When we do transition barriers for a texture view outside of a pass
(say copy, clear, initialization), if the texture view can cover all
subresources, and its old usages across all subresources are the
same, then we can use one transition barrier. We don't need to use
separate barrier per each subresource.
This patch can reduce barrier we delivered, and improve performance
for particular situations.
Bug: dawn:441
Change-Id: I2ae9b39793915553cbaaceacaf58bf87c9ba3bc6
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/23129
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Austin Eng <enga@chromium.org>
Commit-Queue: Yunchao He <yunchao.he@intel.com>
This patch put subresource related variables like baseMipLevel,
levelCount, baseArrayLayer, layerCount into a single struct at
front-end. We have a lot more at backend too, a following patch
will do that.
Bug: dawn:157
Change-Id: Iab5633a4246b6ae89b80c39f5672dbb31d7a3e78
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/22704
Reviewed-by: Kai Ninomiya <kainino@chromium.org>
Commit-Queue: Yunchao He <yunchao.he@intel.com>
If a texture follow this rule:
* old usages of all subresources are the same
* new usages of all subresources are the same
Then we can use one barrier to transit states for all subresources.
We don't need to use one barrier per each subresource.
This patch adds a flag at front-end to capture many (but not all)
situations which follow the rule above, then we don't need to loop
over every subresource at Vulkan backend.
Bug: dawn:441
Change-Id: I6124ee2cb09c9142fefd8f057dc6d2659301e2d4
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/22702
Commit-Queue: Yunchao He <yunchao.he@intel.com>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
This patch combines barriers in a render/compute pass into one
function call.
Previously, we need to dispatch barrier(s) for each buffer/texture
in a pass. So we may need quite a lot function calls to deliver
barriers in a pass in real web applications. One example is that
we did see that too many function calls to deliver barriers in
Aquarium (WebGPU porting) contributed to CPU usage and bottleneck.
Bug: dawn:441
Change-Id: Ibe44967fefd2e1e6e64df4587146c4fb7fbe8e73
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/22700
Commit-Queue: Yunchao He <yunchao.he@intel.com>
Reviewed-by: Austin Eng <enga@chromium.org>
This change implemented texture subresource on Vulkan. It added a new
function to handle barriers for texture subresource for bind groups.
It also simplified barriers which are set for texture clear and copy.
Before this patch, all barriers are done upon all mip levels and all
array layers. With this patch, barriers are done upon particular mip
level(s) and array layer(s).
We may need more texture subresource end2end tests for copy and clear
opterations. I will visit that later.
Bug: dawn:157
Change-Id: Ie2247c6315326494f2d3736334e84b2867a16c17
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/22024
Commit-Queue: Yunchao He <yunchao.he@intel.com>
Reviewed-by: Austin Eng <enga@chromium.org>
Configure was transitioning the swapchain images from undefined to
present layout but this was already happening because TextureVk starts
with a mLastUsage of None that will force a transition from undefined
when used.
Also introduce an internal texture usage bit kPresentTextureUsage to
prepare for the eventual remove of wgpu::TextureUsage::Present when
old swapchains are removed.
Bug: dawn:269
Change-Id: I57d26f18e34cacd5d91419a45787b2ece9558846
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/20881
Reviewed-by: Stephen White <senorblanco@chromium.org>
Reviewed-by: Austin Eng <enga@chromium.org>
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
This aligns this constructor to be used via Create like the other
vulkan::Texture constructors.
Bug: dawn:269
Change-Id: Ib85874bf24bfe49b644b4faa63c9248c540811c8
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/20882
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Austin Eng <enga@chromium.org>
Reviewed-by: Stephen White <senorblanco@chromium.org>
This patch adds the basic supports of read-only and write-only storage
textures on Vulkan.
The subresource tracking and barriers on the subresources used as
read-only and write-only storage textures are not included in this
patch.
BUG=dawn:267
TEST=dawn_end2end_tests
Change-Id: I6831b96202a97182763ecd28bc41ab03df904a7c
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/20560
Commit-Queue: Jiawei Shao <jiawei.shao@intel.com>
Reviewed-by: Austin Eng <enga@chromium.org>
To avoid accidental memory leaks on account of using raw pointers,
use Ref<TextureBase> as method return type except at Dawn interface
boundaries.
Change-Id: I6459062ee28984de2cb1d5a2059bc70cf82b2faf
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/19580
Reviewed-by: Austin Eng <enga@chromium.org>
Commit-Queue: Rafael Cintron <rafael.cintron@microsoft.com>
RefCounted (and derived) destructors should be protected on the class
to ensure the objects can ONLY be destructed by calling Release. This
avoids errors cause by destroying objects out from under code which
has an active reference count.
Unfortunately, many of the 'base' classes must continue having public
destructors because they are used as "blueprint" objects created on
the stack.
Added final on most-derived classes.
Ideas for future improvement:
- Change "base" objects to have protected destructors but create new
blueprint objects that privately derive from base objects. This
limits the blueprint object's usefulness to only be a blueprint.
- Modify createX methods to return Ref<Object> instead of Object*
Change-Id: I6f3b3b178118d135c4342cb912e982a3873d71af
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/18780
Commit-Queue: Rafael Cintron <rafael.cintron@microsoft.com>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Overriding VK_DEFINE_NON_DISPATCHABLE_HANDLE changes the function
signatures of Vulkan functions, changing their ABI and making us
incompatible with real drivers. This removes that magic, and replaces it
with an explicit wrapper, VkHandle, which has much of the same
functionality as the original VkNonDispatchableHandle.
It adds definitions for dawn_native::vulkan::VkBuffer et al, which
shadow the native ::VkBuffer et al. This retains type safety throughout
the Vulkan backend without changing every single usage.
Notably, the following things had to change:
- An explicit conversion from VkBuffer* to ::VkBuffer* is needed for
arrays. This is implemented as a reinterpret_cast, which is still
safe as the new VkHandle still has the same memory layout properties
as VkNonDispatchableHandle did.
- When pointing to a VkHandle as an output pointer, it's now necessary
to explicitly get the native ::VkBuffer (via operator*) and point to
it.
Previously reviewed on:
https://dawn-review.googlesource.com/c/dawn/+/15580
Bug: chromium:1046362
Change-Id: I7d34ec38a805025f92165ea9a7ee07ae5c182076
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/15641
Reviewed-by: Kai Ninomiya <kainino@chromium.org>
Commit-Queue: Kai Ninomiya <kainino@chromium.org>
This reverts commit 4e17d5c2483b63d4863162d692a1a961d1dcb958.
Reason for revert: broken on chromeos
Original change's description:
> Remove VK_DEFINE_NON_DISPATCHABLE_HANDLE magic, use explicit VkHandle wrapper
>
> Overriding VK_DEFINE_NON_DISPATCHABLE_HANDLE changes the function
> signatures of Vulkan functions, changing their ABI and making us
> incompatible with real drivers. This removes that magic, and replaces it
> with an explicit wrapper, VkHandle, which has much of the same
> functionality as the original VkNonDispatchableHandle.
>
> It adds definitions for dawn_native::vulkan::VkBuffer et al, which
> shadow the native ::VkBuffer et al. This retains type safety throughout
> the Vulkan backend without changing every single usage.
>
> Notably, the following things had to change:
> - An explicit conversion from VkBuffer* to ::VkBuffer* is needed for
> arrays. This is implemented as a reinterpret_cast, which is still
> safe as the new VkHandle still has the same memory layout properties
> as VkNonDispatchableHandle did.
> - When pointing to a VkHandle as an output pointer, it's now necessary
> to explicitly get the native ::VkBuffer (via operator*) and point to it.
>
> Bug: chromium:1046362
> Change-Id: I9c5691b6e295aca1b46d4e3d0203956e4d570285
> Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/15580
> Reviewed-by: Austin Eng <enga@chromium.org>
> Reviewed-by: Kai Ninomiya <kainino@chromium.org>
> Commit-Queue: Kai Ninomiya <kainino@chromium.org>
TBR=cwallez@chromium.org,kainino@chromium.org,enga@chromium.org
Change-Id: I500df2e34fd0f245ad04c517ff028ddd7bb5a2bf
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: chromium:1046362
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/15620
Reviewed-by: Kai Ninomiya <kainino@chromium.org>
Commit-Queue: Kai Ninomiya <kainino@chromium.org>
Overriding VK_DEFINE_NON_DISPATCHABLE_HANDLE changes the function
signatures of Vulkan functions, changing their ABI and making us
incompatible with real drivers. This removes that magic, and replaces it
with an explicit wrapper, VkHandle, which has much of the same
functionality as the original VkNonDispatchableHandle.
It adds definitions for dawn_native::vulkan::VkBuffer et al, which
shadow the native ::VkBuffer et al. This retains type safety throughout
the Vulkan backend without changing every single usage.
Notably, the following things had to change:
- An explicit conversion from VkBuffer* to ::VkBuffer* is needed for
arrays. This is implemented as a reinterpret_cast, which is still
safe as the new VkHandle still has the same memory layout properties
as VkNonDispatchableHandle did.
- When pointing to a VkHandle as an output pointer, it's now necessary
to explicitly get the native ::VkBuffer (via operator*) and point to it.
Bug: chromium:1046362
Change-Id: I9c5691b6e295aca1b46d4e3d0203956e4d570285
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/15580
Reviewed-by: Austin Eng <enga@chromium.org>
Reviewed-by: Kai Ninomiya <kainino@chromium.org>
Commit-Queue: Kai Ninomiya <kainino@chromium.org>
The Vulkan spec mandates support for only one or the other, which is
why we have the concept of a depth24plus format. This also adds a Toggle
to test both formats in DepthStencilStateTests.
Finally this renames ForceWorkarounds to ForceToggles because toggles
can be more than just workarounds.
BUG=dawn:286
Change-Id: I5b5dc582ffd4ee61c51e3e75563aec815c580511
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/14103
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: David Turner <digit@google.com>
This CL implements the MemoryService for importing memory and
creating VkImages from a dma-buf handle. Under the hood, it uses the
VK_EXT_external_memory_dma_buf and
VK_EXT_image_drm_format_modifier extensions to find a memory type
that supports dma-buf import. In addition, the extensions are also
used to properly specify the stride and tiling of the dma-buf to
vkAllocateMemory and vkCreateImage.
BUG=chromium:996470
Change-Id: Ie72d73117a4cbafcb40468aab0952b783351d499
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/13785
Commit-Queue: Brian Ho <hob@chromium.org>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
This CL adds an API to import a dma-buf into Dawn as a WGPUTexture.
We also add a descriptor type enum to the base
ExternalImageDescriptor struct. This is because all memory import
code (e.g. MemoryService, Texture::CreateFromExternal) takes the
a base ExternalImageDescriptor as a parameter. The dma-buf external
memory and image services, however, will need to downcast to
ExternalImageDescriptorDmaBuf to access import parameters like
stride. Explicitly adding a type enum will let us more safely verify
the type before downcasting.
BUG=chromium:996470
Change-Id: I2d9883a15e9059a91f2c7bdb7a96d74373e18c56
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/13782
Reviewed-by: Austin Eng <enga@chromium.org>
Commit-Queue: Brian Ho <hob@chromium.org>
This CL introduces SupportsCreateImage and CreateImage to the
external memory service API to provide different implementations for
creating VkImages for different types of external image handles.
While opaque FD and Zircon seem to share the same vkCreateImage
implementation, dma-buf import will require the use of the
VkImageDrmFormatModifierExplicitCreateInfoEXT [1] struct (among
other differences) with vkCreateImage.
[1] https://www.khronos.org/registry/vulkan/specs/1.1-extensions/man/html/VkImageDrmFormatModifierExplicitCreateInfoEXT.html
BUG=chromium:996470
Change-Id: I3eb11f8877d4465f5fcdd4089d5fdd8acbc1da10
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/13781
Reviewed-by: Austin Eng <enga@chromium.org>
Commit-Queue: Brian Ho <hob@chromium.org>
This CL is part of a chain of CLs that imports dma-bufs as VkImages
to support WebGPU on Chrome OS.
There are currently two steps for importing external memory into
Vulkan:
1. DeviceVk::ImportExternalImage: calls into
MemoryServiceOpaqueFD::ImportMemory which in turn calls into
vkAllocateMemory and outputs a VkDeviceMemory handle to the
imported memory.
2. TextureVk::CreateFromExternal: creates the actual TextureVk
object, creates the VkImage, and binds the VkDeviceMemory from
ImportExternalImage to the VkImage.
For dma-buf support, however, we need to re-order these two steps
because importing dma-buf memory requires a handle to the VkImage
that will alias it [1].
This CL splits these two steps into three steps to ensure we create
the VkImage first so we can use it in vkAllocateMemory:
1. TextureVk::CreateFromExternal: creates the TextureVk and
VkImage (no longer concerns itself with vkBindImageMemory).
2. DeviceVk::ImportExternalImage: now takes the VkImage as input
but is otherwise unchanged.
3. TextureVk::BindExternalMemory: calls vkBindImageMemory with
handles to VkDeviceMemory and VkImage.
[1] https://www.khronos.org/registry/vulkan/specs/1.1-extensions/man/html/VkMemoryDedicatedAllocateInfo.html
BUG=chromium:996470
Change-Id: Id2d5951e9b573af79c44ce8c63be5210a279f946
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/13780
Commit-Queue: Brian Ho <hob@chromium.org>
Reviewed-by: Austin Eng <enga@chromium.org>
This makes the Vulkan backend use the BuddyMemoryAllocator to
sub-allocate small resources inside a larger VkDeviceMemory object.
Right now the heuristic to decide to do suballocation is naive and
should be improved.
BUG=dawn:27
Change-Id: Idcc7b6686c086633c85328a7afb91ee84abf7b8c
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/12662
Reviewed-by: Austin Eng <enga@chromium.org>
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
This removes the duplication of the memory allocators in preparation for
using sub-allocation in the Vulkan backend too.
Also renames ResourceMemory to ResourceHeap and MemoryResourceAllocator
to ResourceMemoryAllocator, and fixes a number of unused includes.
BUG=dawn:27
Change-Id: I1a9e7d41e5efafa5192bda1d89dc06455fa2af40
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/12660
Reviewed-by: Bryan Bernhart <bryan.bernhart@intel.com>
Reviewed-by: Austin Eng <enga@chromium.org>
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
Dawn creates a keyed shared mutex for all wrapped resources and acquires
the mutex before the texture is used in an ExecuteCommandList call.
To coordinate with external clients, backend API has been extended
to allow clients to get and set the acquire key.
Pending and queue command lists have now been merged into one.
A future change will adjust GetPendingCommandContext to return a raw
command context without the need for error handling.
Bug:dawn:217
Change-Id: Ia96c449c305586407153f05ce75a40794b96027e
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/12220
Reviewed-by: Austin Eng <enga@chromium.org>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Commit-Queue: Rafael Cintron <rafael.cintron@microsoft.com>
This also introduces another combinator to ConsumeError for
ResultOrError.
BUG=dawn:19
Change-Id: Ic204313436f5e919473d604efd049fe3d3c27a66
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/11862
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Austin Eng <enga@chromium.org>
Reviewed-by: Kai Ninomiya <kainino@chromium.org>
This makes all the regular WebGPU object creation handle errors properly
in the Vulkan backend instead of ASSERTing no Vulkan error is raised.
Static Create functions are added to all these types so that the details
of how the initialization is done is private, and it isn't possible to
construct an object but forget to initialize it.
BUG=dawn:19
Change-Id: I362b2d66b74dd7799ffbf69d732bc58caa97950b
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/11861
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Kai Ninomiya <kainino@chromium.org>
When storeOp is clear, texture subresource is set as not initialized
Bug: dawn:145
Change-Id: I364d7239a7ebdb9d5a28a4af559f3212be7ef15a
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/11560
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Austin Eng <enga@chromium.org>
Clears nonrenderable color formats and merges zero vs nonzero clears
to use the same code path.
Bug: dawn:145
Change-Id: I8f2f36134b56787e07231d82e37c36897ba1d4ba
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/10820
Commit-Queue: Natasha Lee <natlee@microsoft.com>
Reviewed-by: Kai Ninomiya <kainino@chromium.org>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Use dynamic uploader so that we don't have to create a new buffer each
time we lazy clear.
Bug: dawn:145
Change-Id: I2d7a769532bd303fdabb7fc70051b70a1114cc07
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/10760
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Commit-Queue: Natasha Lee <natlee@microsoft.com>
This is to match the naming convention of WebGPU's WebIDL and webgpu.h
BUG=dawn:22
Change-Id: Ia91c5a018403e6a72eb0311b5f1a072d102282a2
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/10461
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Austin Eng <enga@chromium.org>
They were removed from the WebGPU spec because they cannot be supported
everywhere, for example on recent Qualcomm chips.
BUG=dawn:128
Change-Id: I189bd1b34105f215699ddd404341bbee964c8679
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/10280
Reviewed-by: Austin Eng <enga@chromium.org>
Reviewed-by: Kai Ninomiya <kainino@chromium.org>
Commit-Queue: Kai Ninomiya <kainino@chromium.org>
Clears nonrenderable color formats and adds a clearValue enum to help
share the code path between zero and nonzero clears.
Bug: dawn:145
Change-Id: I285521cae0ee71625602b949888b21481a05fb8e
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/10021
Commit-Queue: Natasha Lee <natlee@microsoft.com>
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>
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>
The CL that introduced this formats reordered their components based on
the name of the Vulkan formats, but it turns out that WebGPU lists
components from low bits to high-bits while Vulkan PACK32 formats are
listed from high to low. So the format component orders match, except
for RGB10A2 which is actually BGR10A2 in Vulkan. Instead the Vulkan
backend is updated to correctly use the RGB10A2 format.
BUG=dawn:128
Change-Id: If7f045f020429c44c84b9aed34a4a80107208d5c
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/8601
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Jiawei Shao <jiawei.shao@intel.com>
This is to match the equivalent rename in WebGPU's IDL.
BUG=dawn:22
Change-Id: Ibdf75869e58835d984df559878b884c55310a403
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/8623
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Yunchao He <yunchao.he@intel.com>
This removes the R5G6B5 format that's only available on iOS on Metal and
leaves A2RGB10 unimplemented because it has inverted R and G channels
compared to Metal's RGB10A2 format.
BUG=dawn:128
Change-Id: I9b960d00f281b4d32318bad3360826b8db5bcb72
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/8600
Reviewed-by: Austin Eng <enga@chromium.org>
Reviewed-by: Jiawei Shao <jiawei.shao@intel.com>
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
This patch supports BC5 formats on Vulkan backends and adds related
Dawn end2end tests.
For the textures with BC formats, they could have non-multiple-of-4
sizes on the non-zero mipmap levels in sampling, but we are still
required to provide texture data in complete 4x4 blocks in texture
copies because that is the size of which they are stored in GPU memory.
In this patch, we refer the term "physical memory size" as the memory
size of the texture subresource in GPU memory, and the term "virtual
memory size" as the size used in texture sampling. As Dawn requires
the Extent3D in texture copies must fit in the physical memory size,
while Vulkan requires it must fit in the virtual memory size, this
patch recalculates the imageExtent to ensure it always follow this
Vulkan validation rules.
For Dawn end2end tests, note that we use pure green and pure red for
the textures because BC5 does not support SRGB formats. Furthermore,
"CopyPartofTextureSubResourceIntoNonZeroMipmapLevel" is skipped in
this patch because there is an issue on the T2T copies from a region
within the virtual size of one texture to another one that exceeds
the virtual size of another texture in Vulkan SPEC.
BUG=dawn:42
TEST=dawn_end2end_tests
Change-Id: I17518cd335fb13125cb753bbf879bc06eb20e426
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/8260
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Austin Eng <enga@chromium.org>
Commit-Queue: Jiawei Shao <jiawei.shao@intel.com>
This adds the formats to dawn.json, implements support in the Vulkan
backend and adds tests performing basic sampling checks for all formats.
The R8UnormSrgb and RG8UnormSrgb formats skipped because they are not
required in Vulkan (and RG8UnormSrgb is in fact not supported on the
machine used for developing this CL). A PR will be sent to the WebGPU
repo to remove the from the initial list of formats.
The RG11B10Float and RGB10A2Unorm formats of WebGPU are replaced with
B10GR11Float and A2RGB10Unorm that are the formats exposed by Vulkan. It
is likely that all APIs implement them with components stored in that
order.
Each format except depth-stencil ones is tested by uploading some
interesting texel data and checking that sampling from the texture
produces correct results. The goal is to make sure that backends don't
make a mistake in the giant switch statements. There was no effort made
to check the hardware implementation of the formats.
Tests will later be extended to cover rendering and clearing operations
as well as multisample resolve.
It isn't clear if depth-stencil format will support TRANSFER operations
in WebGPU so these are left untested for now.
BUG=dawn:128
Change-Id: I78ac5bf77b57398155551e6db3de50b478d69452
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/8363
Reviewed-by: Austin Eng <enga@chromium.org>
Reviewed-by: Jiawei Shao <jiawei.shao@intel.com>
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
Texture formats have plenty of properties we'd like to query on them
like their texel size, whether they are compressend or if they are depth
stencil. Instead of making switch statements for each of these
properties, we store them in a new `Format` structure. Textures compute
their format only onces and then pass a const reference to it on
GetFormat().
This is in preparation of adding all WebGPU texture formats.
BUG=dawn:128
Change-Id: Iad2831cf16f14e1a1bfce2c10b22527fc982d1aa
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/8166
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Kai Ninomiya <kainino@chromium.org>
Also ensure depth stencil texture is clear in begin render pass.
Bug: dawn:175, dawn:145
Change-Id: I19fc857dedf9f353a4834b8e1c9616d02c5a15ac
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/8140
Commit-Queue: Natasha Lee <natlee@microsoft.com>
Reviewed-by: Kai Ninomiya <kainino@chromium.org>
This patch adds VK_IMAGE_USAGE_TRANSFER_DST_BIT usage when creating
Vulkan images because when we do resource initialization with
vkCmdClearColorImage() and vkCmdClearDepthStencilImage(), the image must
have been created with VK_IMAGE_USAGE_TRANSFER_DST_BIT usage flag
according to Vulkan SPEC.
BUG=dawn:171
TEST=dawn_end2end_tests
Change-Id: Iaba6c04c6942354cc6be502eeee04457d3f43ba7
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/8100
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Austin Eng <enga@chromium.org>
Commit-Queue: Jiawei Shao <jiawei.shao@intel.com>
This prevents dirty textures to be used when memory is recycled
while destroying/creating textures. If a texture is not cleared at load,
it will be cleared to 0 before it is used.
Bug: dawn:145
Change-Id: Ia3f02427478fb48649089829186ccb377caa1912
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/6960
Commit-Queue: Natasha Lee <natlee@microsoft.com>
Reviewed-by: Kai Ninomiya <kainino@chromium.org>
