|
2b723c13
|
2024-02-02T14:12:06
|
|
CL/VK: Add initial enqueueCopyBuffer
Generalizing the copy buffer with appropriate
barrier inserts and commandbuffer command.
Bug: angleproject:42267074
Change-Id: I02fd36a221ea70f007aee78ceeef84c24fa85682
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5410935
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
da572160
|
2024-07-23T16:36:10
|
|
Reland: GL: Forward client-side arrays to the driver when possible
The OpenGL driver can handle client-side arrays when the context is
OpenGL ES or a desktop GL compatibility profile. When in these
situations, use the driver default VAO for all frontend context VAOs
and forward client-side data directly to the driver.
Fix synchronizing the default VAO state for external contexts. There
is no valid VertexArrayStateGL for external VAOs so make sure it's
nulled and the VAO dirty bits are set so the correct VAO state is
reapplied.
Disable syncing to the default VAO for external contexts. The only
VAO that they can share with ANGLE's internal state is the default
VAO so avoid having to save and restore its state.
Bug: angleproject:355034686
Change-Id: I015bbbc854938fe4bc1e92d0ca8fe04628d0db16
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5743284
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Commit-Queue: Geoff Lang <geofflang@chromium.org>
|
|
|
6fddb18d
|
2024-07-26T12:44:23
|
|
GL: Disable heavily emulated extensions for non-WebGL.
The multidraw and base vertex base instance extensions are heavily
emulated in ANGLE's GL backend, manipulating vertex arrays multiple
times per draw. This is often a performance improvement for deep
pipelines such as WebGL but slower for shallow stacks like Chrome's
rasterization.
Mark these extensions as "emulated" and disable them in the frontend
for non-WebGL contexts. Only EXT_base_instance can have true native
support.
Add a frontend feature to re-enable these extensions and use it in the
existing end2end tests.
Bug: angleproject:355645824
Change-Id: I897559fb270e2aa0f476d7784d25d97574e28f43
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5743837
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Commit-Queue: Geoff Lang <geofflang@chromium.org>
|
|
087afc91
|
2024-08-10T20:40:53
|
|
Vulkan: Prefer fallback to renderable texture than CPU copy
When a texture copy is done, the transfer path is preferred. If
transfer is not possible, ANGLE falls back to a draw-based copy.
However, if the format is not renderable, ANGLE falls back to CPU copy.
Instead, this change makes ANGLE try to fall back to a renderable format
before giving up on the draw-based copy.
Bug: b/351872449
Change-Id: I3d09dd88cc0b7feb5e92e58c4f19380973eb5e94
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5775592
Reviewed-by: Amirali Abdolrashidi <abdolrashidi@google.com>
Reviewed-by: Charlie Lao <cclao@google.com>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
5967afe7
|
2024-04-02T17:45:48
|
|
CL: Update Platform ctor to handle rx errors
Bug: angleproject:359294584
Change-Id: I7aad107e57ea6f29b8fc69205ba72c27de6a9428
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5421576
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Commit-Queue: Austin Annestrand <a.annestrand@samsung.com>
|
|
|
072dc5d1
|
2024-08-13T13:53:52
|
|
Metal: Disable multisampled render to texture.
Corruption has been seen on M1 Macs for WebGL content which uses a
multisampled backbuffer.
Bug: chromium:358957665, angleproject:42261786
Change-Id: Ia17581b15cd032e91e2642d80abace8d57ceb23b
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5783540
Commit-Queue: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Quyen Le <lehoangquyen@chromium.org>
|
|
|
b36d94dc
|
2024-08-02T16:23:03
|
|
WebGPU: Implement viewport and scissor
Tested using samples. dEQP tests covering viewport and scissor require
more shader translation to be completed.
Bug: angleproject:0
Change-Id: Ia76f2edeea5a07af1881704a21fafb64ab7df43a
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5756127
Reviewed-by: Liza Burakova <liza@chromium.org>
Commit-Queue: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
5b4517ec
|
2024-02-02T14:04:29
|
|
CL/VK: Add helper for inserting barriers
Adding a helper function to insert execution and memory barriers when
resource access commands are recorded into the command buffer in
CLCommandQueueVk.
An example of a copy buffer command with src and dst
```
set access.readBuffer <- src
set access.writeBuffer <- dst
insert barriers - onResourceAccess(access)
add copy command - commandBuffer->copyBuffer()
```
Bug: angleproject:42267073
Change-Id: I61d8064fd015a4bf3dab018cbb83d147915ed627
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5409249
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Commit-Queue: Austin Annestrand <a.annestrand@samsung.com>
|
|
f8fc8ac3
|
2024-08-05T11:50:11
|
|
Vulkan: Remove dependency on ContextVk for CommandBufferHelper
Following on the changes in [1], this makes the
`CommandBufferHelperCommon` and `OutsideRenderPassCommandBufferHelper`
interfaces independent of `ContextVk` state. Any dependency is made
explicit.
In addition, interfaces that are not specific to GLES context are also
updated.
[1]: Commit (bcf814fda5 Vulkan: Constrain the dependency on ContextVk in
BufferHelper)
Bug: angleproject:8544
Change-Id: I7d90ad915e8c14187ab5584453b9e8802bd91e2b
Signed-off-by: Gowtham Tammana <g.tammana@samsung.com>
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5319147
Reviewed-by: Charlie Lao <cclao@google.com>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
ee25dddb
|
2024-08-09T01:03:26
|
|
Metal: Properly clear render pass' attachment with an emulated alpha.
When a texture is created with emulated format such as RGB, its alpha
channel is initialized to 1 and the write mask is configured such that
the alpha value won't be overwritten or set to other values.
However, if we use this texture as an attachment in a render pass, and
DontCare loadAction is used, the alpha value would be initialized
with garbage values. No draw calls within the render pass would change
these garbage values because the configured write mask already prevents
that. Thus the garbage alpha values would be stored in the memory after
the render pass ends.
This bug also affects MSRTT's unresolve step, because it uses a draw
call to blit the resolve texture to the MSAA texture. However, the
alpha write is disabled thus the MSAA texture will remain having garbage
alpha values.
The fix for this bug is that: if we detect that the texture has emulated
format and its loadAction is DontCare, we change it to Clear action to
clear the alpha channel to 1.
Bug: angleproject:42261786
Change-Id: I994849bd4b3c3ab51698833d1cc520376ae6de44
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5773347
Reviewed-by: Kenneth Russell <kbr@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Auto-Submit: Quyen Le <lehoangquyen@chromium.org>
Commit-Queue: Quyen Le <lehoangquyen@chromium.org>
|
|
6e248bcb
|
2024-06-20T00:00:00
|
|
Metal: Fix ASTC on visionOS
Adjusted compile guards to
include visionOS target.
Fixed: angleproject:358393363
Change-Id: If7ead234f5deaee524172d2fb38d4a041c915716
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5777140
Commit-Queue: Alexey Knyazev <lexa.knyazev@gmail.com>
Reviewed-by: Kimmo Kinnunen <kkinnunen@apple.com>
Reviewed-by: Kenneth Russell <kbr@chromium.org>
|
|
2b15c795
|
2024-08-09T17:13:17
|
|
Make explicit that GenerateCrc() return correct CRC32 hash
According to the documentation of `crc32_z()`, to get required
initial value need to call `crc32_z()` with nullptr "buf" parameter.
Using this initial value will produce correct CRC32 hash.
On practice, `crc32_z()` with nullptr "buf" produces "0", so the current
code actually calculates correct CRC32 hash. However, this is still
against the documentation, and in different versions of zlib or
architectures may produce incorrect results.
Technically, `GenerateCrc()` does not require to produce correct CRC32
hash - the main purpose is to calculate some hash value.
But because updating this function to produce correct CRC32 hash is very
easy, this change updates the implementation and renames it to
`GenerateCRC32()` to explicitly state that it generates correct CRC32
hash. After this, function may be used in scenarios where CRC32 hash
is really required.
Bug: angleproject:4722
Change-Id: Ib159f2d3ad1a8388f505abb09a6d0b19a51b7eec
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5777145
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
7bab5bce
|
2024-06-20T00:00:00
|
|
Metal: Support BC texture formats with non-macOS targets
Used supportsBCTextureCompression device query
to check BC formats support on Apple GPUs.
Fixed: angleproject:42266573
Change-Id: I60fd5614d10d0c4558992e2c5d3e4b4c1a396aaa
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5771136
Commit-Queue: Alexey Knyazev <lexa.knyazev@gmail.com>
Reviewed-by: Kimmo Kinnunen <kkinnunen@apple.com>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
|
|
|
bacd3a7d
|
2024-08-07T19:01:18
|
|
Vulkan: Fix CompressAndStorePipelineCacheVk() crashing
`ContextVk` may be destroyed while
`CompressAndStorePipelineCacheTask()` is still running. Context is only
used to output a performance warning. So in case of such performance
warning, `gl::State::mDebug` object may be already destroyed, causing
undefined behavior.
Change removes `ContextVk` from the function and uses `WARN()` instead.
The `CompressAndStorePipelineCacheTask` stores pointer to the
`vk::Renderer` instead of `ContextVk`.
Bug: angleproject:4722
Change-Id: Iba9cad78ce5c6ea29cc2143f7bd2340cf0484855
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5768881
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Charlie Lao <cclao@google.com>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
98b06696
|
2024-06-17T17:28:56
|
|
CL/Vulkan: Add initial profiling support
This PR introduces initial cl_event profiling
support. For now, use CPU timestamps
(later look into Vulkan GPU device timestamps).
Bug: angleproject:357902512
Change-Id: I2197e396e222d2842ab31304f91934f9e000ea0d
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5767598
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Commit-Queue: Austin Annestrand <a.annestrand@samsung.com>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
|
|
|
d1d9cd63
|
2024-06-17T17:22:16
|
|
CL/Vulkan: Init events to NULL before enqueue
On a CL enqueue API call, if the passed cl_event
object ref is not NULL, set the object to NULL
initially before we create it during enqueue call.
Bug: angleproject:357902512
Change-Id: I7d94012c38eb96e7021da20ea22bd741e536de53
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5767597
Commit-Queue: Austin Annestrand <a.annestrand@samsung.com>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
|
|
cb27b4a4
|
2024-08-08T10:17:47
|
|
Metal: Remove MTLGPUFamilyApple9, it is unused
Fixes visionOS compiles, the SDKs do not define the symbols.
Changes the SDK #if checks to be more consistent:
- Remove version checks from WATCHOS, as it's expected to be compiled
with fairly recent SDK (Metal is not supported).
- Compile fallbacks only when needed by the minimum deployment
target.
- Add Apple3 -> :MTLFeatureSet_tvOS_GPUFamily2_v1 mapping just for
consistency.
Bug: angleproject:347739670
Change-Id: I054c9faeba4245be8be6cfa7128f17df03f448da
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5771352
Auto-Submit: Kimmo Kinnunen <kkinnunen@apple.com>
Commit-Queue: Kimmo Kinnunen <kkinnunen@apple.com>
Commit-Queue: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Alexey Knyazev <lexa.knyazev@gmail.com>
|
|
|
1010a275
|
2024-06-18T10:27:10
|
|
CL/Vulkan: Set proper build status prior to API return
For cl[Compile/Link/Build]Program, we need to
set the build status to CL_BUILD_IN_PROGRESS
before returning to user/app (since in build
callback case, user can query status as either
in-progress, done, or error).
CL_BUILD_NONE would not be a valid query
return if the user just launched a build in this case.
Bug: angleproject:357905825
Change-Id: I85f23db7f9a543ff8872557de5c1cf41d2f2645a
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5767599
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Commit-Queue: Austin Annestrand <a.annestrand@samsung.com>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
|
|
bc0c6c5b
|
2024-08-07T09:48:25
|
|
VVL: sync_queue_submit -> syncval_submit_time_validation
Avoids this warning printed to stdout:
Validation Setting Warning - sync_queue_submit was set, this is deprecated, please use syncval_submit_time_validation
Bug: b/316013423
Change-Id: I9a8be1a49be72e8ec2fed74ace36989d8e37163a
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5769522
Auto-Submit: Roman Lavrov <romanl@google.com>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
27c0f10a
|
2024-08-06T11:12:32
|
|
WebGPU: Generate texture formats when initializing surfaces.
Bug: angleproject:344814096
Change-Id: I873c6308e9c203fd759ccbf5a378056f2aca849c
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5763218
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Commit-Queue: Liza Burakova <liza@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
|
|
|
057bc0f0
|
2024-08-07T11:38:54
|
|
WebGPU: Set multiple parameters in RenderPipelineDesc.
This change adds multiple setters to RenderPipelineDesc to set
various parameters within the DepthStencilState.
Bug: angleproject:42267012
Change-Id: I71eed7195477572295de6575bfa18d2c58ca93de
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5765553
Reviewed-by: Matthew Denton <mpdenton@chromium.org>
Commit-Queue: Liza Burakova <liza@chromium.org>
|
|
|
8a8b85bb
|
2024-06-20T00:00:00
|
|
Metal: Adjust supports32BitFloatFiltering compile guards
* Used MAX_ALLOWED instead of MIN_REQUIRED macros.
* Explicitly enabled visionOS SDK.
* Used correct tvOS SDK version.
Bug: angleproject:347739670
Change-Id: I599ce31c66fc4341a43607c46df67740b8478670
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5756573
Reviewed-by: Kimmo Kinnunen <kkinnunen@apple.com>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Commit-Queue: Alexey Knyazev <lexa.knyazev@gmail.com>
|
|
0592b897
|
2024-08-02T14:50:08
|
|
Metal: fix frame capture bundle file permissions
Capturing Metal frame captures of ANGLE end2end test results in a
.gputrace bundle devoid of any posix file permissions. This causes Xcode
to throw an error when trying to open the frame capture.
On |StopFrameCapture|, ensure that the .gputrace files are RW for the
user and the bundle directory is RWX.
Fixed: angleproject:356907817
Change-Id: Ia994aa5b8bd09a3dc34fd79aac831a6f7dcc792f
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5759063
Reviewed-by: Quyen Le <lehoangquyen@chromium.org>
Commit-Queue: Quyen Le <lehoangquyen@chromium.org>
|
|
|
70f2e203
|
2024-08-06T17:21:50
|
|
Metal: Fix memoryless usage on ios on tv
Bug: angleproject:42261786
Change-Id: I8a289eb888dbb9b2d28066850f2218a1aa92da3b
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5765976
Auto-Submit: Shahbaz Youssefi <syoussefi@chromium.org>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Quyen Le <lehoangquyen@chromium.org>
Commit-Queue: Quyen Le <lehoangquyen@chromium.org>
|
|
|
7a81e79d
|
2024-08-02T19:39:57
|
|
Vulkan: Fix mPipelineCache not populated from blob cache
Fixes regression introduced in change:
Vulkan: Remove support for pipeline cache control
https://chromium-review.googlesource.com/c/angle/angle/+/5375102
Returns logic implemented in:
Vulkan: pipeline cache not populated as blob cache is not set
https://chromium-review.googlesource.com/c/angle/angle/+/1683807
In the current state, call to `ensurePipelineCacheInitialized()` from
`getPipelineCache()` is essentially a dead code, since
`mPipelineCacheInitialized` will always be `true` after call from
`createDeviceAndQueue()`. So if blob functions are not set before
`createDeviceAndQueue()` then blob cache data will never be loaded.
This change simply removes `ensurePipelineCacheInitialized()` from the
`createDeviceAndQueue()`, since cache can only be accessed after the
initialization (`mPipelineCacheInitialized` is true).
Bug: angleproject:8601
Change-Id: I3db1cd69346ec1aa2c36e4134c77680526b22cc8
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5756574
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
721becf2
|
2024-08-06T17:49:56
|
|
Vulkan: Update Renderer::syncPipelineCacheVk() method
Add check for `mPipelineCacheInitialized` to avoid overwriting existing
pipeline cache data in the blob cache by empty cache data if sync is
called before pipeline cache is loaded from the blob cache.
Move checking of `mCompressEvent` before getting pipeline size and
updating `mPipelineCacheSizeAtLastSync`. This will avoid unnecessary
(potentially heavy) vulkan call and fix bug when next sync is skipped,
because `mPipelineCacheSizeAtLastSync` is already updated.
Bug: angleproject:4722
Change-Id: I7390c4bc53ba57732b659d79c65ed4da8d7a6ba6
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5756575
Reviewed-by: Charlie Lao <cclao@google.com>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
3e4fa128
|
2024-08-05T16:20:30
|
|
Store ImageHelper's queue by mip levels.
This change converts ImageHelper's mSubresourceQueue to store
a vector of updates for each mip level. This allows updates
to be flushed per mip level.
This also adds a call in the RenderTargetWgpu to flush updates
such that now when a framebuffer is flushing updates to
color attachments the associated render target ensures that
only updates associated with its mip level are flushed.
Bug: angleproject:42267012
Change-Id: I1abdbc842cf18b1bd897037bf11aeea9c6d09e14
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5759469
Reviewed-by: Matthew Denton <mpdenton@chromium.org>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Commit-Queue: Liza Burakova <liza@chromium.org>
|
|
|
67d8d8c8
|
2024-07-29T13:44:37
|
|
WebGL: Reject shaders with infinite loops
Bug: chromium:350528343
Change-Id: I1b2fc152cf285b0e69c4c294351c1cf2389cc234
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5746714
Commit-Queue: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Kenneth Russell <kbr@chromium.org>
Auto-Submit: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
c5d4c16f
|
2024-08-02T11:27:56
|
|
WebGPU: Initialize all GLES 2 and 3 caps
Instead of using the min caps, intialize them with limits from the
WebGPU limits struct or values from the spec.
Bug: angleproject:357017322
Change-Id: I15597b192d915dad7c3f6bacab048cf09999da6d
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5759280
Reviewed-by: Liza Burakova <liza@chromium.org>
Commit-Queue: Geoff Lang <geofflang@chromium.org>
|
|
9b06e003
|
2024-08-02T15:47:36
|
|
Vulkan: Avoid unnecessary observer changes for Buffers
Noticed in black_desert_mobile hat it calls glVertexAttribPointer with
some buffer. What we end up doing is remove observer from old buffer and
add observer to new buffer. This CL checks if old and new buffer are the
same and skip some of the logic if both buffers are the same.
Also added check in VertexArrayBufferContentsObservers::disableForBuffer
to early out if it was not enabled in the first place.
Bug: b/356473483
Change-Id: I872dcee7d87b404f525d0323d2952c8249ef1c9a
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5760652
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Yuxin Hu <yuxinhu@google.com>
Commit-Queue: Charlie Lao <cclao@google.com>
|
|
|
a89ca6c8
|
2024-07-30T17:14:50
|
|
Metal: Fully implement unresolve step for MSRTT.
This CL also uses memoryless for implicit MS textures in
EXT_multisampled_render_to_texture render passes.
Bug: angleproject:42261786
Change-Id: Ic20450574c4c670d05ca846587ced05488012270
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5746181
Reviewed-by: Alexey Knyazev <lexa.knyazev@gmail.com>
Reviewed-by: Kenneth Russell <kbr@chromium.org>
Commit-Queue: Quyen Le <lehoangquyen@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
|
|
|
1bfe5c57
|
2024-07-30T15:58:47
|
|
Metal: partially implement EXT_multisampled_render_to_texture.
Implement by implicitly attaching a multisampled texture to the render
pass. The content will be preserved across render passes by
loading/storing to the implicit multisampled texture.
However this won't work if the single sampled texture is used in
multiple render passes with different
glFramebufferTexture2DMultisampleEXT's sample counts. For that to work
we need to implement unresolve step to load the resolve texture's texels
into the implicit multisampled texture. That will be implemented in a
separate CL.
Bug: angleproject:42261786
Change-Id: I12be75af17ce5b98266946846417d0a43fcba455
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5746180
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Commit-Queue: Quyen Le <lehoangquyen@chromium.org>
|
|
|
26aa4df5
|
2024-06-20T00:00:00
|
|
Metal: Update floating-point texture format caps
* Fixed detection of Apple7+ GPUs. Added support
for Apple8 and Apple9 GPUs.
* Adjusted filterability and resolvability caps for
R32F, RG32F, and RGBA32F formats. This enables
GL_OES_texture_float_linear on capable devices.
* Adjusted RGBA32F blendability cap. This enables
GL_EXT_float_blend on capable devices.
Bug: angleproject:347739670
Change-Id: I84d23814d119452f47fff5a1d7dc33f614c35bb7
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5646489
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Commit-Queue: Alexey Knyazev <lexa.knyazev@gmail.com>
Reviewed-by: Kimmo Kinnunen <kkinnunen@apple.com>
|
|
22ed9877
|
2024-03-07T15:52:13
|
|
Vulkan: Generate fragment shading rates with compute
Generating fragment shading rates on the CPU could cause avoidable
stalls or increased latencies due to host<->device synchronization.
Instead generate the rates using a compute shader.
Optimize foveated rendering by treating a foveated draw with focal
points whose gainX or gainY is 0 as if it were an un-foveated draw
Bug: angleproject:42266906
Change-Id: If8c5504087997666c7d0c1cae8dcc5fab847187e
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5754322
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Yuxin Hu <yuxinhu@google.com>
Commit-Queue: mohan maiya <m.maiya@samsung.com>
|
|
|
5a159d69
|
2024-05-03T10:24:01
|
|
CL/Vulkan: Pass proper arch flag to clspv
Needed to update the processedOptions
field so that we append the "-arch=" flag
based off the device's address bits value.
Bug: angleproject:356328474
Change-Id: I5f42f4cdf30ccef9e9aad8eb040f6cbdf8d3b750
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5753767
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Commit-Queue: Austin Annestrand <a.annestrand@samsung.com>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
|
|
|
cc50c75b
|
2024-04-19T16:40:26
|
|
CL/Vulkan: Fix missing kernel enqueue retain/release
According to spec:
https://registry.khronos.org/OpenCL/specs/3.0-unified/html/OpenCL_API.html#clReleaseKernel
We also need to implicitly retain/release the Kernel object
on Kernel enqueue APIs along with other kernel resource
retain/release(s) (i.e. memobjs).
Bug: angleproject:356328473
Change-Id: Ie6ca228ebece2988d511d9ffd2a617abdc057d0a
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5753766
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Commit-Queue: Austin Annestrand <a.annestrand@samsung.com>
|
|
|
d4f93407
|
2024-03-25T18:14:26
|
|
CL: Conditional enable of Vulkan Validation layers
Rather than force YesIfAvailable always,
we restrict the Vk validation layers if
user specifies it in the GN args:
angle_enable_vulkan_validation_layers
Bug: angleproject:356320155
Change-Id: I9035d0e7d220ce59db39f1a86f31b403c44088dd
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5753765
Commit-Queue: Austin Annestrand <a.annestrand@samsung.com>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
31c0f439
|
2024-08-02T09:18:33
|
|
WebGPU: Centralize caps generation
All GL and EGL caps can be generated at display creation time for
WebGPU. Move all existing logic into a centralized GenerateCaps
function.
Bug: angleproject:357017322
Change-Id: I7398d4ee539192d64971023d9a598ac1f11592ad
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5759279
Commit-Queue: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Liza Burakova <liza@chromium.org>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
ea63e372
|
2024-08-01T21:10:02
|
|
Enable VK_KHR_synchronization2 if it is available
If the device supports VK_KHR_synchronization2 and
VkPhysicalDeviceSynchronization2Features.synchronization2
queried from the device is VK_TRUE,
Enabled the extension VK_KHR_synchronization2
and feature VkPhysicalDeviceSynchronization2Features
when calling vkCreateDevice.
Bug: b/356985874
Bug: b/336844257
Change-Id: Iac95f373a283f5a97e126770321769e0f760e5f9
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5759319
Reviewed-by: Charlie Lao <cclao@google.com>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Commit-Queue: Yuxin Hu <yuxinhu@google.com>
|
|
|
06ae828f
|
2024-07-31T16:42:54
|
|
Vulkan: Avoid breaking render pass for vertex buffer conversion
Adreno driver does not support VK_FORMAT_FEATURE_VERTEX_BUFFER_BIT for
VK_FORMAT_R8G8B8A8_USCALED. This cases we hit fallback code path using
GPU to do conversion. The conversion buffer gets reused since all access
are from GPU, but that causes render pass break since there is a WAR
hazard on the conversion buffer. This CL adds the
isRenderPassStartedAndUsesBuffer check and allocate a new conversion
buffer if reuse current buffer may break the render pass.
Bug: b/356473483
Change-Id: Iaa0b9235ba42787f0e3629f0d9174ae768456f8b
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5754324
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Commit-Queue: Charlie Lao <cclao@google.com>
Reviewed-by: Yuxin Hu <yuxinhu@google.com>
|
|
|
59f6d962
|
2024-07-30T21:36:00
|
|
WebGPU: Run dEQP GLES2 tests.
Add the flags and platform support needed to initialze the WebGPU
backend.
Set up initial test expectations with broad suppressions.
Current pass rate: 14.27%
Bug: angleproject:356399840
Change-Id: I1339cd978ca75b59a2e99ee81d93981dc6066f36
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5751601
Commit-Queue: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
cc7d0220
|
2024-07-31T14:22:38
|
|
Vulkan: Fix serial mismatch during mid-loop flush
Currently, if the total buffer updates to the image surpasses a
certain threshold, it results in a flush. However, this can cause
discrepencies in the queue serial, which can result in incorrect
behavior on some platforms.
* Updated flushStagedUpdatesImpl() so that the image serial after
applying the updates matches that of the current outside command
buffer.
* That includes when there is a flush in the middle of the update
loop, resulting in submission and new queue serial for the CB.
* Added a unit test to check if a large texture can uploaded and
deleted after a second small texture is uploaded.
* Texture1UploadThenTexture2UploadThenTexture1Delete
* Added a unit test for flushing when uploading cubemap textures.
Bug: b/351650806
Bug: b/356192937
Change-Id: I7f9b20e4b7fd49115f22081a9733b4d44b740e4a
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5744377
Commit-Queue: Amirali Abdolrashidi <abdolrashidi@google.com>
Reviewed-by: Charlie Lao <cclao@google.com>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
26cdf621
|
2024-07-31T14:43:12
|
|
Vulkan: Bugfix in dynamic rendering support
Fix 2 bugs -
1. Set VkGraphicsPipelineCreateInfo::flags when linking GPLs
2. Chain VkRenderingFragmentShadingRateAttachmentInfoKHR struct in
VkRenderingInfo::pNext
Bug: angleproject:42267038
Change-Id: I456c75d57ec64926a21dffe93b05e1bbab2cb262
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5754635
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
7c77bb75
|
2024-07-24T12:05:29
|
|
Vulkan: Remove implicit buffer barrier for shader write
When app uses shaders to write to SSBO, right now we are inserting an
implicit barrier to ensure WAW are in order. But Spec says that
"Explicit synchronization is required to ensure that the effects of
buffer and texture data stores performed by shaders will be visible to
subsequent operations using the same objects". This CL removes the
implicit barrier for buffer write if the current write comes from
shaders and relies on explicit glMemoryBarrier to insert a global
barrier.
Bug: angleproject:350994515
Change-Id: I8ab039610be9be2ded27ea60dab54bdad08502f6
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5719258
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Commit-Queue: Charlie Lao <cclao@google.com>
Reviewed-by: Yuxin Hu <yuxinhu@google.com>
|
|
|
279c69e5
|
2024-07-30T21:30:54
|
|
WebGPU: Fix intended/actual format mismatch for textures
Update WebGPU texture allocation size to use the actual format instead
of intended. Early-out with an UNIMPLEMENTED if the format is not
supported yet.
Bug: angleproject:356399840
Bug: angleproject:344814096
Change-Id: Idef9ad40c6b1d490964fc398fb077b117612a44b
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5748448
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Commit-Queue: Geoff Lang <geofflang@chromium.org>
|
|
|
cab7d1b6
|
2024-07-30T21:19:15
|
|
WebGPU: Obey buffer alignment rules
WebGPU buffers require a size aligned to 4 bytes and map buffer offsets
aligned to 8 bytes. Hide this requirement in The BufferHelper class by
doing the apropriate offsets during buffer creation and map calls.
This allows many dEQP tests to run without crashing due to WebGPU
buffer creation errors.
Bug: angleproject:356399840
Bug: angleproject:42267091
Change-Id: Id71c3b5db31aa712d5a88631efa4897d6205a41d
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5748443
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Commit-Queue: Geoff Lang <geofflang@chromium.org>
|
|
|
641bc04a
|
2024-07-30T21:25:44
|
|
WebGPU: Use the correct source format for ReadPixels
ReadPixels was using the format of the destination buffer which works
correctly for RGBA8 -> RGBA/UNSIGNED_BYTE but runs the wrong conversion
functions for other formats.
Bug: angleproject:356399840
Bug: angleproject:42267090
Change-Id: I7c46add3f643017f454136135afef697847a2383
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5748446
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Commit-Queue: Geoff Lang <geofflang@chromium.org>
|
|
|
9da7fb15
|
2024-07-30T21:21:45
|
|
WebGPU: Don't log all renderpass closure reasons
This logging line spams the console excessively. For now, just make
sure there is a message associated with each renderpass closure reason.
Bug: angleproject:356399840
Change-Id: I2313d439837ffb11f11cd56194511e80fb312eba
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5748444
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Commit-Queue: Geoff Lang <geofflang@chromium.org>
|
|
|
1953c606
|
2024-07-30T21:28:13
|
|
WebGPU: Fix ReadPixels from window surfaces.
The intended/actual formats were not initialized in this code path and
the created swap chain texture did not have the usage flags needed for
copying to a buffer for ReadPixels.
Bug: angleproject:356399840
Bug: angleproject:42267090
Change-Id: I8e4279eedb2b23f0d2ed690ccee086ae115969c3
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5748447
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Commit-Queue: Geoff Lang <geofflang@chromium.org>
|
|
|
0898dab0
|
2024-07-30T21:34:00
|
|
WebGPU: Skip primitive topology conversion for DrawArrays
For non-indexed draw calls, no index format is needed for the
stripIndexFormat.
Bug: angleproject:356399840
Change-Id: Ie2c9805a2c257d94ea5f1bca026287873e4d6781
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5748449
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Commit-Queue: Geoff Lang <geofflang@chromium.org>
|
|
|
8446bbd6
|
2024-07-30T21:22:50
|
|
WebGPU: Remove ASSERT in depth/stencil clears
Instead of asserting that we are not clearing depth or stencil, just
log an UNIMPLEMENTED and skip the call. This allows many dEQP tests to
run without crashing on the ASSERT.
Bug: angleproject:356399840
Change-Id: Icfe7a77d7128e679501da42023601217c7b0e6c8
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5748445
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Commit-Queue: Geoff Lang <geofflang@chromium.org>
|
|
|
aee59926
|
2024-06-28T17:00:56
|
|
Metal: Enable GL_ANGLE_reverse_row_order for Metal
The feature is already implemented due to generic pack parameters
handling and FramebufferMtl::setFlipY().
Bug: angleproject:349980052
Change-Id: I08cc81dcb26ae81e8d81b060d986e01479487dbe
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5667597
Reviewed-by: Quyen Le <lehoangquyen@chromium.org>
Commit-Queue: Kimmo Kinnunen <kkinnunen@apple.com>
|
|
|
65f08d08
|
2024-07-26T22:14:49
|
|
Vulkan: Work around driver bug with nested switch
Bug: chromium:350528343
Change-Id: Ie7bd58934ccb2b8f06f6ad6a8c3bf38e81b84969
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5744620
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
|
|
4c883359
|
2024-07-03T13:50:58
|
|
Add dma_buf build support to generate_android_bp
Minimized version of go/anglecl/5729843 that limits introduced changes.
Make generate_android_bp.py emit a `angle_android_dma_buf` Soong
config variable that, when enabled, creates a "Linux-like" build
using Android toolchain.
The build does not assume the presence of regular Android graphics
stack. eglGetDisplay defaults to creating DisplayVkOffscreen. This
enables EGL rendering context setup based on an externally allocated
DMA-BUF, using EGL_EXT_image_dma_buf_import and
GL_OES_EGL_image_external extensions.
See ag/28533701 to view the generated code.
Also, revert changes from "Add build flag for a "dma_buf" Android build"
(commit 4d2b365582669a04224d1e1fa3c92447c359a4ba).
The flag added there was submitted a bit prematurely, and as
go/anglecl/5729843 was deemed too complex a different approach was
necessary.
Test: CQ
Test: Build the entire CL chain with the new Soong config flag enabled
Test: run the system, attach debugger to eglGetDisplay(), see
Test: DisplayVkOffscreen getting instantiated
Bug: b/353262025
Change-Id: I974d91cff79a0bd8b5059a145f0ae6114b2bcbe5
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5746214
Reviewed-by: Roman Lavrov <romanl@google.com>
Commit-Queue: Roman Lavrov <romanl@google.com>
Auto-Submit: Marcin Radomski <dextero@google.com>
|
|
|
198de044
|
2024-07-30T15:44:03
|
|
Metal: fix wrong render encoder's serial assignment
mtl::RenderCommandEncoder object is created once and reused multiple
times. We should re-assign a new serial whenever we reuse it, instead of
assigning once in constructor.
bug: angleproject:40644888
Change-Id: I18120c91dddd5b279fcec7a9099833e97ec1c37c
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5746179
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Kenneth Russell <kbr@chromium.org>
Commit-Queue: Quyen Le <lehoangquyen@chromium.org>
|
|
|
2bc1519c
|
2024-07-28T05:16:07
|
|
Metal: fix base/max level change for mutable textures.
Previously, the following scenario would accidentally delete and
recreate the native texture storage for a non-immutable TextureMtl:
1. set a texture base/max level.
2. attach the texture to a FBO and clear it.
- native storage is created.
3. sample the texture.
- native storage is accidentally deleted and recreated.
This happened because front-end only notifies the backend about the
base/max level dirty bits at step 3. At this point, Metal backend
thought that the native storage created at step 2 is no longer valid and
attempted to recreate the storage.
This CL fixes the bug by checking whether the base/max level is still
compatible with the existing native texture storage. If it is then the
dirty bits result in no-op.
Bug: chromium:355605685
Change-Id: I1daac80a74219109631a101e18e944560606d926
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5741121
Reviewed-by: Kenneth Russell <kbr@chromium.org>
Commit-Queue: Quyen Le <lehoangquyen@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
|
|
|
bd3a3308
|
2024-07-22T14:03:20
|
|
Vulkan: Remove implicit image barrier for shader write
When app uses compute or fragment shader to write to an image and makes
multiple dispatchCompute or draw calls, right now we are inserting an
implicit barrier to ensure WAW is hazard free. But Spec says that
"Explicit synchronization is required to ensure that the effects of
buffer and texture data stores performed by shaders will be visible to
subsequent operations using the same objects". This CL records the bits
from the last glMemoryBarrier call and will skip the barrier calls in
ContextVk::updateActiveImages if there is no layout change, unless there
is requirement from prior glMemoryBarrier.
Bug: angleproject:350994515
Change-Id: I8bdeeb658993824369824aaa0f25cb4b6e3785f7
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5719024
Reviewed-by: Yuxin Hu <yuxinhu@google.com>
Commit-Queue: Charlie Lao <cclao@google.com>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
57202584
|
2024-07-26T13:07:44
|
|
Vulkan: Fix dispatch-after-closed-render-pass bug
Bug: b/355567160
Change-Id: I4bc6acec53a50330507bfadcc0a4c1093366aae6
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5741786
Reviewed-by: mohan maiya <m.maiya@samsung.com>
Reviewed-by: Yuxin Hu <yuxinhu@google.com>
Reviewed-by: Charlie Lao <cclao@google.com>
|
|
|
b20cd34d
|
2024-07-18T10:03:17
|
|
Add build flag for a "dma_buf" Android build
Add angle_android_dma_buf gn flag, defaulting to false. When
enabled:
- Enables a "linux-like" build of ANGLE while using the Android
toolchain, by making the OS detection logic define PLATFORM_OS_POSIX
but not PLATFORM_OS_ANDROID.
- Makes ANGLE default to using Vulkan backend, and
- Enables DMA-BUF based external image support.
This allows setting up GL framebuffers backed by DMA-BUFs, enabling
rendering without full Android graphics stack. See the bug for details.
Bug: b/353262025
Test: treehugger
Test: Build the entire CL chain with the new Soong config flag enabled
Test: run the system, attach debugger to eglGetDisplay(), see
Test: DisplayVkOffscreen getting instantiated
Change-Id: I584f9491b188f730f4ee9630173cc486ae7e8908
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5729841
Reviewed-by: Roman Lavrov <romanl@google.com>
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
Commit-Queue: Roman Lavrov <romanl@google.com>
|
|
|
a0a832de
|
2024-07-26T12:33:37
|
|
Revert "GL: Forward client-side arrays to the driver when possible"
This reverts commit a6c2b4346516f228054b5bdd754bbc6f3ba1cba7.
Reason for revert: Fails some Chrome tests on Android after rolling: https://chromium-review.googlesource.com/c/chromium/src/+/5742024
Original change's description:
> GL: Forward client-side arrays to the driver when possible
>
> The OpenGL dirver can handle client-side arrays when the context is
> OpenGL ES or a desktop GL compatability profile. When in these
> sitatuions, use the driver default VAO for all frontend context VAOs
> and forward client-side data directly to the driver.
>
> Bug: angleproject:355034868
> Change-Id: I21a4459c4f7db780b51441d76e63d17bf737c101
> Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5736058
> Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
> Reviewed-by: Vasiliy Telezhnikov <vasilyt@chromium.org>
> Commit-Queue: Geoff Lang <geofflang@chromium.org>
Bug: angleproject:355034868
Change-Id: I76f95d66ca277bcbb67300179e49287b433c1ede
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5742647
Bot-Commit: Rubber Stamper <rubber-stamper@appspot.gserviceaccount.com>
Auto-Submit: Geoff Lang <geofflang@chromium.org>
Commit-Queue: Rubber Stamper <rubber-stamper@appspot.gserviceaccount.com>
|
|
|
a6c2b434
|
2024-07-23T16:36:10
|
|
GL: Forward client-side arrays to the driver when possible
The OpenGL dirver can handle client-side arrays when the context is
OpenGL ES or a desktop GL compatability profile. When in these
sitatuions, use the driver default VAO for all frontend context VAOs
and forward client-side data directly to the driver.
Bug: angleproject:355034868
Change-Id: I21a4459c4f7db780b51441d76e63d17bf737c101
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5736058
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Vasiliy Telezhnikov <vasilyt@chromium.org>
Commit-Queue: Geoff Lang <geofflang@chromium.org>
|
|
|
4f498eaa
|
2024-07-24T14:31:23
|
|
WebGPU: Add more format support
This CL adds a new FormatTable class that initializes webgpu texture
and vertex formats. It also adds this class to the display so it can
be used in the ImageHelper.
This CL changes the previously hardcoded RGBA8 texture format that
was initially used when creating textures to use the format passed
into the methods.
Bug: angleproject:344814096
Change-Id: I768b85335329116496dbf721aac54d1137aaae9f
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5660397
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Commit-Queue: Liza Burakova <liza@chromium.org>
|
|
|
84e54d88
|
2024-07-22T14:39:22
|
|
WebGPU: Add command buffers and renderpass management
Add a CommandBuffer class to serialize and replay WebGPU commands. Only
Draw and SetPipeline are implemented in this patch.
Manage render pass begin and end events due to framebuffer changes and
swapping.
Handle the color mask dirty bits so that a non-zero color mask will be
used.
All togther, this is enough to draw a triangle using a hard-coded
shader without inputs.
Bug: angleproject:0
Change-Id: I0fbf0296563c02c7f0774ad4197b83f4c93c22bb
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5731594
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Liza Burakova <liza@chromium.org>
Commit-Queue: Geoff Lang <geofflang@chromium.org>
|
|
|
703c5960
|
2024-07-24T17:02:28
|
|
Vulkan: PPO must warmup pipeline cache according to GPL support
This commit caused regression:
https://chromium-review.googlesource.com/c/angle/angle/+/5366173
`ProgramPipelineVk::link()` uses Complete subset for warmup regardless
of the GPL support. However, `ContextVk::createGraphicsPipeline()` will
respect the GPL support. Before the above change, PPO was respecting
GPL for warmup.
This change removed `subset` parameter from the
`getPipelineCacheWarmUpTasks()` not only to avoid copy-paste logic, but
also to ensure consistency with `waitForGraphicsPostLinkTasks()`.
Bug: angleproject:8601
Change-Id: Iab5df1b55921649a8f98a34bb07d8e6a145bfd4d
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5738153
Reviewed-by: mohan maiya <m.maiya@samsung.com>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
fc381db4
|
2024-07-23T00:00:00
|
|
Metal: Treat link jobs as thread-safe
Fixed: angleproject:351165323
Change-Id: I2ad7b8bd4b89be7d98876287c641b86a33e1108e
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5735475
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Commit-Queue: Alexey Knyazev <lexa.knyazev@gmail.com>
|
|
b98c4d81
|
2024-07-23T09:09:37
|
|
FrameCapture: Add restore of VAO buffer binding
If the VAO was changed before the trace started, but no
buffers were bound during replay, the VAO buffer binding
was not properly reset.
Test: angle_trace_tests --gtest_filter=*shovel_knight_pocket_dungeon
Bug: b/354888763
Change-Id: I76f94e5d219b88c641528049034d92ee55161f56
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5734653
Commit-Queue: Mark Łobodziński <mark@lunarg.com>
Reviewed-by: Cody Northrop <cnorthrop@google.com>
|
|
|
6eda2557
|
2024-07-22T00:00:00
|
|
Metal: Pass an existing context to CompileMslTask
Bug: angleproject:351165323
Change-Id: Icd4a2c8f9e158dc1989d720dc8235ba09dc0d82a
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5736272
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Commit-Queue: Alexey Knyazev <lexa.knyazev@gmail.com>
|
|
|
7b0dc666
|
2024-07-19T15:39:42
|
|
Vulkan: Enable explicitlyEnablePerSampleShading for most vendors
Bug: b/267953710
Bug: angleproject:40096890
Tests: SampleMultisampleInterpolationTest.SampleMaskInPerSample*
Change-Id: I51cc393934e7fd24146ab5c94893bb3014c5c044
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5726628
Commit-Queue: mohan maiya <m.maiya@samsung.com>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
7691cea7
|
2024-07-22T13:46:14
|
|
Vulkan: Remove seamful cubemap emulation
Practically, the Vulkan backend is never expected to run on ES2
hardware. It _may_ for WebGL, but seamful cubemap emulation was
disabled for webgl anyway.
Bug: angleproject:354729454
Change-Id: Iafa20fbdbe232c4df4c777b12e7698ef7a87cf24
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5730143
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
Auto-Submit: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Yuxin Hu <yuxinhu@google.com>
Reviewed-by: Charlie Lao <cclao@google.com>
|
|
|
0d458614
|
2024-07-18T00:00:00
|
|
Vulkan: Fix PBO readbacks with small row length
Use CPU path when the row length is
smaller than the source area width.
Fixed: angleproject:354005999
Change-Id: I5c4686ca5387a98c6137868afb19c333aed8ac21
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5724591
Commit-Queue: Alexey Knyazev <lexa.knyazev@gmail.com>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
975674ea
|
2024-07-18T00:00:00
|
|
Metal: Fix PBO readback failures with small row length
Do not use blits when the row length is
smaller than the source area width to
avoid Metal validation failures.
Bug: angleproject:354005999
Change-Id: Ifa0f29a0d1d2fb3275ea254957c24f36807e6f66
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5724790
Commit-Queue: Alexey Knyazev <lexa.knyazev@gmail.com>
Reviewed-by: Quyen Le <lehoangquyen@chromium.org>
Reviewed-by: Kenneth Russell <kbr@chromium.org>
|
|
|
e37c79d4
|
2023-11-23T15:59:12
|
|
Fix DecompressBlob() when compressed data is invalid
Function uses `zlib_internal::GetGzipUncompressedSize()` to calculate
the `uncompressedSize`. This size comes directly from `compressedData`.
If source data is invalid or has different format, then calculated
size may be any value, including 0.
Zero `uncompressedSize` cases ASSERT() in `angle::MemoryBuffer::data()`
and may case UB in `zlib_internal::GzipUncompressHelper()`.
Bug: chromium:1485277
Change-Id: Iaa92d93a5a1b530b8686cfb544eb66075b0874e6
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5729953
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Commit-Queue: Igor Nazarov <i.nazarov@samsung.com>
Reviewed-by: Roman Lavrov <romanl@google.com>
|
|
|
85edb907
|
2024-07-17T19:09:09
|
|
Vulkan: Remove the adjustClearColorPrecision feature for new driver
The vulkan driver issue that this feature tries to workaround
is resolved in the new driver. Remove the feature workaround.
Bug: b/292282210
Change-Id: I541526e71d5312f1a23b5a415e96d5b966e77f2c
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5719652
Reviewed-by: Amirali Abdolrashidi <abdolrashidi@google.com>
Commit-Queue: Amirali Abdolrashidi <abdolrashidi@google.com>
Reviewed-by: Solti Ho <solti@google.com>
Auto-Submit: Yuxin Hu <yuxinhu@google.com>
|
|
93491ae3
|
2024-06-21T08:04:24
|
|
FrameCapture: Handle active texture in MEC
When restoring texture bindings and contents, we need to first
set the correct GL_ACTIVE_TEXTURE, otherwise it will use existing
active texture bindings and update the wrong textures.
Test: Grand Mountain Adventure MEC
Bug: angleproject:345851268
Change-Id: I7c6c4a918513d4216405efe51774fcce4b49b45e
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5647491
Commit-Queue: Cody Northrop <cnorthrop@google.com>
Reviewed-by: Mark Łobodziński <mark@lunarg.com>
|
|
|
0e598a0c
|
2024-07-17T09:49:19
|
|
FrameCapture: Unbind PIXEL_UNPACK_BUFFER for texture restore
When restoring textures, we need to ensure the PIXEL_UNPACK_BUFFER
is not bound, as that will influence where we pull data from.
Allow per-context reset to restore the correct binding.
Test: Grand Mountain Adventure MEC
Bug: angleproject:345851268
Change-Id: I636f3780f8454d04152ff7bf7ba15d039ee41872
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5718285
Commit-Queue: Cody Northrop <cnorthrop@google.com>
Reviewed-by: Mark Łobodziński <mark@lunarg.com>
|
|
|
98ff1c14
|
2024-07-17T09:45:03
|
|
FrameCapture: Improve buffer binding reset
We aren't correctly tracking and resetting buffer bindings, which are
per-context. To restore them, we need to track the bindings at the
beginning and mark them dirty. Before this we were doing it in
ResetReplayContextShared based on the object dirty state.
Test: Grand Mountain Adventure MEC
Bug: angleproject:345851268
Change-Id: Ic3db583c88bec9c680197f71ff11fdf69efdf630
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5610332
Reviewed-by: Mark Łobodziński <mark@lunarg.com>
Commit-Queue: Cody Northrop <cnorthrop@google.com>
|
|
|
86b22745
|
2024-07-17T19:05:51
|
|
Vulkan: Fix mPipelineCache is not initialized after terminate
Very old regression:
https://chromium-review.googlesource.com/c/angle/angle/+/1683807
Problem happens only if call `eglInitialize()` again on terminated
display without also calling `eglGetDisplay()`.
Note: calling `eglGetDisplay()` on terminated display will call
`setupDisplayPlatform()`, which will destroy old `mImplementation`.
Test: angle_end2end_tests --gtest_filter=EGLDisplayTest.InitializeDrawSwapTerminateLoop*
Bug: angleproject:3318
Change-Id: I94bc7eba97be2d9b194c1a22a37662a2836f60b4
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5717753
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
62d9f847
|
2024-07-17T13:27:07
|
|
Speed up UniformBuffer support on Metal
If the sizes are different but all fields match
then only padding remains. So, as long as the
buffer can safely handle the offset and size of
usage, allow drawing without converting the
uniform buffer.
Fixed: angleproject:348270693
Change-Id: I91196ca9839cf064b573d912f125ff4879a85b02
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5719193
Reviewed-by: Kenneth Russell <kbr@chromium.org>
Commit-Queue: Quyen Le <lehoangquyen@chromium.org>
Auto-Submit: Gregg Tavares <gman@chromium.org>
Reviewed-by: Quyen Le <lehoangquyen@chromium.org>
|
|
|
9968e98c
|
2024-07-16T00:00:00
|
|
Metal: Fix compute path for 16-bit PBO readbacks
Added 16-bit normalized pixel formats
to internal copy shaders.
Fixed and simplified floatToNormalized
usage for signed inputs.
Bug: angleproject:352700368
Change-Id: Icbb79381991c6621004d53706b97662d33c25cd8
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5717929
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Quyen Le <lehoangquyen@chromium.org>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
1db80b88
|
2024-07-10T12:47:42
|
|
Reland "Vulkan: Use VK_KHR_dynamic_rendering[_local_read]"
This is a reland of commit c379ff48043a47e444c388c45270db40d3172d50
Original change's description:
> Vulkan: Use VK_KHR_dynamic_rendering[_local_read]
>
> Bug: angleproject:42267038
> Change-Id: I1f4eb0f309992a9c1c287a69520dadf5eff23b26
> Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5637155
> Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
> Reviewed-by: Amirali Abdolrashidi <abdolrashidi@google.com>
> Reviewed-by: Charlie Lao <cclao@google.com>
Bug: angleproject:42267038
Change-Id: I083e6963b5421386695e49a9872edbb2016c9763
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5691342
Reviewed-by: Charlie Lao <cclao@google.com>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Amirali Abdolrashidi <abdolrashidi@google.com>
|
|
|
1f87cbc9
|
2024-07-15T13:07:35
|
|
Vulkan: Fix late-added resolve attachment tracking
Resolve attachments may be added after the fact to a render pass due to
glBlitFramebuffer or eglSwapBuffer. Previously, only the resolve image
views were tracked by the render pass, and otherwise the state tracking
(layout, content defined, etc) treated the resolve images as generically
written-to by the render pass.
As a result, the render pass was unable to finalize the layout of the
resolve images early. Optimizing the layout of the swapchain image when
the surface is multisampled for example was not done due to this issue.
In this change, when resolve attachments are added late, they are
tracked identically to when they are added at the beginning of the
render pass, fixing the issues described above.
Bug: angleproject:42265625
Bug: angleproject:42266019
Change-Id: I765560762bb8caf39ba1096fb028177201c082d7
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5707470
Reviewed-by: Charlie Lao <cclao@google.com>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
c4274d54
|
2024-04-29T11:47:34
|
|
WebGPU: Add pipeline creation and caching.
Bug: angleproject:342213825
Change-Id: I303f193d30fd6b9820efaefcae64e11042888009
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5497535
Reviewed-by: Liza Burakova <liza@chromium.org>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Commit-Queue: Geoff Lang <geofflang@chromium.org>
|
|
|
58a56ca4
|
2024-07-16T11:26:44
|
|
Vulkan: Restrict VkEvent to TBRs
VkEvent has much bigger overhead. Until we know that it helps desktop
GPUs, we restrict it to TBRs. Also enabled for SwiftShader so that we
get more test coverage in bots.
Bug: b/336844257
Change-Id: Ie1859a515c9969ca37f07eae0ff729dd934eb26e
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5714391
Commit-Queue: Charlie Lao <cclao@google.com>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
84c77437
|
2024-07-15T16:41:17
|
|
GL: Disable GL_KHR_blend_equation_advanced on various devices
Intel desktop OpenGL drivers fail tests for advanced blend. Skia also
disables usage of this extension on Intel drivers.
Disable on older Qualcomm and ARM devices following Skia's blocklists.
Bug: angleproject:42267098
Change-Id: I7642017d06e11ad0651fba6f1ec5cc3f589700ec
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5709950
Reviewed-by: John Stiles <johnstiles@google.com>
Commit-Queue: Geoff Lang <geofflang@chromium.org>
|
|
9880af58
|
2024-07-09T13:59:54
|
|
GL: Fix map input buffer out of range in streamAttributes
Early return when validating map range size for input attribute
buffer fails.
Bug: b/349653220
Change-Id: I75ac73c58abb469f0757978175fde53cd012df31
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5690279
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Commit-Queue: Shrek Shao <shrekshao@google.com>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
|
|
|
85875b96
|
2024-07-11T14:59:27
|
|
Fix validation regarding cube map array dims
When defining cube map array textures, the depth represents
the number of layer-faces, and does not change with mip level.
* Updated validation for glTexImage3D().
* The width and the height for a mip level should not exceed the
max cube map size for that mip level.
* The depth limit has been set to max 3D texture size (similar to
glTexStorage3D())
* The spec states that this limit should be at least the value of
the implementation-based constant MAX_RECTANGLE_TEXTURE_SIZE.
* Added unit tests to check multiple cases for cube map array
validation for TexStorage3D, TexImage3D, and TexSubImage3D.
Bug: b/338621235
Bug: angleproject:42262247
Change-Id: I0857e68de02238ffd288e34879fb7dfa6e3f545f
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5698571
Commit-Queue: Amirali Abdolrashidi <abdolrashidi@google.com>
Reviewed-by: Charlie Lao <cclao@google.com>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
a8eded77
|
2024-06-26T12:58:25
|
|
Vulkan: remove unused function
Bug: b/293297177
Change-Id: I9eeda38b78cc68eae3f65571b82b62e5ea45ad61
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5659504
Commit-Queue: Yuxin Hu <yuxinhu@google.com>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Charlie Lao <cclao@google.com>
|
|
06921c61
|
2024-07-08T12:13:40
|
|
vulkan: filter out devices with insufficient API version
We weren't taking the version into consideration inside
ChoosePhysicalDevice, but RendererVk really cares about that and will
error out if the selected device doesn't meet the minimum API version
requirements.
Frontload some of that API version check work in ChoosePhysicalDevice,
giving it the opportunity to find a device that matches the
requirements.
Bug: angleproject:351866412
Change-Id: I471e93b03eca5e18e98202f9848ba0fd1b55d5a8
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5686556
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Auto-Submit: Steven Noonan <steven@uplinklabs.net>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Amirali Abdolrashidi <abdolrashidi@google.com>
|
|
|
df90bbc5
|
2024-07-12T18:04:53
|
|
Refactoring: move angle::HashMap and HashSet to own header
Underlying abseil includes pull in a large set of headers
Bug: angleproject:42266508
Change-Id: Icee47143a8a59bb0795a054b67c0aa4ddcfca4d4
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5704137
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
a5161f01
|
2024-06-04T15:46:04
|
|
Vulkan: Enable event based image barrier
This CL enables VkEvent based image barrier for all GPUs. The testing
result on HK3 shows manhattan31_off score impoves ~5%. We can exclude
certain barrier insensitive GPUs if needed.
Bug: b/336844257
Change-Id: Iebc4957dbb683637c9314e706be31d3376ee15aa
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5634089
Commit-Queue: Charlie Lao <cclao@google.com>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
24322b7d
|
2024-06-28T20:10:34
|
|
Vulkan: feature for cached non-coherent for dyn/stream buffers
On platforms lacking cached coherent memory, ANGLE falls back to
non-cached coherent memory for dynamic/stream buffers. This impacts
CPU readback performance.
Add VK feature preferCachedNoncoherentForDynamicStreamBufferUsage. When
enabled, ANGLE prioritizes cached non-coherent memory for these
buffers.
Enable this feature for Intel Meteorlake SOCs.
Bug: b/347601787
Change-Id: If62af9f3df57c0bcebf18af747cac56e45f93ea7
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5667457
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Charlie Lao <cclao@google.com>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
8ed4d427
|
2024-07-11T00:00:00
|
|
Metal: Remove link tasks dependency on context
Bug: angleproject:351165323
Change-Id: If9e5d141cb15a6fc6df8a5fd9a2cfe70efe19dfc
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5701902
Commit-Queue: Alexey Knyazev <lexa.knyazev@gmail.com>
Reviewed-by: Kenneth Russell <kbr@chromium.org>
|
|
|
245c1243
|
2024-07-12T14:28:31
|
|
Revert "Disallow read color type conversions for norm16 formats"
This reverts commit e5b442dedd8030d64a6ca2cb34ba0ec37895adfa.
Reason for revert: Broke Skia tests; possibly not going for this any more, as discussed on https://issuetracker.google.com/351644552#comment11
Original change's description:
> Disallow read color type conversions for norm16 formats
>
> OpenGL ES is not supposed to convert 16-bit color buffers to
> 8-bit pixel types or vice versa during readPixels operation.
>
> Fixed: angleproject:351644552
> Change-Id: I9e51ddc82f62f958b983b5d4609f7e983a941e83
> Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5686233
> Commit-Queue: Alexey Knyazev <lexa.knyazev@gmail.com>
> Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Bug: angleproject:351644552
Change-Id: Ib0222e4cc5eae944db96fdd3f72c8980dfe09adf
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5702736
Reviewed-by: Solti Ho <solti@google.com>
Bot-Commit: Rubber Stamper <rubber-stamper@appspot.gserviceaccount.com>
|
|
|
c5ab1ceb
|
2024-07-11T00:00:00
|
|
Metal: Remove library creation dependency on ContextDevice
Bug: angleproject:351165323
Change-Id: Idc560a75e569fe917685bb247625d7df8472af46
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5698885
Reviewed-by: Kenneth Russell <kbr@chromium.org>
Commit-Queue: Alexey Knyazev <lexa.knyazev@gmail.com>
|
|
|
a93476a1
|
2024-07-11T00:00:00
|
|
D3D11: Pass 3D slice index to Image11::copyWithoutConversion
3D slice range is passed in a D3D11_BOX
structure for CopySubresourceRegion.
Fixed: angleproject:352467349
Bug: angleproject:352496176
Change-Id: Ie96d9567cf73b0612bfb1cbd7fbacc6c0f01dec4
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5697425
Commit-Queue: Alexey Knyazev <lexa.knyazev@gmail.com>
Reviewed-by: Kenneth Russell <kbr@chromium.org>
|
|
|
6578b9c0
|
2024-07-09T17:19:47
|
|
Vulkan: Exclude compute/preFrag only access images from event
This further restricts VkEvent usage for certain usage patterns. If
image is only used by compute, use VkEvent also will not benefit it
since compute itself can not overlap with compute (assume there is only
one compute engine and compute work can not overlap with each other).
Similarly this also applies to KPreFragment stages. Basically after this
CL, use of VkEvent is limited to usages that crosses different execution
units (modeled against tiler based GPUs where there are pre-fragment
stages and fragment stages and compute and all others). Before this CL,
we are seeing performance regression with antutu_refinery and
streets_of_rage_4 due to overhead of VkEvent, which is fixed with this
CL.
Bug: b/336844257
Change-Id: I5ca5d813daefe9bfcaf48f831340cdf9559f8104
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5692760
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Amirali Abdolrashidi <abdolrashidi@google.com>
Commit-Queue: Charlie Lao <cclao@google.com>
|
|
|
402c8ccd
|
2024-06-26T19:28:21
|
|
Vulkan: Limit VkEvent for images that has fragment access only
One of the problem with VkEvent is that the overhead comes with
VkCmdSetEvent causes some app traces regress performance. The goal in
this CL is to further limit VkCmdSetEvent to images that that we think
are potentially subject to the pipeline bubble. The bubble usually
occurs when accesses are alternated between different stages,
specifically a mix between vertex/transfer/compute/fragment. If all
accesses are from fragment shader or color attachment, then use VkEvent
will not be beneficial, but only adds extra overhead. This CL adds the
heuristic tracking for image access. Every time an image is used, a bit
is used to indicate the usage involves fragment only or not. A bitfield
is used to track the window of the history of the usage. When image is
used (usually at the time queueSerial is set), we shift the history bits
left and the new bit is added to the right most bit. If all accesses are
from the fragment shader or color attachment, then no need to use
VkEvent. For example, if a texture is always sample from fragment shader
only, then VkEvent will not used. Another common usage is you render to
it and then texture from it, it will also excluded from VkEvent with
this CL.
Bug: b/336844257
Change-Id: I175194f30b8f1d9b8fbf38ad594778474548016f
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5664170
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Amirali Abdolrashidi <abdolrashidi@google.com>
Commit-Queue: Charlie Lao <cclao@google.com>
|
|
|
eda67d7e
|
2024-07-11T11:16:08
|
|
Avoid including fstream from angleutils.h
SaveFileHelper is mostly useful for / specific to FrameCapture,
so move it there.
The other couple of uses just write a string.
Bug: angleproject:42266508
Change-Id: Ia1dcd4531f9d5671f40611a1887dcfe7c5dbc1ef
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5696025
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Commit-Queue: Roman Lavrov <romanl@google.com>
|
|
|
87e9ee9c
|
2024-07-08T14:07:05
|
|
egl: add missing extension string for D3D LUID
This extension was never announcing itself.
Bug: angleproject:351866412
Change-Id: I9a6e4aae1ea3a8aba49ed036cb7cc14ec1a9151f
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5686555
Commit-Queue: Kenneth Russell <kbr@chromium.org>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Kenneth Russell <kbr@chromium.org>
Auto-Submit: Steven Noonan <steven@uplinklabs.net>
|
|
|
584fbcee
|
2024-07-10T12:43:34
|
|
Vulkan: Rework swap-time barrier logic
Avoids unnecessary transitions when overlay is enabled
Bug: angleproject:42267038
Change-Id: I0534911c0142c5e94cf3be112283fb98fcde0f6c
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5691346
Reviewed-by: Amirali Abdolrashidi <abdolrashidi@google.com>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Charlie Lao <cclao@google.com>
|