Currently it's not clear how to convert timestamps to nanoseconds on
Metal, disable timestamp query extension on Metal until we know how to
do that. WebGPU issue: https://github.com/gpuweb/gpuweb/issues/1325
Bug: dawn:434
Change-Id: If63bd5fe9dccc1fff16a42cddf63e9d8265e65c5
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/37020
Commit-Queue: Austin Eng <enga@chromium.org>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Austin Eng <enga@chromium.org>
- Get timestamp period on each backend
D3D12: Get GPU frequency(HZ) from queue and calculate the period in ns
Vulkan: Get timestampPeriod from device properties
Metal and others: don't need the period
Bug: dawn:434
Change-Id: Ia5588a3dccadfe92d7384b9fdf1e6848c6e5c6e2
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/36220
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Commit-Queue: Hao Li <hao.x.li@intel.com>
Adds some maxAnisotropy implementation.
Adds an end2end test, drawing a slanted plane with a texture of which each mipmap has a different color, with different maxAnisotropy values.
You can get an idea of what it does at https://jsfiddle.net/t64kpu81/85/
Needs further CTS.
Bug: dawn:568
Change-Id: I89ac56d8cf0fbb655358bf6effa016ddc1f8426f
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/35143
Commit-Queue: Kai Ninomiya <kainino@chromium.org>
Reviewed-by: Kai Ninomiya <kainino@chromium.org>
Fixes validation errors when drawing with point topologies, and without explicitly writing to the PointSize builtin.
Fixed: tint:321
Change-Id: I3c00c5ee56966a82d9e3024cb277eae8921a9af2
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/35800
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Austin Eng <enga@chromium.org>
Commit-Queue: Austin Eng <enga@chromium.org>
Changes the internal BindingInfo structure and any references to it. The
BindGroupLayoutEntry information is normalized when converting it into
the internal representation, but still accepted as either the old or
new layout. A "bindingType" member is added to the BindingInfo that's
not present in the BindGroupLayoutEntry itself to indicate which of
buffer, sampler, texture, or storageTexture is populated. This proves
useful for a myriad of switch statements in the various backends.
Bug: dawn:527
Change-Id: I6ae65adae61d0005fc50ed6d1bc2ec9b2a1295ad
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/35862
Commit-Queue: Austin Eng <enga@chromium.org>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Austin Eng <enga@chromium.org>
Auto-Submit: Brandon Jones <bajones@chromium.org>
Updates BindGroupLayoutEntry to allow for the newly split-up descriptors
that define each binding type in it's own member (buffer, texture, etc.)
The previous style of descriptor is still supported but is deprecated.
For the sake of keeping the scope reasonable, this change does not alter
the BindingInfo structure that's used internally by the various
backends. That will come as a followup.
Bug: dawn:527
Change-Id: I2f301f5f36fa2ce7ff15126ac90dc4c19d5e32ca
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/34921
Commit-Queue: Brandon Jones <bajones@chromium.org>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Turns on Tint generation of MSL if UseTintGenerator is on
Bug: dawn:571
Change-Id: Icfa523c36a509baf5da3b2a54152a7fb462c86f4
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/32303
Commit-Queue: Austin Eng <enga@chromium.org>
Reviewed-by: dan sinclair <dsinclair@chromium.org>
This factors code to move parsing of tint::ast::Module to the
frontend. All backends will use this code path when
UseTintGenerator is enabled for both SPIR-V and WGSL ingestion.
To avoid too much code explosion, parsing and validating the
shader is moved into ValidateShaderModuleDescriptor which
returns a result struct that gets passed into creation.
Bug: dawn:571
Change-Id: I598693ef36954fd0056a0744a2a0ebd7cc7d40a4
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/32301
Commit-Queue: Austin Eng <enga@chromium.org>
Reviewed-by: dan sinclair <dsinclair@chromium.org>
These were accidentally left out in the original code and fixes
WebGPU CTS zero init tests with renderable formats on Mac.
Bug: dawn:145
Change-Id: Ied886098408a749c094addb2ceaeb7e8c14938d7
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/34726
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Stephen White <senorblanco@chromium.org>
Commit-Queue: Austin Eng <enga@chromium.org>
Renames indexTypeSize to indexFormatSize. I forgot to add the changes
prior to pushing the commit.
Bug: dawn:502
Change-Id: I38f19edc021141c5a5ab57c6220978b7c88b116e
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/33923
Reviewed-by: Austin Eng <enga@chromium.org>
Reviewed-by: Stephen White <senorblanco@chromium.org>
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
This matches the defaults of webgpu.h and allows its use from inside
dawn_native for the implementations of SwapChain.
Bug: dawn:269
Change-Id: Icfe15f1775e78690fe4fb7b871bd0c0365d444e4
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/33783
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Stephen White <senorblanco@chromium.org>
This also simplifies a bunch of code in backends that was used to handle
getting the indexFormat from the pipeline "late".
Bug: dawn:502
Change-Id: Ibae50c8df21323fd391515f6036552e9fb868d93
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/32023
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Austin Eng <enga@chromium.org>
Reviewed-by: Stephen White <senorblanco@chromium.org>
There was a data-race between getting and using the pointer to
mLastSubmittedCommands in WaitForCommandsToBeScheduled and setting it
inside the scheduling handler.
Lock the mutex before calling waitUntilScheduled so that the object
doesn't get destroyed from underneath us.
Bug: chromium:1142024
Change-Id: Iadbf473530342de6d108d39285d723815452ac7c
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/31260
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Austin Eng <enga@chromium.org>
This helps push for comparing against nullptr more consistently.
Also replaces .Get() == nullptr and .Get() != nullptr with just ==
nullptr and != nullptr.
Bug: dawn:89
Change-Id: I884a4819f97305a73c11bad84391d1d2113ab7e2
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/32922
Reviewed-by: Austin Eng <enga@chromium.org>
Reviewed-by: Stephen White <senorblanco@chromium.org>
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
This makes refcounting of these objects more automatic to try and
prevent leaks or use-after-frees in the future.
Also removes operator* from RefBase (and Ref) because it is never used
and cannot work in a normal way for ObjectiveC protocols that cannot be
dereferenced.
Bug: dawn:89
Change-Id: I2e3fbfd638e2ba76d8c563f30bc489a384152552
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/32161
Reviewed-by: Austin Eng <enga@chromium.org>
Reviewed-by: Stephen White <senorblanco@chromium.org>
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
This helper function makes the code easier to read because the name
encodes the semantic of the operation compared to the bit-twiddling that
it replaces.
Bug: None
Change-Id: Iab587e04a91cf60acf8920de1f20bb55f3ea3816
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/31668
Reviewed-by: Stephen White <senorblanco@chromium.org>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
Previsouly having a ShaderModule with multiple entrypoints with the same
name and different stages was valid in Dawn. However it is disallowed by
the WGSL specification so change Dawn to index the ShaderModule's
entrypoints only by their name (instead of name and stage).
Bug: dawn:216
Change-Id: Id6fc80a03436b008c2f057bd30c70fdf240919e8
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/31665
Reviewed-by: dan sinclair <dsinclair@chromium.org>
Reviewed-by: Austin Eng <enga@chromium.org>
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
This CL adds sampling of depth-only and stencil-only texture
views on all backends. However, Metal on macOS <= 10.11 will
need a workaround to use separate depth/stencil textures for
each aspect since it is impossible to sample the stencil
aspect of a combined depth/stencil texture.
Also fixes sampling of depth24plus on D3D12 which had an
incomplete check for determining if a TYPELESS format is
necessary.
Bug: dawn:439, dawn:553
Change-Id: Id4991c565f822add200054296714e2dcd330119a
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/30725
Commit-Queue: Austin Eng <enga@chromium.org>
Reviewed-by: Stephen White <senorblanco@chromium.org>
This solves an issues where when switching swapchains the previous
one was destroyed before the new one was created, doing so detached
itself from the Surface, which in turn made the new swapchain not do
a graceful transition via vkSwapchainCreateInfoKHR::oldSwapchain.
Keeping the reference on the surface makes sure we always have
knowledge of the previous swapchain when replacing it. It requires
re-working the lifetime model of NewSwapChainBase to not require a
call to DetachFromSurface in the destructor, and having the Device
explicitly tell a swapchain it got attached on creation (otherwise
there are ASSERTs firing when swapchain creation fails).
In addition, backends are changed to use a SwapChain::Create method
and fail with a validation error (for now) when the previous swapchain
didn't use the same API.
vulkan::SwapChain is updated to use the previous swapchain's device's
fenced deleter to destroy it which is important in the device
switching tests.
The SwapChainValidationTests are updated because with the lifetime
changes the texture view can be kept alive after the application has
lost the last reference to the wgpu::SwapChain.
TBRing since it was reviewed in a different CL (but for the wrong
branch).
TBR=enga@chromium.orgTBR=senorblanco@chromium.org
Bug: dawn:269
Change-Id: Ie4374b5685af990d68969ab9cd7767e53c287ace
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/31041
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
Previously Format::Type was used instead of wgpu::TextureComponentType
so that ::Other could server as a tag value. This was confusing and
almost the single use of ::Other.
In a follow-up CL the format baseType is changed to be a per-aspect
value, and Format::Type. This CL is a self-contained step in that
direction.
Bug: dawn:527
Change-Id: Ida834087f45a8fca17670ffe8ebd4d5c4f1cd423
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/30102
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Stephen White <senorblanco@chromium.org>
Reviewed-by: Austin Eng <enga@chromium.org>
In follow up CLs additional will be added to the AspectInfo, like the
supported component types.
Also simplify the logic for GetTexelInfo since all aspects are the first
aspects, except stencil which is always stencil8.
Bug: dawn:517
Change-Id: Iebbcb8a7f8fa2c4b7b06f65d6e4e8917c0a85366
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/30100
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Austin Eng <enga@chromium.org>
And use it to print the driver version at the start of
dawn_end2end_tests. This will help when figuring out issues
happening on CQ but not necessarily locally.
Bug: None
Change-Id: Ibdb9ab8cab53cc1e1cf8a807da53edeca616bed9
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/29602
Reviewed-by: Stephen White <senorblanco@chromium.org>
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
This allows empty scissors, so add a test for it.
This disallows scissor boxes that are bigger than the renderpass
attachment so remove an end2end test for that behavior.
Update the SetScissorRect validation tests.
Bug: dawn:542
Change-Id: I5b8578a4df1b94510a9356bd4007efddf2711588
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/29820
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Stephen White <senorblanco@chromium.org>
- Add WriteTimestamp and ResolveQuerySet on Metal
- Enable end2end tests of timestamp query on Metal
- Lazy initialize the distination buffer in ResolveQuerySet
- Update part of end2end tests to test from render pass instead of
command encoder due to we cannot write timestamp on CommandEncoder
without any copy commands on Metal.
Bug: dawn:434
Change-Id: Ie9217e1f5a00bc252d6293ef7521c2e343ba9259
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/29020
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Jiawei Shao <jiawei.shao@intel.com>
Commit-Queue: Hao Li <hao.x.li@intel.com>
Adds validation to ensure clear colors do not exceed 2^24 and a
corresponding unit test. Also removes intermediate float conversions
that are no longer necessary.
Bug: dawn:525
Change-Id: I020b98de85384c20da51158de79eab87f60dcf6d
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/29040
Commit-Queue: Brandon Jones <brandon1.jones@intel.com>
Reviewed-by: Austin Eng <enga@chromium.org>
This will prevent mixing it up with other serial types in the future.
Bug: dawn:442
Change-Id: I74e964708acc62eb0f33127cc48f1b9a7b171d11
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/28923
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Jiawei Shao <jiawei.shao@intel.com>
Reviewed-by: Austin Eng <enga@chromium.org>
A lot of our switches over enum values use the following pattern:
default:
UNREACHABLE();
return foo;
This is problematic because when adding a new value to one of the WebGPU
enums, there is no compilation error for switches that are missing it.
Currently we're supposed to write code and tests and fix UNREACHABLEs when
we see them.
Instead we should strive to have most switches on enums to be complete
and explicitily tag unreachable values as UNREACHABLE. Some switches
might still want to use default: UNREACHABLE() if only a couple values
need to be handled out of very many.
In this CL we go through all the UNRAECHABLEs and change them if need
be. Also an ErrorQueue class is added to avoid having
QueueBase::SubmitImpl just be UNREACHABLE (and force overriding
instead).
Bug: dawn:527
Change-Id: I33dfb4703104912cc5f001f9faf907a61324de68
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/28501
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Jiawei Shao <jiawei.shao@intel.com>
Returning the layouts from an export operation and then using
them in a subsequent import operation allows the import to preserve
the texture contents.
This fixes Vukan image wrapping on some AMD/NVIDIA devices.
Bug: dawn:200
Change-Id: Icbb6e759856d410bb69724b9f439bc3088756d19
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/28380
Commit-Queue: Austin Eng <enga@chromium.org>
Reviewed-by: Kai Ninomiya <kainino@chromium.org>
Change dawn.json to reflect changes in the WebGPU IDL. Also fixes any
conversion issues in Dawn.
Bug: dawn:525
Change-Id: Ifb46329f073bcf31d43342f20f4819eac061d5a1
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/28400
Reviewed-by: Austin Eng <enga@chromium.org>
Commit-Queue: Brandon Jones <brandon1.jones@intel.com>
It isn't in the upstream WebGPU spec and completely unimplemented in
Dawn.
Bug: dawn:527
Change-Id: I2023c7b1de2a9fa50d26ab1678b7ef7e32c64af6
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/28500
Reviewed-by: Austin Eng <enga@chromium.org>
Reviewed-by: Jiawei Shao <jiawei.shao@intel.com>
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
This change the Metal backend in preparation for supporting multiple
entrypoints:
- Explicitly set the spirv_cross entry point before compiling.
- Moves gathering of the local size to the frontend as it will be
useful for validation in the future.
- Query spirv-cross for the modified entrypoint name instead of
duplicating the code in Dawn.
- Move some conversion helpers from ShaderModule.cpp to their own
SpirvUtils file.
Bug: dawn:216
Change-Id: I87d4953428e0bfeb97e39ed22f94d86ae7987782
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/28241
Commit-Queue: Kai Ninomiya <kainino@chromium.org>
Reviewed-by: Kai Ninomiya <kainino@chromium.org>
Also moves BindingNumber, BindGroupIndex, and BindingIndex to
IntegerTypes.h. Future TypedIntegers should be declared here.
Bug: dawn:442
Change-Id: I5ba8de3412fb48b7957b67e7c413a5097f8ec00f
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/27880
Reviewed-by: Austin Eng <enga@chromium.org>
Commit-Queue: Austin Eng <enga@chromium.org>
Remove all usages of SPVC from the code and update the fuzzers. Some
of the include paths and deps came transitively from spvc, so needed
to update build rules.
This patch does NOT remove the flags related to spvc usage, they are
just no-ops as the moment. After this patch lands I will remove the
usage of those flags from the bots, then remove the flags.
BUG=dawn:521
Change-Id: I0d7c3e28f79354c78f00c48b6a383b823094a069
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/27900
Reviewed-by: Austin Eng <enga@chromium.org>
Commit-Queue: Ryan Harrison <rharrison@chromium.org>
PipelineBase now collects the EntryPointMetadata for all its
stages which makes the rest of the code agnostic to the entrypoint
name (except D3D12 and OpenGL that required transition hacks and
will be fixed in follow-up CLs).
Bug: dawn:216
Change-Id: I643da198cb2a20a9d94d805a2dc783d6d4346ae9
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/27260
Reviewed-by: Ryan Harrison <rharrison@chromium.org>
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
This patch removes the Toggle LazyClearBufferOnFirstUse and uses
the Toggle::LazyClearResourceOnFirstUse, which means buffer lazy
initialization has been enabled by default in Dawn.
BUG=dawn:414
TEST=dawn_end2end_tests
Change-Id: I6b247d9442b57b6bb3cb5a2208467036fef6b293
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/27820
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Austin Eng <enga@chromium.org>
Commit-Queue: Jiawei Shao <jiawei.shao@intel.com>
This patch fixes the issues on the initialization of buffers with
MapRead usage and mappedAtCreation == true.
1. The buffers with MapRead usage and mappedAtCreation == true can be
read from the CPU side just after the creation of the buffer,
however at that time the internal pending command buffers may not
be executed, thus causing the buffer is not cleared as is expected.
2. On D3D12 the buffer with MapRead and mappedAtCreation == true is
created on the READBACK heap, so all the data written in the CPU
side cannot be uploaded to the GPU memory. When the buffer is mapped
again all the original data written through the CPU pointer will be
overwritten by the data in the GPU memory (which means it is also
cleared to 0).
This patch fixes this issue by:
1. clearing the buffers with mappedAtCreation == true on the CPU side.
2. on D3D12 making the buffer with MapRead and mappedAtCreation == true
use the staging buffer instead of mapping itself.
Note that this change is only related to the code path with Toggle
"nonzero_clear_resources_on_creation_for_testing" enabled, currently
we don't plan to do the similar change when we enable Dawn wire.
BUG=dawn:414
TEST=dawn_end2end_tests
Change-Id: I2b3d0840333e8d99759800ab9fc141d0a7cf2f8d
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/27220
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Commit-Queue: Jiawei Shao <jiawei.shao@intel.com>
Hao Li
shrekshao
Ben Clayton
Brandon Jones
Austin Eng
Corentin Wallez
Kai Ninomiya
Enrico Galli
Brandon Jones
Ryan Harrison
Jiawei Shao