This PR adds conversion template to convert between Low Precision float
formats. These include Binary32 floats and lower. It also adds support
to convert between unsigned and signed float formats and float formats
with different mantissa and exponents.
Additionally overflows (values that don't fit in the target float
format) will be clamped to the maximum value.
**Reasoning**:
Up to now the Vulkan backend only supported float and half float
formats, but to support workbench, 11 and 10 unsigned floats have to be
supported as well. The available libraries that support those float
formats targets scientific applications. Where the final code couldn't
be optimized that well by the compiler.
Data conversion for color pixels have different requirements about
clamping and sign, what could eliminate some clamping code in other
areas in Blender as well. Also could fix some undesired overflow when
using pixels with high intensity that didn't fit in the texture format
leading to known artifects in Eevee and slow-down in the image editor.
**Future**
In the future we might want to move this to the public part of the GPU
module so we can use this as well in other areas (Metal backend), Imbuf clamping
See 3c658d2c2e69e9cf97dfaa7a3c164262aefb9e76 for a commit that uses
this and improves image editor massively as it doesn't need to reiterate over
the image buffer to clamp the values into a known range.
Pull Request: https://projects.blender.org/blender/blender/pulls/108168
There produce unneeded empty lines in the
console. They are just relic from the time
these message were using printf.
Also remove some redundant informations in
the messages themselves.
This PR silences console output during statup phase of blender. During
startup logging isn't yet initialized and print statements where used.
Logging is initialized during the first construction of a Metal Context.
The console prints are now hidden by behind the '--debug-gpu' command
line option.
Pull Request: https://projects.blender.org/blender/blender/pulls/108593
In the specification of validation layers it is mentioned that
the messenger should always return VK_FALSE. When this isn't the
case VK_ERROR_VALIDATION_FAILED_EXT will be returned and the
API command will fail.
This PR will always return false.
Pull Request: https://projects.blender.org/blender/blender/pulls/108545
This change will register a callback to Vulkan debug messaging
allowing us to filter out messages and log them using CLog.
This change only provides the initial changes, and more tweaks
are expected to on formatting and filtering.
Pull Request: https://projects.blender.org/blender/blender/pulls/107635
Excessive memory pool bloating could occur for certain workloads in
Metal. Particularly those which continuously allocate increasingly
large buffers with minimal re-use of existing buffers.
New logic added to the memory pool flushes old buffers if they
have not been used for a set period of time. Timing is calibrated
against system resources and overall memory pressure.
Metal memory pressure will run higher than OpenGL, however,
this is an active decision to provide significant performance
improvements for scenarios which allocate lots of memory and
for keeping frames queued in flight without stalling on pending
GPU work.
Authored by Apple: Michael Parkin-White
Pull Request: https://projects.blender.org/blender/blender/pulls/108083
Certain textures require explicit HOST READ support for reading
data back to CPU. Given the compute based texture read path may
be used for this, if format conversion is required, this flag should
also imply general texture read support.
This resolves a bug with GreasePencil fill which relies on this.
Authored by Apple: Michael Parkin-White
Pull Request: https://projects.blender.org/blender/blender/pulls/108463
A lot of files were missing copyright field in the header and
the Blender Foundation contributed to them in a sense of bug
fixing and general maintenance.
This change makes it explicit that those files are at least
partially copyrighted by the Blender Foundation.
Note that this does not make it so the Blender Foundation is
the only holder of the copyright in those files, and developers
who do not have a signed contract with the foundation still
hold the copyright as well.
Another aspect of this change is using SPDX format for the
header. We already used it for the license specification,
and now we state it for the copyright as well, following the
FAQ:
https://reuse.software/faq/
When a descriptor pool cannot allocate a descriptor set in stead
of resulting `VK_ERROR_OUT_OF_POOL_MEMORY` it is adviced that
drivers will return `VK_ERROR_FRAGMENTED_POOL`.
Before this PR the Vulkan Backend crashed as it only checked the
out of pool memory. According to the Vulkan specification it is
adviced to driver developers to report fragmented pool.
The crash happened as no new pool was allocated and no descriptor
set could be allocated anymore.
This change improved the reliability of the vulkan backend to be
able draw an animation in the 3d viewport for half an hour without
crashing. Before this change Blender would crash in a few seconds.
Pull Request: https://projects.blender.org/blender/blender/pulls/108416
This PR adds the ability to only read back an area of a framebuffer
texture. It also adds the ability to read back from the depth
attachment.
Also reduces the amount of needed memory and reduces the CPU cycles
by reading back directly into the memory provided by the user. The
previous implementation wasn't able to do so as the `VKTexture::read`
function always returned a new buffer. The introduced
`VKTexture::read_sub` works on a pre-allocated buffer.
Pull Request: https://projects.blender.org/blender/blender/pulls/108418
Current implementation of the Color Blend State worked for framebuffers
with a single attachment. This PR adds support for Color Blend State for
multiple attachments. It is assumed that the Blend State is the same for
all attachments.
NOTE: Integer based attachments aren't yet supported. In OpenGL it is
assumed that Integer based attachments aren't blended. It is currently
to early to tell if this is also the case for Vulkan. If this assumption
is incorrect we should use multi blend state.
Pull Request: https://projects.blender.org/blender/blender/pulls/108419
This PR adds support for restart indices. Restart indices are for
example used during hair rendering where an index buffer contains
multiple hair strands. When in the index buffer a restart index is
added, a new line will start, instead of continuing extending the
previous line.
Pull Request: https://projects.blender.org/blender/blender/pulls/108421
Adds the initial stage for the grease pencil 3.0 project.
This patch includes:
* New ID and new object type.
* New DNA structures.
* New drawing engine for grease pencil (gpencil-next).
* Tests for the new grease pencil data-type.
* A few operators for conversion, switching modes and (simple) drawing.
Exposed to the user:
* An experimental option to switch to the new grease pencil.
* This will switch the grease pencil render engine to gpencil-next which can only render the new object type.
Current grease pencil objects will no longer render.
* Changing this option currently requires a restart of blender (for the keymap to update).
* A conversion setting in the `Object` > `Convert To` operator.
* A drawing operator in `Draw Mode`.
Pull Request: https://projects.blender.org/blender/blender/pulls/106848
This patch improves how debug groups are displayed
within captures. Passes are now split to align with
debug groups, such that navigation of captures is
more intuitive.
To closer represent useful information, debug groups are now
deferred to align with passes, with the addition of Macros to
control capture display options.
METAL_DEBUG_CAPTURE_MAX_NESTED_GROUPS limits debug
group nesting, and METAL_DEBUG_CAPTURE_HIDE_EMPTY allows
hiding of debug groups which do not contain any commands.
Authored by Apple: Michael Parkin-White
Pull Request: https://projects.blender.org/blender/blender/pulls/108287
Previously, each GHOST Context instantiated its own Metal device
queue. Commands are only synchronized within a queue, this was the
root cause of a number of flickering issues which had previously
been worked-around with synchronization primitives.
New solution uses a shared queue to simplify dependencies and
alleviate possibility of stalls and bugs when resources are modified
or shared across separate GPU command queues.
Authored by Apple: Michael Parkin-White
Pull Request: https://projects.blender.org/blender/blender/pulls/108223
Extract vendor, driver and device information from the physical device
properties.
Note that driver version is implementation dependent and might fail as
it is unclear which driver is being used. An open source driver could
store the driver version in a different way than a closed source driver.
But as it is not clear which driver version is being used it might
extract the incorrect version.
To solve this issue we check if the extracted version makes sense
depending on the version schema of the driver and if they don't match we
use another approach.
Pull Request: https://projects.blender.org/blender/blender/pulls/107872
Resolve an issue where a high resolution texutre 16k x 8k
did not update in metal due to integer overflow of size parameter.
This patch contains several changes to address size correctness
across multiple use cases within the Metal backend.
Authored by Apple: Michael Parkin-White
Pull Request: https://projects.blender.org/blender/blender/pulls/108238
Method didn't modified the ShaderInterface but wasn't usable when trying
to debug a const reference to the ShaderInterface. This PR just adds the
const to the method.
Pull Request: https://projects.blender.org/blender/blender/pulls/108190
This contains the basis of the new overlay engine.
Only a few overlays are ported for proof of concept of the new design.
This new design unifies the selection drawing with the overlay-next engine.
It now becomes responsible of selection in object mode.
For this we create a dedicated shader module that patches the shaders
for selection.
Note that the gizmo selection still uses the occlusion queries and edit-mode
the current selection engine (select_engine.c).
Related task #102179
Related task #102177
Co-authored-by: Clément Foucault <foucault.clem@gmail.com>
Pull Request: https://projects.blender.org/blender/blender/pulls/107734
Adds stencil texture view support for Metal, allowing reading of
stencil component during texture sample/read.
Stencil view creation refactored to use additional parameter in
textureview creation function, due to deferred stencil parameter
causing double texture view creation in Metal, when this should
ideally be provided upfront.
Authored by Apple: Michael Parkin-White
Pull Request: https://projects.blender.org/blender/blender/pulls/107971
