This CL implements RenderBundle and RenderBundleEncoder in the frontend
and adds unittests for validation.
Bug: dawn:154
Change-Id: Ice5ecd384cd627ad270b73052408f8139d1ea5f4
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/9221
Commit-Queue: Austin Eng <enga@chromium.org>
Reviewed-by: Kai Ninomiya <kainino@chromium.org>
As part of enabling WebGPU on Chrome OS, we would like to run the Dawn
unit and e2e tests as part of the Chrome OS test suite. This CL removes
the GLFW dependency because Chrome OS does not support GLFW.
The GLFWwindow is only used to create swap chains for the various
backends, but these swap chains are not actually used in the tests
(the e2e tests render to textures instead). The swap chains are only
referenced as part of an unused debugging function:
SwapBuffersForCapture which we can safely remove as per my discussions
with kainino@ and enga@.
We still need GLFW for OpenGL, so we conditionally include it on
platforms that enable the OpenGL backend (which Chrome OS is not).
Note: enga@ suggested to create a VulkanWindowlessBinding that has an
empty GetSwapChainImplementation, but after exploring the option, it
seems like a bit too many ifdefs. In the end, I think it's cleaner to
just remove the *Binding classes entirely.
BUG=chromium:993457
TEST=tests compile and pass for all values of dawn_enable_opengl
Change-Id: I067b12a23f2c236f5506252cd7727b847e79a667
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/10080
Reviewed-by: Kai Ninomiya <kainino@chromium.org>
Commit-Queue: Brian Ho <hob@chromium.org>
This patch provides Chromium TRACE_EVENT macros and hooks for
implementing the TRACE_EVENT api.
Bug: chromium:958013
Change-Id: I033b1c7ca57c550504a1bea1898a1a152831922b
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/7060
Commit-Queue: Kai Ninomiya <kainino@chromium.org>
Reviewed-by: Kai Ninomiya <kainino@chromium.org>
This test is flaky on Windows on Intel HD 530. Skip it on Windows for now.
BUG=dawn:55
Change-Id: I537ef66ee6b0844e8b1dfd46fae47c5921451e55
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/10040
Reviewed-by: Austin Eng <enga@chromium.org>
Commit-Queue: Austin Eng <enga@chromium.org>
Cannot reproduce failure on GTX1060 and GTX2080Ti with latest driver. So disable
inherit test on compute shader and reland this patch
Original change's description:
> Fix dynamic buffer offset issues on D3D12 backend
>
> This patch fixes
> 1. Inherit dynamic buffer offsets : dawn chooses vulkan
> like inherit behaviour, so dynamic offsets need to be inherited.
> This patch adds inherit dynamic offsets between pipelines support and
> adds tests to cover it.
>
> 2. Dynamic offsets are skipped when groups have been set : in D3D12
> backend, when invoke SetBindGroup, dawn will check whether this group
> has already been set and skip updating root signature if the answer is
> yes. However, this behaviour will affect dynamic offsets update. With
> the latest patch, we always update dynamic offsets, even if they didn't
> change and adds tests to cover it.
>
> This patch also hit a dawn's issue about storage buffer validation in compute pass.
> Currently the validation is a workaround to avoid access conflicts but will impact
> using dynamic buffer offset in compute pipeline. Fix this issue is hard so disable
> related test for now and will enable it after the issue been fixed. File dawn bug
> 198 to track this
>
> BUG=dawn:55
BUG=dawn:55
Change-Id: Ia105786c035eafc6f68dcb54e6c1145b06c6a630
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/9960
Reviewed-by: Austin Eng <enga@chromium.org>
Reviewed-by: Kai Ninomiya <kainino@chromium.org>
Commit-Queue: Shaobo Yan <shaobo.yan@intel.com>
Validation of GPURenderBundle will share code with RenderPass
validation. Factor validation of commands for GPURenderBundle
into a separate function.
Bug: dawn:154
Change-Id: I79a229592ead27d462da0dd2d12fbdb95443ff19
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/9980
Commit-Queue: Kai Ninomiya <kainino@chromium.org>
Reviewed-by: Kai Ninomiya <kainino@chromium.org>
This change adds platform-dependent services that handle creating
semaphores and importing image memory. Then, we use them to wrap a
texture from an outside source, and release a signal semaphore back
when we're done with it. This will be used to allow chrome to render
dawn on Vulkan platforms.
Bug: chromium:976495
Change-Id: I9f07eaf436e10aa6bd88cffdc74fd23834d62ee0
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/8340
Commit-Queue: Idan Raiter <idanr@google.com>
Reviewed-by: Kai Ninomiya <kainino@chromium.org>
Clear through loadop instead of standalone clear operation to optimize
efficiency on modern desktop GPUs and mobile GPUs.
Removed clear calls in TransitionForPass for render pass
to help optimize clearing using loadop instead.
Compute pass textures and sampled textures are still cleared in transition.
Bug: dawn:145
Change-Id: I84082bdea3ed7be75683389132d8b296051731b7
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/8641
Reviewed-by: Austin Eng <enga@chromium.org>
Reviewed-by: Kai Ninomiya <kainino@chromium.org>
Commit-Queue: Natasha Lee <natlee@microsoft.com>
This reverts commit 421584173c.
Reason for revert: Some of the newly added tests are failing on Windows NVIDIA GTX 1660
https://ci.chromium.org/p/chromium/builders/ci/Win10%20FYI%20x64%20Release%20%28NVIDIA%20GeForce%20GTX%201660%29/21
Original change's description:
> Fix dynamic buffer offset issues on D3D12 backend
>
> This patch fixes
> 1. Inherit dynamic buffer offsets : dawn chooses vulkan
> like inherit behaviour, so dynamic offsets need to be inherited.
> This patch adds inherit dynamic offsets between pipelines support and
> adds tests to cover it.
>
> 2. Dynamic offsets are skipped when groups have been set : in D3D12
> backend, when invoke SetBindGroup, dawn will check whether this group
> has already been set and skip updating root signature if the answer is
> yes. However, this behaviour will affect dynamic offsets update. With
> the latest patch, we always update dynamic offsets, even if they didn't
> change and adds tests to cover it.
>
> This patch also hit a dawn's issue about storage buffer validation in compute pass.
> Currently the validation is a workaround to avoid access conflicts but will impact
> using dynamic buffer offset in compute pipeline. Fix this issue is hard so disable
> related test for now and will enable it after the issue been fixed. File dawn bug
> 198 to track this
>
> BUG=dawn:55
>
> Change-Id: I2b0f179b3555d37d5b350292eb729767b0d60ab6
> Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/9621
> Commit-Queue: Austin Eng <enga@chromium.org>
> Reviewed-by: Austin Eng <enga@chromium.org>
> Reviewed-by: Kai Ninomiya <kainino@chromium.org>
TBR=kainino@chromium.org,shaobo.yan@intel.com,enga@chromium.org
# Not skipping CQ checks because original CL landed > 1 day ago.
Bug: dawn:55
Change-Id: I018e6fa9b4b0905de602ab4ba3294ef537c53759
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/9900
Reviewed-by: Austin Eng <enga@chromium.org>
Commit-Queue: Austin Eng <enga@chromium.org>
GPURenderBundleEncoder and GPUCommandEncoder will need to share code
for tracking resource usages.
Bug: dawn:154
Change-Id: I0286f71c4c0638f89be2754c8e9691e67e5db335
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/9700
Commit-Queue: Austin Eng <enga@chromium.org>
Reviewed-by: Kai Ninomiya <kainino@chromium.org>
Unlike API objects, the AttachmentState object is only used internally
and should have no external references. We should not increment the
reference count on creation because it is Ref'ed internally. This patch
also adds ASSERTs that all Device caches are empty when the Device is
destroyed.
Bug: dawn:154
Change-Id: I5dd46bde03c0be920356444e6b1214ed38e833e8
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/9761
Commit-Queue: Kai Ninomiya <kainino@chromium.org>
Reviewed-by: Kai Ninomiya <kainino@chromium.org>
This patch fixes
1. Inherit dynamic buffer offsets : dawn chooses vulkan
like inherit behaviour, so dynamic offsets need to be inherited.
This patch adds inherit dynamic offsets between pipelines support and
adds tests to cover it.
2. Dynamic offsets are skipped when groups have been set : in D3D12
backend, when invoke SetBindGroup, dawn will check whether this group
has already been set and skip updating root signature if the answer is
yes. However, this behaviour will affect dynamic offsets update. With
the latest patch, we always update dynamic offsets, even if they didn't
change and adds tests to cover it.
This patch also hit a dawn's issue about storage buffer validation in compute pass.
Currently the validation is a workaround to avoid access conflicts but will impact
using dynamic buffer offset in compute pipeline. Fix this issue is hard so disable
related test for now and will enable it after the issue been fixed. File dawn bug
198 to track this
BUG=dawn:55
Change-Id: I2b0f179b3555d37d5b350292eb729767b0d60ab6
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/9621
Commit-Queue: Austin Eng <enga@chromium.org>
Reviewed-by: Austin Eng <enga@chromium.org>
Reviewed-by: Kai Ninomiya <kainino@chromium.org>
This patch does some code clean-ups in CompressedTextureFormatTests as
the preparation of adding new tests in the following patches.
BUG=dawn:42
TEST=dawn_end2end_tests
Change-Id: Ie00eeca2d3a161f3673c554b91b54eebea4bf8bb
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/9740
Commit-Queue: Jiawei Shao <jiawei.shao@intel.com>
Reviewed-by: Kai Ninomiya <kainino@chromium.org>
This patch skips several flaky and failed tests on the related Dawn
backends.
1. Tests on all integer/unsigned integer formats in TextureFormatTest on
Intel OpenGL drivers.
2. All CompressedTextureBCFormatTest tests on Intel Vulkan Windows
drivers.
3. BufferSetSubDataTests/ManySetSubData on Intel Vulkan Linux drivers.
BUG=dawn:128, dawn:199, chromium:980737
Change-Id: Ia8912804ea762b458be0b48d9df341ff7637f608
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/9800
Commit-Queue: Kai Ninomiya <kainino@chromium.org>
Reviewed-by: Kai Ninomiya <kainino@chromium.org>
They aren't guaranteed renderable in Vulkan nor core OpenGL 4.5 without
extensions. They happened to work on a lot of our CQ builders but failed
on Intel OpenGL Linux which helped understand this was an issue.
BUG=dawn:128
Change-Id: I83c4f8116c1125c3bac2f1dd6197976c9063e129
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/9682
Commit-Queue: Austin Eng <enga@chromium.org>
Reviewed-by: Jiawei Shao <jiawei.shao@intel.com>
Reviewed-by: Austin Eng <enga@chromium.org>
Metal uses a physical addressing mode and buffers are just pointers in
GPU memory that don't have a size. This prevents implementing
OpArrayLength without any additional information. When a shader uses
OpArrayLength on unsized arrays, SPIRV-Cross requires an extra buffer
argument that gets the length of the buffers.
Use that workaround mechanism in the Metal backend by keeping track of
the storage buffer sizes and applying the extra "buffer length buffer"
when a pipeline requires it.
Also adds tests that OpArraySize works in all shader stages.
BUG=dawn:195
Change-Id: I6aa6089aaea85d0589ccad1756e55dd0befefcb5
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/9386
Commit-Queue: Kai Ninomiya <kainino@chromium.org>
Reviewed-by: Kai Ninomiya <kainino@chromium.org>
Also adds GetGlobalInfo, similar to GetDeviceInfo, but for code that wants to query instance extensions or API version.
Bug: chromium:976495
Change-Id: Ia7131aa1780737ec2691cde0007acb86a120d307
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/9620
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Kai Ninomiya <kainino@chromium.org>
Commit-Queue: Kai Ninomiya <kainino@chromium.org>
The only missing one is BGRA8UnormSrgb because it doesn't exist in
OpenGL and will need further emulation.
Also suppress most tests on the Mesa Intel GL driver that fails tests
pending further debugging.
BUG=dawn:128
Change-Id: Id454a89711e25b14c7de75a36c7642f2429a0e1f
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/8364
Reviewed-by: Austin Eng <enga@chromium.org>
Reviewed-by: Jiawei Shao <jiawei.shao@intel.com>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
This CL makes sure debug markers commands are handled not only for
render pass encoder, but also compute pass encoder.
Bug: dawn:44
Change-Id: Iad1865d7c20e36d580e1f3f16496e4b90cc1ff34
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/9680
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
This CL makes fragment stage optional but still raises a validation
error if set to null in a RenderPipeline.
Bug: dawn:136
Change-Id: Ie900d2e768d251db1e0c4a90738f80b2de15dc2f
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/9641
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Austin Eng <enga@chromium.org>
Commit-Queue: François Beaufort <beaufort.francois@gmail.com>
Adds a structure that contains wait / signal semaphores and the current command buffer. Will allow us to have a list of pending semaphores after recording, either to consume or do something smarter with in the future.
Bug: chromium:976495
Change-Id: Ib61455039bd97ac8b0ff701af2b694cc8794226d
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/9600
Commit-Queue: Kai Ninomiya <kainino@chromium.org>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Kai Ninomiya <kainino@chromium.org>
This patch refactors the current implementation of BC formats to treat
it as the first extension in Dawn and adds all the related tests.
Note that in Dawn all the extensions are disabled unless we enable them
when we create the device, which means the BC formats can only be used
when we enable the related extension on the creation of the device, and
the creation of the device will fail if the adapter does not support the
extension
BUG=dawn:42
TEST=dawn_end2end_tests
TEST=dawn_unittests
Change-Id: I04d818b0218ebb3b1b7a70a4fea71779f308f85f
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/9520
Commit-Queue: Jiawei Shao <jiawei.shao@intel.com>
Reviewed-by: Austin Eng <enga@chromium.org>
In a typical graphics application it is a common usage to update some uniforms once per draw,
and such uniforms include the word positions, orientations, and so on. In the current state of
WebGPU, this means that for each draw call we have to create a new bind group to set the right
uniform values. Bind group creation is expected to be more expensive than
recording draws because a memory allocation is required.
The functionality of dynamic buffer offset is to reduce the number of bind groups that need to
be created.
The patch implements dynamic buffer offset on D3D backend using root descriptor.
Bug=dawn:55
Change-Id: Ia713a4edb3c0ab8f3bba048d7813f343e9dee166
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/9040
Commit-Queue: Shaobo Yan <shaobo.yan@intel.com>
Reviewed-by: Austin Eng <enga@chromium.org>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
This patch moves all the code related to Toggle from Instance.cpp to
Toggles.cpp to make the code in Instance.cpp cleaner.
BUG=dawn:56
Change-Id: Ica2a779569478a58ddf21e632a211c0c5b889b00
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/9580
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Austin Eng <enga@chromium.org>
Commit-Queue: Jiawei Shao <jiawei.shao@intel.com>
Introduces the buddy system for fast power-of-two sub-allocation.
BUG=dawn:27
Change-Id: I56836ae317ecc5a91d8341c843fc37d4f91fb5af
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/9260
Reviewed-by: Kai Ninomiya <kainino@chromium.org>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
bindgroup->GetLayout() was called before checking the bindgroup is an
error, causing an assert to fire in Debug. Adds a regression unittest.
BUG=dawn:196
Change-Id: I58e95879a7f2a7cf5c47c7b228f7e3b5bb72d8ea
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/9560
Commit-Queue: Kai Ninomiya <kainino@chromium.org>
Reviewed-by: Kai Ninomiya <kainino@chromium.org>
Reviewed-by: Austin Eng <enga@chromium.org>
In a typical graphics application it is a common usage to update some uniforms once per draw,
and such uniforms include the word positions, orientations, and so on. In the current state of
WebGPU, this means that for each draw call we have to create a new bind group to set the right
uniform values. Bind group creation is expected to be more expensive than
recording draws because a memory allocation is required.
The functionality of dynamic buffer offset is to reduce the number of bind groups that need to
be created.
The patch implements dynamic buffer offset on OpenGL backend using glBindBufferRange and adds
validation to check whether visibility of resources are none.
Bug=dawn:55
Change-Id: I77e10a9677d1737f377301ee89e29d904c91c298
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/9540
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Austin Eng <enga@chromium.org>
Commit-Queue: Shaobo Yan <shaobo.yan@intel.com>
WebGPU have a 2D pipeline layout plus a vertex buffer table while Metal
has a single vertex buffer table that contains everything (including
uniform and storage buffers). Previously the space for vertex buffers
was statically allocated in that table which made the last vertex buffer
go out of bound of the Metal vertex buffer table.
This fixes the issue by packing all the vertex buffers that are used
right after the vertex buffers used by the bind groups. This is a
drive-by fix found while looking at reserving Metal vertex buffer 30 to
contain the shader storage buffer lengths.
BUG=dawn:195
Change-Id: If5c67bbc0d15c976793ef43889e50e4a360217d7
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/9387
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Austin Eng <enga@chromium.org>
This both deduplicates shared state by multiple passes or pipelines and
makes checking pipeline compatibility a single pointer check. It will be
useful for also checking RenderBundle compatibility.
Bug: dawn:154
Change-Id: I0fb289fab5ac76a7fbd500f64b8a6409a246ab32
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/9461
Commit-Queue: Austin Eng <enga@chromium.org>
Reviewed-by: Kai Ninomiya <kainino@chromium.org>
This patch adds the support of recreating a swapchain with SwapChain.
Configure() on Vulkan backends.
1. Query the Vulkan surface information again.
2. Clean up old swapchain.
BUG=dawn:177
Change-Id: Id7e9e07e9dbbba4a71322367cae10fb179106220
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/9340
Reviewed-by: Austin Eng <enga@chromium.org>
Commit-Queue: Jiawei Shao <jiawei.shao@intel.com>
Add test for many buffer writes.
BUG=dawn:27
Change-Id: I026c3b9a6d30c7c6634b89db9798902b036150c3
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/9280
Commit-Queue: Bryan Bernhart <bryan.bernhart@intel.com>
Reviewed-by: Austin Eng <enga@chromium.org>
This patch factors the CommandAllocator, CommandIterator, and error
handling out of CommandEncoderBase so it can later be used by the
RenderBundleEncoder.
Bug: dawn:154
Change-Id: Ia4f8c3ce7f432f0887b619bd8090aa9bec7330fc
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/9181
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Kai Ninomiya <kainino@chromium.org>
Commit-Queue: Austin Eng <enga@chromium.org>
The information about formats in OpenGL will grow to include more
information than just (internalFormat, format, type) and will gain more
logic to depend on GL version and available extensions over time.
BUG=dawn:128
Change-Id: I63a6ac7d48797fb4a9f97a65871306e640cf41d6
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/9201
Reviewed-by: Austin Eng <enga@chromium.org>
Reviewed-by: Jiawei Shao <jiawei.shao@intel.com>
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
This patch adds the missing header <tuple> in OpenGLFunctions.cpp to
fix a build error on Windows that the compiler cannot find the
definition of std::tie().
BUG=dawn:190
Change-Id: I35d1f258d3b6197187d626870f3347e195881da8
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/9300
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Austin Eng <enga@chromium.org>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
This uses the OpenGL debug ouput functionality to make the driver call
us back when an error happens so we can ASSERT and fail.
BUG=dawn:190
Change-Id: I4b6d7a860384dfeccc1c37383fd4cbdc09d7dc05
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/9204
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Kai Ninomiya <kainino@chromium.org>
We would incorrectly use the size of the first mip level to clear
non-zero mip levels. This was found while enabling debug output in
the OpenGL backend.
BUG=dawn:190
Change-Id: I29a87b87057d425d1c131d11afdc7ca4716607c8
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/9202
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Kai Ninomiya <kainino@chromium.org>
In OpenGL we have to query a resource's location or index before we can
bind it. The driver can tell us that the resource doesn't exist using
special values (-1 for locations, GL_INVALID_INDEX for indices). This
happens when Dawn compiles a pipeline that has bind group bindings that
none of the shader modules uses.
This was found while enable OpenGL debug output.
BUG=dawn:190
Change-Id: I58b3a1a65f87b4408985c1258f1a95379e6b540e
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/9203
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Austin Eng <enga@chromium.org>
Reviewed-by: Kai Ninomiya <kainino@chromium.org>
Austin Eng
Brian Ho
Yan, Shaobo
Bryan Bernhart
Idan Raiter
Natasha Lee
Jiawei Shao
Corentin Wallez
François Beaufort
Stephen White