be improved later.
This PR adds a context function to consider all
buffer bindings obsolete. This is in order to
track missing binds and invalid lingering states
accross `draw::Pass`es.
The functions `GPU_storagebuf_debug_unbind_all`
and `GPU_uniformbuf_debug_unbind_all` do nothing
more than resetting the internal debug slot bits
to zero. This is what OpenGL backend does as it
doesn't track the bindings themselves.
Other backends might have other way to detect
missing bindings. If not they should be
implemented separately anyway.
I renamed the function to `debug_unbind_all` to
denote that it actually does something related to
debugging.
This also add SSBO binding check for OpenGL as it
was also missing.
#### Future
This error checking logic is pretty much backend
agnostic. While it would be nice to move it at
`gpu::Context` level, we don't have the resources
for that now.
Pull Request: https://projects.blender.org/blender/blender/pulls/120716
Adds support for subpass transition for AMD/Intel IMR
GPUs. This enables correct functioning of EEVEE Next
deferred lighting pass on AMD platforms.
The emulation is consistent with the OpenGL approach
of generating additional texture bindings in the shader
for subpass inputs, and splitting render passes across
sub-pass boundaries.
Authored by Apple: Michael Parkin-White
Pull Request: https://projects.blender.org/blender/blender/pulls/119784
Now that all relevant code is C++, the indirection from the C struct
`GPUVertBuf` to the C++ `blender::gpu::VertBuf` class just adds
complexity and necessitates a wrapper API, making more cleanups like
use of RAII or other C++ types more difficult.
This commit replaces the C wrapper structs with direct use of the
vertex and index buffer base classes. In C++ we can choose which parts
of a class are private, so we don't risk exposing too many
implementation details here.
Pull Request: https://projects.blender.org/blender/blender/pulls/119825
* Only works on machines with a Qualcomm Snapdragon 8cx Gen3 or above.
Older generation devices are not and will not be supported due to
some driver issues
* Requires VS2022 for building.
* Uses new MSVC preprocessor for sse2neon compatibility.
* SIMD is not enabled, waiting on conversion of blenlib to C++.
Ref #119126
Pull Request: https://projects.blender.org/blender/blender/pulls/117036
This patch adds the maximum number of supported image units to the GPU
capabilities module. Currently, the GPU module assume a maximum of 8
units, so the patch is not currently particularly useful, but we can
consider committing it for the future anyways.
Pull Request: https://projects.blender.org/blender/blender/pulls/119057
Adds an option to set the capture title when using renderdoc
`GPU_debug_capture_begin` has an optional `title` parameter to set
the title of the renderdoc capture.
Pull Request: https://projects.blender.org/blender/blender/pulls/118649
Span is preferrable since it's agnostic of the source container,
makes it clearer that there is no ownership, is 8 bytes smaller,
and can be passed by value.
`GLBatch::draw_indirect` has additional overhead compared to
`GLBatch::draw`, and can become a bottleneck in scenes that require
many draw calls (ie. with too many unique meshes).
The performance difference is almost exclusively caused by the
`GL_COMMAND_BARRIER_BIT` barrier that happens on every call.
This PR adds a `GPU_storagebuf_sync_as_indirect_buffer` function that
can be used to place the barrier only once after filling the indirect
buffer content.
This function is a no-op in Vulkan and Metal since they don't need the
barrier.
Pull Request: https://projects.blender.org/blender/blender/pulls/117561
When a shader performs a geometry shader injectoin to work around
features that are not supported natively on the GPU (viewport,
barycentric coordinates, layered rendering), linking would fail.
The reason was that the geometry shader was stored in a slot that was
patched by the specialization constants, resulting in an empty geometry
shader. An empty shader can be compiled, but doesn't match the interface
with other stages, so the linking would fail.
This fixes the issue that EEVEE crashed on Intel iGPUs. These GPUs
don't support viewports.
Pull Request: https://projects.blender.org/blender/blender/pulls/117440
This PR improves the place when shader stages are attached to glPrograms.
Previously it was done when shaders stages where created, in the function
create_shader_stage.
This PR will attach the shader stages inside link program.
Ensuring that create_shader_stage doesn't alter the program, which isn't
clear in its name.
Pull Request: https://projects.blender.org/blender/blender/pulls/117407
Ensure attachment states and load/store configs don't get out of sync
with the framebuffer layout.
In theory, a Framebuffer could have empty attachments interleaved with
valid ones so checking just the attachments "length" is not enough.
What this does instead is to ensure that valid attachments have a valid
config and that null attachments either don't have a matching config or
have an IGNORE/DONT_CARE one.
Pull Request: https://projects.blender.org/blender/blender/pulls/117073
Generated copies of GLSL sources are kept in a std::string and
it was always accessed by a long living StringRefNull which lead
to potential read from unallocated memory as std::strings are
not null terminated.
Pull Request: https://projects.blender.org/blender/blender/pulls/117120
This PR adds support for specialization constants for the OpenGL
backend. The minimum OpenGL version we are targetting doesn't
have native support for specialization constants. We simulate this
by keeping track of shader programs for each set of specialization
constants that are being used.
Specialization constants can be used to reduce shader complexity
and improve performance as less registry and/or spilling is done.
This requires the ability to recompile GLShaders. In order to do this
we need to keep track of the sources that are used when the shader
was compiled. For static sources we only store references
(`GLSource::source_ref`), for dynamically generated sources we keep
a copy of the source (`GLSource::source`).
When recompiling the shader GLSL source-code is generated for the
constants stored in `Shader::constants`. When compiling the previous
GLSource that contains specialization constants is then replaced
by the new version.
Pull Request: https://projects.blender.org/blender/blender/pulls/116926
This avoid the cost of creating the tiles themselves which uses a lot
texture write. This was a bottleneck on Apple GPUs.
Also the per pixel classification allows us to remove certain checks in
the deferred lighting shader making it faster.
### TODO
- [x] Add gl_FragStencilRefARB support on other backend
- [x] Add workaround for when gl_FragStencilRefARB isnt supported
Pull Request: https://projects.blender.org/blender/blender/pulls/116704
Along with the 4.1 libraries upgrade, we are bumping the clang-format
version from 8-12 to 17. This affects quite a few files.
If not already the case, you may consider pointing your IDE to the
clang-format binary bundled with the Blender precompiled libraries.
This adds some `#line` directive between the
source file injection so that the log parser knowns
which file the errors originated from.
This is then followed by a scan over the combined
source to find out the real row number.
This needed some changes in the `Shader::plint_log`
to skip lines to avoid outputing redundant information.
