| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Unlike the shader resource binding lists that automatically recognize
in setShaderResources() when a referenced QRhiResource has been rebuilt
in the meantime (create() was called i.e. there may be completely
different native objects underneath), QRhiTextureRenderTarget has no
such thing. This leads to an asymmetric API and requires also rebuilding
the rt whenever an attachment is rebuilt:
rt = rhi->newTextureRenderTarget({ { texture } })
rt->create()
cb->beginPass(rt, ...)
texture->setPixelSize(...)
texture->create()
rt->create() // this should not be needed
cb->beginPass(rt, ...)
Avoid having to do that second rt->create().
Change-Id: If14eaa7aac3530950498bbdf834324d0741a7c4d
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
Reviewed-by: Andy Nichols <andy.nichols@qt.io>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Arrays of textures have always been supported, but we will encounter
cases when we need to work with texture array objects as well.
Note that currently it is not possible to expose only a slice of the
array to the shader, because there is no dedicated API in the SRB,
and thus the same SRV/UAV (or equivalent) is used always, capturing
all elements in the array. Therefore in the shader the last component
of P in texture() is in range 0..array_size-1.
Change-Id: I5a032ed016aeefbbcd743d5bfb9fbc49ba00a1fa
Reviewed-by: Andy Nichols <andy.nichols@qt.io>
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
QVLA itself is non-relocatable due to self references. (ptr pointing
to array[Prealloc] as long as capacity < Prealloc)
Seems we shot ourselves in the foot in multiple places with this.
Pick-to: 6.2 6.2.0
Fixes: QTBUG-96619
Change-Id: I57a2ce539b671326cd352dbe57a1f3d4c46a6456
Reviewed-by: Tobias Koenig <tobias.koenig@kdab.com>
Reviewed-by: Lars Knoll <lars.knoll@qt.io>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Supported on OpenGL (and ES) 3.0+ and everywhere else.
Can also be a render target, targeting a single slice at a time.
Can be mipmapped, cannot be multisample.
Reading back a given slice from a 3D texture is left as a future
exercise, for now it is documented to be not supported.
Upload is going to be limited to one slice in one upload entry,
just like we specify one face or one miplevel for cubemap and
mipmapped textures.
This also involves some welcome hardening of how texture subresources
are described internally: as we no longer can count on a layer index
between 0..5 (as is the case with cubemaps), simply arrays with
MAX_LAYER==6 are no longer sufficient. Switch to sufficiently dynamic
data structures where applicable.
On Vulkan rendering to a slice needs Vulkan 1.1 (and 1.1 enabled on the
VkInstance).
Task-number: QTBUG-89703
Change-Id: Ide6c20124ec9201d94ffc339dd479cd1ece777b0
Reviewed-by: Andy Nichols <andy.nichols@qt.io>
|
|
|
|
|
|
|
| |
Fixes: QTBUG-90770
Change-Id: Icba328c417bcce256e7b44f1d540af7f8e83376b
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
Reviewed-by: Christian Strømme <christian.stromme@qt.io>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Add QRhi APIs to retrieve and reload the contents of the "pipeline
cache".
The only API where there is a true pipeline cache is object is Vulkan
(VkPipelineCache). For OpenGL, the other backend where we support this,
it is simulated with program binaries. The Qt 5 style OpenGL program
binary disk cache continues to work like before, but one has now the
option to do things in a more modern, graphics API agnostic way, that
leads to generating a single blob instead of a large set of files in
some system location, allowing easier "pre-baking" of the cache content.
It is expected that Qt Quick exposes the two new functions in form
if QSG_RHI_ environment variables, thus allowing easy testing and
cache file generation.
As an example for the performance improvements this can give, consider
Vulkan, where we do not have any existing persistent caching mechanism
in place:
Running BenchmarkDemoQt6.exe --scene flythrough --mode demo creates 18
QRhiGraphicsPipeline objects from Qt Quick and Qt Quick 3D.
The total time spent in QRhiGraphicsPipeline::create() during application
startup for these 18 pipelines is 35-40 ms on a given Windows (NVIDIA)
system.
When exporting the pipeline cache contents to a file, and then, in a
subsequent run, reloading the cache contents, this is reduced to 5-7 ms
on the same system, meaning we get a 6-7x improvement.
The generated data is always specific to a given Qt version, RHI
backend, graphics device, and driver version. Much of the implementation
consists of adding and verifying the appropriate header to the blobs
retrieved from the driver, to allow gracefully ignoring data that was
generated with a device or driver that differs from the one used at
run time. This should provide robustness, even if the Vulkan or OpenGL
implementation is for some reason not prepared to identity and reject
incompatible cache/program blobs.
Fixes: QTBUG-90398
Change-Id: I67b197f393562434f372c7b7377f638abab85cb3
Reviewed-by: Andy Nichols <andy.nichols@qt.io>
|
|
|
|
|
|
|
| |
Task-number: QTBUG-90321
Pick-to: 6.0
Change-Id: If3b0841f3e9139bb1911c6a5d03a16daf8c1b3d6
Reviewed-by: Laszlo Agocs <laszlo.agocs@qt.io>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
...to the extent it is sensible. We have to make compromises still,
meaning some fields will only be applicable with certain APIs.
Most of this is already shown upon QRhi::create() as info debug
prints, when enabled. Now expose it all through the QRhi API as
well.
This is useful for printing in qtdiag, and, while it should be
avoided as much as possible, to make decisions about disabling
3D rendering features depending on the driver and GPU in use.
Change-Id: Iebe1e192965c928b82a094d1c7c50ddf4b38b9a2
Reviewed-by: Andy Nichols <andy.nichols@qt.io>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Q_MOVABLE_TYPE was conceived before C++ had move semantics. Now, with
move semantics, its name is misleading. Q_RELOCATABLE_TYPE was
introduced as a synonym to Q_MOVABLE_TYPE. Usage of Q_MOVABLE_TYPE
is discouraged now. This patch replaces all usages of Q_MOVABLE_TYPE
by Q_RELOCATABLE_TYPE in QtBase. As the two are synonymous, this
patch should have no impact on users.
Pick-to: 6.0
Change-Id: Ie653984363198c1aeb1f70f8e0fa189aae38eb5c
Reviewed-by: Lars Knoll <lars.knoll@qt.io>
Reviewed-by: Andrei Golubev <andrei.golubev@qt.io>
|
|
|
|
|
|
| |
Change-Id: Iac2645ebd1d42753817078f194ba61520f5f70c9
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
Reviewed-by: Laszlo Agocs <laszlo.agocs@qt.io>
|
|
|
|
|
| |
Change-Id: I2e2ff5f4b8aa91d0accb01108a5199b98c371455
Reviewed-by: Christian Strømme <christian.stromme@qt.io>
|
|
|
|
|
| |
Change-Id: Ib11ea33eb695d4599f4f040415d497aaf19cb15b
Reviewed-by: Andy Nichols <andy.nichols@qt.io>
|
|
|
|
|
| |
Change-Id: I02c1f8c32c08d39cde9845d20ba8b02541d9d325
Reviewed-by: Andy Nichols <andy.nichols@qt.io>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
ExternalContentsInPass becomes a per-pass flag now. Why is this
beneficial? Because while Qt Quick has no choice for its render
pass, not being able to guess if the application wants to do some
native rendering in there, Quick 3D's render passes, all the ones
that are under Quick3D's control, do not have native rendering
from the application in them, and so using secondary command
buffers with Vulkan is not necessary.
Introduce something similar for compute and OpenGL. By knowing that
none of the resources used in a pass are used with a compute pass
(e.g. because we know that there are no compute passes at all) a small
amount of time can be saved by skipping tracking buffers and textures
because the only purpose of said tracking is to generate barriers that
are relevant only to compute.
Change-Id: I0eceb4774d87803c73a39db527f5707a9f4d75c1
Reviewed-by: Andy Nichols <andy.nichols@qt.io>
|
|
|
|
|
|
|
|
| |
Use a simple and straightforward container that only does what
we need here.
Change-Id: I1a81b53a58bc91d533e3d7df5471a1362046825d
Reviewed-by: Andy Nichols <andy.nichols@qt.io>
|
|
|
|
|
| |
Change-Id: I4ae92e6c8c91111a4593c51ee05443b3bc806c35
Reviewed-by: Andy Nichols <andy.nichols@qt.io>
|
|
|
|
|
|
|
|
|
| |
Also bump the non-heap buffer size in the binding list to 16,
in order to accommodate complex Quick3D materials with many
associated texture maps.
Change-Id: Id190e5f8304f5941cffc41a2605fce45dfeb72f0
Reviewed-by: Andy Nichols <andy.nichols@qt.io>
|
|
|
|
|
| |
Change-Id: Ia1aab06214be802aaabc97ffefa28947e11148e3
Reviewed-by: Andy Nichols <andy.nichols@qt.io>
|
|
|
|
|
|
|
|
| |
Because having profilers bombarded with mallocs (due to creating
deep copy QByteArrays) is not nice.
Change-Id: I848f41f3465d6dc2a58a193cc863495aacf13d79
Reviewed-by: Andy Nichols <andy.nichols@qt.io>
|
|
|
|
|
|
|
|
| |
More of an enabler for reusing the data in the individual entries since
not clearing the QVLA does not give us much on its own.
Change-Id: Ief9761f75382c3373cc2bc7b866eb59fdd8b3277
Reviewed-by: Andy Nichols <andy.nichols@qt.io>
|
|
|
|
|
| |
Change-Id: If47eddf3fe7d365c80b0a15712ef155a6898d904
Reviewed-by: Andy Nichols <andy.nichols@qt.io>
|
|
|
|
|
|
|
|
|
| |
Rather, utilize all the available ones in the pool, picking
the next available batch after the one we picked previously
(with wrapping over as necessary).
Change-Id: I5f26e127a406c2dd07d155712429c72ad4f0f0f1
Reviewed-by: Andy Nichols <andy.nichols@qt.io>
|
|
|
|
|
|
|
|
|
|
|
| |
Instead of cryptic assertions and crashes depending on the backend,
show some useful warnings (in debug builds only) when one tries to
create an srb with a list where there are duplicated bindings. (a
mistake that happens relatively often during the development of
frameworks, such as Quick 3D, on top)
Change-Id: If1b50a2e8165b001878ad566e048f146e636514f
Reviewed-by: Andy Nichols <andy.nichols@qt.io>
|
|
|
|
|
|
|
|
|
| |
For all of these we know in advance that the vast majority of usages
will not exceed a certain number of elements. Also, none of these are
copied or moved ever.
Change-Id: I48aedf143e221dc178d661e23454d1e4fb7a271b
Reviewed-by: Andy Nichols <andy.nichols@qt.io>
|
|
|
|
|
|
|
|
| |
Applied to headers only. Source file to be changed separately.
Task-number: QTBUG-84469
Change-Id: Ic08a899321eaffc46b8461aaee3dbaa4d2c727a9
Reviewed-by: Laszlo Agocs <laszlo.agocs@qt.io>
|
|
|
|
|
|
|
|
|
|
| |
For historical reasons we use build and release instead of create and
destroy. This becomes confusing now that more modules in Qt start taking
QRhi into use. Migrate to the more familiar naming, so those who have
used QWindow or QOpenGLContext before will find it natural.
Change-Id: I05eb2243ce274c59b03a5f8bcbb2792a4f37120f
Reviewed-by: Eirik Aavitsland <eirik.aavitsland@qt.io>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Fixes Coverity warnings, and avoids hard-to-debug errors, at minimal
overhead.
Change-Id: I3ff530a9263693d1123932458b3e186e79a14b7e
Coverity-Id: 263692
Coverity-Id: 263693
Coverity-Id: 263699
Coverity-Id: 263700
Coverity-Id: 263702
Coverity-Id: 263705
Pick-to: 5.15
Reviewed-by: Laszlo Agocs <laszlo.agocs@qt.io>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
When x or y are >= the width or height of the render target, then the
width or height of the scissor/viewport rect is zero, no further logic
is needed.
This is different from the case of x or y being negative, because then
there is still a chance that there is an in-bounds area (if width or
height are large enough).
It is important to make this check based on the original value of x
and y, not the clamped ones. Otherwise we end up with a 1 pixel wide
region even when the expected result is a width or height of 0.
Previously the incorrect subtraction of 1 in the final clamping of w and h
masked this, but once that is fixed, the issue fixed here becomes visible
in the cubemap_scissor manual test.
Change-Id: I3d4b0a163a16aa1116b1e838fa95c0faf7b56a3d
Reviewed-by: Eirik Aavitsland <eirik.aavitsland@qt.io>
|
|
|
|
|
|
|
|
|
|
|
| |
In cases where qrhi_toTopLeftRenderTargetRect() would clip the width or height
to the available space, it would subtract 1 from the result, leading to
painting errors.
Fixes: QTBUG-83928
Change-Id: I65d23151d838386b516ded0588702bc0bf4c0d93
Reviewed-by: Jonas Karlsson <jonas.karlsson@qt.io>
Reviewed-by: Laszlo Agocs <laszlo.agocs@qt.io>
|
|
|
|
|
|
| |
Task-number: QTBUG-83707
Change-Id: I63548f4ace70af614a2aa082663bb3ae9fbedc25
Reviewed-by: Paul Olav Tvete <paul.tvete@qt.io>
|
|
|
|
|
|
|
|
|
|
|
| |
Internally this is already supported by all backends. The frontend was just
not exposing addressW, instead defaulting to the (arbitrarily chosen) ClampToEdge.
Add the parameter to newSampler(), but make it optional, defaulting to the more
natural Repeat (because that's what one would get with OpenGL for WRAP_R by default)
Change-Id: I0b991d8b649db37d4da86ac8e98ab7845601cf67
Reviewed-by: Christian Strømme <christian.stromme@qt.io>
|
|
|
|
|
|
|
|
|
| |
Also improve (docs and runtime checks) and test the minimum set
of required data to create a graphics pipeline.
Task-number: QTBUG-78971
Change-Id: If5c14f1ab1ff3cf70f168fde585f05fc9d28ec91
Reviewed-by: Paul Olav Tvete <paul.tvete@qt.io>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This also marks the beginnings of significantly extending autotesting
of the resource and rendering functionality in QRhi.
Also involves fixing up the buffer operation lists like we did
for textures before. This is to ensure updates and reads on the
same batch execute in the correct order. So just have two lists:
one with buffer, one with texture operations.
Also simplify the struct layouts. No need for those inner structs
with many duplicate members. This reduces the size even, since using a
union was never an option here. Also switch to a VLA, the size is around
253 KB per batch.
The Null backend now keeps track of the QRhiBuffer data so it can return
valid results in readbacks.
Task-number: QTBUG-78984
Task-number: QTBUG-78986
Task-number: QTBUG-78971
Task-number: QTBUG-78883
Change-Id: I9694bd7fec523a4e71cf8a5c77c828123ebbb3bd
Reviewed-by: Paul Olav Tvete <paul.tvete@qt.io>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
As usual, keep some QVector overloads around to allow Qt Quick to compile.
Color attachments and vertex input bindings get an at(index) type of
accessor, unlike any other of similar lists. This is because there the
index is significant, and sequential iteration is not the only type of
operation that is performed. Sometimes a lookup based on an index will
be needed as well.
Task-number: QTBUG-78883
Change-Id: I3882941f09e94ee2f179e0e9b8161551f0d5dae7
Reviewed-by: Christian Strømme <christian.stromme@qt.io>
Reviewed-by: Paul Olav Tvete <paul.tvete@qt.io>
|
|
|
|
|
|
| |
Task-number: QTBUG-78862
Change-Id: If278bd55530081cbbdbab8dd6e14d86e28da558e
Reviewed-by: Christian Strømme <christian.stromme@qt.io>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Sad to see this go since the d pointer pattern with implicit sharing
would have been perfect for this class, had this been a public API.
However, as binary compatibility will not be a concern for QRhi classes,
it is wasteful to allocate memory on every QRhiShaderResourceBinding.
This allows users, such as Qt Quick, to use QRhiShaderResourceBinding as
a cheap, simple, value class, without having to invent their own
alternatives in performance critical places.
The change brings a not insignficant improvement in certain qmlbench scenes
(the ones with thousands of unbatched geometry nodes).
Change-Id: I6d1dced6498d9ad625f90ead78bc0a417ea99ed8
Reviewed-by: Christian Strømme <christian.stromme@qt.io>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
With OpenGL a scissor (or viewport) rectangle is not allowed to have a negative
width or height. Everything else is allowed. This is also the semantic we wish to
keep for QRhiViewport and QRhiScissor.
This raises some problems. For instance, when we do bottom-left - top-left
rectangle conversion, the case of partially out of bounds rects needs to be
taken into account.
Otherwise, Qt Quick ends up in wrong scissoring in certain cases, typically when
the QQuickWindow size is decreased so the content does not fit because that will
then start generating negative x, y scissors for clipping (which is perfectly
valid but the QRhi backends need to be able to deal with it)
Then there is the problem of having to clamp width and height carefully, because
some validation layers for some APIs will reject a viewport or scissor with
partially out of bounds rectangles.
To verify all this, add a new manual test, based on the cubemap one. (cubemap was
chosen because that is an ideal test scene as it fills the viewport completely, and
so it is visually straightforward when a scissor rectangle is moving around over it)
Fixes: QTBUG-78702
Change-Id: I60614836432ea9934fc0dbd0ac7e88931f476542
Reviewed-by: Christian Strømme <christian.stromme@qt.io>
|
|
|
|
|
|
|
|
| |
...and change the return value of makeThreadLocalNativeContextCurrent() to
a bool since we expect this to mirror QOpenGLContext::makeCurrent().
Change-Id: I339507152e461fe28fcf7fe777165e6d0072f055
Reviewed-by: Christian Strømme <christian.stromme@qt.io>
|
|
|
|
|
|
|
| |
Starting with D3D11. The other backends will follow later.
Change-Id: I4f165c9f1743df0fb00bdce1e898917575bf5f6e
Reviewed-by: Christian Strømme <christian.stromme@qt.io>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Now that Qt Quick's batch renderer misses one level of shader source
caching due to the nature of pipeline state objects, it can be useful
to keep and reuse shader objects when the hash of the source code
matches.
The goal here is to allow Qt Quick to be on par with what the direct
OpenGL path has when it comes to caching shader sources and compilation
results. The program binary disk cache is not in scope in this patch.
Also adds QRhi::releaseCachedResources(), similarly to what the scenegraph
has. This can be called to clear caches such as the shader object
cache we keep here.
Change-Id: Ie3d81d823f61fa65ec814439e882c498f7774d43
Reviewed-by: Christian Strømme <christian.stromme@qt.io>
|
|
|
|
|
|
|
|
|
|
|
| |
As an option. Must opt in via setting ExternalContentsInPass in
the flags for beginFrame(). It is somewhat unfortunate to require
declaring this up front, but forcing using secondary command buffers
always, even though beginExternal() may not be used in many applications,
would be an overkill.
Change-Id: I8d52bcab40c96f89f140c4c7877b6c459925e3c7
Reviewed-by: Andy Nichols <andy.nichols@qt.io>
|
|
|
|
|
|
|
|
|
|
|
| |
Following the Quick scenegraph (qt.scenegraph.*), we now have
qt.rhi.general. Other categories may get added later.
This does not change the printing of real errors, those will
continue to use qWarning().
Change-Id: Id95416fc82ba8add9527212e431bcbd47d416f1a
Reviewed-by: Andy Nichols <andy.nichols@qt.io>
|
|
|
|
|
|
|
| |
...and storage buffers and images.
Change-Id: If38a51322e3187088a13cf4e9b88cb40c8af8621
Reviewed-by: Andy Nichols <andy.nichols@qt.io>
|
|
|
|
|
|
|
|
|
|
|
|
| |
Define the static QAtomic at file scope to avoid GCC's pessimisation with
function-static QAtomic (https://gcc.gnu.org/bugzilla/show_bug.cgi?id=79561),
and make sure the initial value is 0, so it ends up in BSS, not TEXT.
In QRhi..., don't create a static instance of the wrapper class, use a file-
static atomic, too. This turns the class into a glorified namespace.
Change-Id: I707f628e2b434330028077223071716d5704ba32
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
D3D11 and GL (4.3+, ES 3.1+) will come separately at a
later time.
Change-Id: If30f2f3d062fa27e57e9912674669225b82a7b93
Reviewed-by: Lars Knoll <lars.knoll@qt.io>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Needed by Qt Quick to handle cases where the application (or other Qt)
code contains OpenGL calls, and Qt Quick facilitates this by ensuring
the scenegraph's GL context is current. The expectation is that when
running with the GL backend of the rhi, all such code remains fully
functional. So add a makeCurrent type of call into the QRhi API that is
a no-op with anything other than OpenGL.
Change-Id: I6f774bf828e31802bdab0c3fef9421cdc0cebe5c
Reviewed-by: Lars Knoll <lars.knoll@qt.io>
|
|
Comes with backends for Vulkan, Metal, Direct3D 11.1, and OpenGL (ES).
All APIs are private for now.
Shader conditioning (i.e. generating a QRhiShader in memory or on disk
from some shader source code) is done via the tools and APIs provided
by qt-labs/qtshadertools.
The OpenGL support follows the cross-platform tradition of requiring
ES 2.0 only, while optionally using some (ES) 3.x features. It can
operate in core profile contexts as well.
Task-number: QTBUG-70287
Change-Id: I246f2e36d562e404012c05db2aa72487108aa7cc
Reviewed-by: Lars Knoll <lars.knoll@qt.io>
|