|
3adcf1e4
|
2019-01-16T00:40:38
|
|
Revert "Don't call Texture::onDestroy twice for the zero textures."
This reverts commit d3e0e84ce7e1fcde7bbe71ebfba009ddd5942de3.
Reason for revert: likely cause for crbug.com/922296
BUG=chromium:922296
Original change's description:
> Don't call Texture::onDestroy twice for the zero textures.
>
> The BindingPointer will call onDestroy when all references are removed, no
> need to do it explicitly.
>
> BUG=angleproject:3070
>
> Change-Id: Ic5dd4da8f73e204d26d9d3a3c6e1eff9fded5411
> Reviewed-on: https://chromium-review.googlesource.com/c/1409868
> Reviewed-by: Jamie Madill <jmadill@chromium.org>
> Commit-Queue: Geoff Lang <geofflang@chromium.org>
TBR=geofflang@chromium.org,jmadill@chromium.org
Change-Id: I38097218d49940e7ac3ad914714294b019c154fd
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: angleproject:3070
Reviewed-on: https://chromium-review.googlesource.com/c/1413239
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
|
|
7098cff5
|
2019-01-15T18:33:42
|
|
Make BUILD.gn targets into templates.
This allows us to "globally" add and remove certain configs as long as
we use the new templates. This simplifies the logic of adding configs
for stuff like extra warnings and default include dirs. As well it
simplifies removing certain common unwanted configs.
Generally simplifies the logic in BUILD.gn. Will allow for easily
suppressing the clang-plugins config instead of using a global setting
in .gn. Then we can enable the additional warnings config-by-config.
Also fixes some warnings that turned up after we enabled the extra
warnings config in our tests. Also moves the dEQP tests main to be
consistent with the other test main files.
Bug: angleproject:3069
Change-Id: I5a8166cd0f5a7926822c171fcaf473fc86b3ffc1
Reviewed-on: https://chromium-review.googlesource.com/c/1409871
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
|
|
d3e0e84c
|
2019-01-14T15:59:18
|
|
Don't call Texture::onDestroy twice for the zero textures.
The BindingPointer will call onDestroy when all references are removed, no
need to do it explicitly.
BUG=angleproject:3070
Change-Id: Ic5dd4da8f73e204d26d9d3a3c6e1eff9fded5411
Reviewed-on: https://chromium-review.googlesource.com/c/1409868
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Geoff Lang <geofflang@chromium.org>
|
|
f3179a6a
|
2018-07-12T16:22:06
|
|
ES31: Implement bindImageTexture binds a single layer on D3D backend
Dynamically generate image2D variables' declaration and function
definition in libANGLE.
Bug: angleproject:1987
TEST=angle_end2end_tests.ComputeShaderTest.*
Change-Id: Idacc756f7bd15f22eccb1d689e18e997f3e74159
Reviewed-on: https://chromium-review.googlesource.com/c/1142885
Commit-Queue: Xinghua Cao <xinghua.cao@intel.com>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
|
|
15460e36
|
2019-01-03T23:38:44
|
|
In PixelTransfer11, center offset should be 1/size, not 1/(size-1).
Bug: angleproject:3039
Change-Id: Ie97bfb5aa3cfe7cd2c4b1d22b5b4a8d82a5e5170
Reviewed-on: https://chromium-review.googlesource.com/c/1407251
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>
|
|
c2ee2ccb
|
2019-01-11T10:39:15
|
|
Vulkan: Don't support EGL_SWAP_BEHAVIOR_PRESERVED_BIT
dEQP-EGL.functional.preserve_swap.preserve.no_read_before_swap#no_draw
dEQP-EGL.functional.preserve_swap.preserve.no_read_before_swap#post_clear
dEQP-EGL.functional.preserve_swap.preserve.no_read_before_swap#post_render
dEQP-EGL.functional.preserve_swap.preserve.no_read_before_swap#pre_clear
dEQP-EGL.functional.preserve_swap.preserve.no_read_before_swap#pre_clear_post_clear
dEQP-EGL.functional.preserve_swap.preserve.no_read_before_swap#pre_clear_post_render
dEQP-EGL.functional.preserve_swap.preserve.no_read_before_swap#pre_render
dEQP-EGL.functional.preserve_swap.preserve.no_read_before_swap#pre_render_post_clear
dEQP-EGL.functional.preserve_swap.preserve.no_read_before_swap#pre_render_post_render
Bug: angleproject:3051 angleproject:2716
Change-Id: I3607f45cd2f447fb0a95f91ffd9aa93000dfa18c
Reviewed-on: https://chromium-review.googlesource.com/c/1407248
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Tim Van Patten <timvp@google.com>
|
|
|
9567b0c5
|
2019-01-14T13:45:58
|
|
Clear VAO dirty bits in NULL back-end.
This is the back-end responsibility now.
Bug: chromium:921338
Change-Id: I6181210a80ec7d635382df74d842d8a899ad2894
Reviewed-on: https://chromium-review.googlesource.com/c/1409405
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>
|
|
|
17a50e17
|
2019-01-10T22:05:43
|
|
Vulkan: Enable robust buffer access.
This is not a complete implementation because it does not have the
ability to disable the physical device feature. It does pass the
current set of robust access tests. But there could be a performance
regression on platforms that have a slower impelmentation. We would
want the ability to support cases with bufer robustness on or off.
Bug: angleproject:3062
Change-Id: I7d6eb889debcbd32f6ed809b526677123f872726
Reviewed-on: https://chromium-review.googlesource.com/c/1403967
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>
|
|
|
c759b8b4
|
2019-01-03T15:16:50
|
|
Vulkan: More Vertex Array optimizations.
Inlines a number of Vulkan vertex array methods.
Also changes the way vertex buffers are bound. Note that Vulkan doesn't
support NULL buffer bindings. Thus we create an emulated NULL buffer
to work around the problem of having gaps in the bound vertex buffers.
This allows us to use a single bind call for ranges of vertex buffers
even when there are gaps.
Also changes how vertex array dirty bits are reset. Instead of calling
memset to clear the affected buffers we pass a mutable pointer to the
Vertex Array sync state. This allows us to only reset the dirty bits
that we sync. This saves on the memory clearing time.
Improves perf by about 10% in the Vulkan VBO state change test.
Bug: angleproject:3014
Change-Id: Ib7b742dff7897fc891606a652ea0b64255a24c86
Reviewed-on: https://chromium-review.googlesource.com/c/1390360
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
|
|
|
633d5e69
|
2018-12-23T19:58:01
|
|
Vulkan: Put viewport and scissor back in pipeline desc.
Turns out this is much faster than using dynamic state. When we
support multiple viewports it might be easier to use dynamic
state since we won't need to make an overly large pipeline
description. We could support both methods using a flag to
indicate the viewport and/or scissor regions are invalid.
Until then we can remove the pipeline and scissor dirty bits.
Improves perf by about 15% in the Vulkan VBO state change test.
Bug: angleproject:3013
Change-Id: I4572250626a9a0f0ca3451b17e8f0de186416cae
Reviewed-on: https://chromium-review.googlesource.com/c/1390359
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
5bca4fed
|
2019-01-09T17:07:06
|
|
Vulkan: Switch to EXT_debug_utils
EXT_debug_utils is a (relatively) new extension that subsumes
EXT_debug_report and EXT_debug_marker extensions.
This change uses EXT_debut_utils (if available) to receive debug
messages, keeping EXT_debug_report as fallback.
Bug: angleproject:2798
Change-Id: Ia6767604548704e18f8150a54d2533c0b1197841
Reviewed-on: https://chromium-review.googlesource.com/c/1403955
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
|
|
|
2450d8dd
|
2019-01-11T14:41:04
|
|
Vulkan: Squash Vertex Pipeline info.
Instead of using 8 bytes per vertex we can reduce the space used for
the divisor to 8 bytes. For larger values than 255 we can emulate the
divisor by unrolling the draw call. We will likely need to do this
in any case for instanced draws when the instancing extension isn't
available.
The tighter packing will allow for us to move the viewport and scissor
back into the pipeline description. It seems this is much faster than
using dynamic state. Every state change that would pull in a new
Pipeline would need the viewport and scissor re-applied. It seems
these driver calls are costly.
Does not improve perf significantly but enables future improvements.
Bug: angleproject:3013
Change-Id: I1a41c3acadc6fbd47c7a7b961c706e82f78de936
Reviewed-on: https://chromium-review.googlesource.com/c/1390358
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
|
|
a3b2e71f
|
2018-12-08T16:46:48
|
|
ES31: Support for GL_BUFFER_DATA_SIZE on GL_ATOMIC_COUNTER_BUFFER in D3D
This commit adds support for querying the GL_BUFFER_DATA_SIZE of
GL_ATOMIC_COUNTER_BUFFER in the D3D renderer.
Bug: angleproject:1729
Test: angle_end2end_tests
Change-Id: Id6aae0d92c5e0960b2b245ba7d83970b04ba4eed
Reviewed-on: https://chromium-review.googlesource.com/c/1399143
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
|
|
|
3f0c4a56
|
2019-01-10T10:20:35
|
|
Vulkan: Faster state transitions.
Implements a transition table from Pipeline Cache entry to
state change neighbouring Pipeline Cache entries. We use
a 64-bit mask to do a quick scan over the pipeline desc.
This ends up being a lot faster than doing a full hash
and memcmp over the pipeline description.
Note that there could be future optimizations to this design.
We might keep a hash map of the pipeline transitions instead
of a list. Or use a sorted list. This could speed up the search
when there are many transitions for cache entries. Also we could
skip the transition table and opt to do a full hash when there
are more than a configurable number of dirty states. This might
be a bit faster in some cases. Likely this will be something we
can add performance tests for in the future.
Documentation is also added in a README file for the Vulkan back
end. This will be extended over time.
Improves performance about 30-35% on the VBO state change test.
Bug: angleproject:3013
Change-Id: I793f9e3efd8887acf00ad60e4ac2502a54c95dee
Reviewed-on: https://chromium-review.googlesource.com/c/1369287
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
|
|
|
80766cfa
|
2019-01-10T10:20:34
|
|
Vulkan: Update individual VAO attribs when dirty.
This simplifies the dirty state updates in VertexArrayVk. It also lets
us use a smaller dirty bit mask when a single attribute is marked dirty
in a vertex array.
Improves performance by about 1-2% in the VBO state change test. Will
allow for better performance using a pipeline transition table.
Bug: angleproject:3013
Change-Id: I25c5172b3f41b7abac6b8273c8f9cd42eb46cc9f
Reviewed-on: https://chromium-review.googlesource.com/c/1403958
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>
|
|
|
b827f490
|
2018-12-14T12:09:41
|
|
GL: Work around errors generated during query caps initialization.
Nexus5X drivers generate INVLAID_ENUM errors when querying
GL_QUERY_COUNTER_BITS.
Add device detection from renderer string.
BUG=angleproject:3027
Change-Id: I367e20c79e1c4e53c26d94603d9a893604b51165
Reviewed-on: https://chromium-review.googlesource.com/c/1374274
Commit-Queue: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
|
|
|
b5ba549a
|
2019-01-02T15:19:22
|
|
Vulkan: Shader path for texture copy when image is not initialized
This change implements staging image/texture copies when the destination
image is not yet fully initialized. With this change, CPU readback for
glCopyTex[Sub]Image2D and glCopy[Sub]TextureCHROMIUM should happen only
if the texture formats don't allow a fragment-shader based copy.
Bug: angleproject:2958
Change-Id: I04087e14ea8fb6fbc731598c5493e44651c22c01
Reviewed-on: https://chromium-review.googlesource.com/c/1393909
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
|
|
|
4f3b207d
|
2019-01-01T14:48:25
|
|
Vulkan: Shader path for texture-to-texture copy
This change implements glCopy[Sub]TextureCHROMIUM in GPU. As with the
previous change implementing glCopyTex[Sub]Image2D, it currently only
selects the shader path if the texture is already defined.
Bug: angleproject:2958
Change-Id: Ia1b5625f92e6c9f91807c9b601e5c34d2d5e5c30
Reviewed-on: https://chromium-review.googlesource.com/c/1392394
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
|
|
|
29b49417
|
2019-01-07T14:03:06
|
|
Make copy texture test more extensive
By doing the copy multiple times, we exercise both paths where the
destination is already initialized and when it's not.
This adds tests for all combinations of formats and flags.
Bug: angleproject:2958
Change-Id: I56afb44496acd1b4d5a8527f4dbee29afbac9c81
Reviewed-on: https://chromium-review.googlesource.com/c/1398643
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
|
|
|
eb278cb0
|
2019-01-07T13:49:36
|
|
D3D: Log GetLastError when compiler load fails.
Also use the non-unicode version of LoadLibrary explicitly.
Bug: chromium:919163
Change-Id: I4841c3eef586ff57563915da12765baaa6e0b146
Reviewed-on: https://chromium-review.googlesource.com/c/1398642
Reviewed-by: Bruce Dawson <brucedawson@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>
|
|
5df1d281
|
2019-01-04T14:22:59
|
|
Add std430 support (continue)
This patch enables that the interface block information can be correctly
collected in API side for std430 layout. So we can get right offset value when
we use glGetProgramResourceiv to query.
BUG=angleproject:1920
Change-Id: Ib936f6e25936c07c5bbc29f6b567d282a0f257c2
Reviewed-on: https://chromium-review.googlesource.com/c/1345891
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Jiajia Qin <jiajia.qin@intel.com>
|
|
|
dbc605ce
|
2019-01-04T16:39:14
|
|
Vulkan: Optimize VBO state changes.
Also has some minor optimizations for the front-end.
12% improvement on the Vulkan VBO change test.
Bug: angleproject:3014
Change-Id: I38e1a8194edfc14bfe57424be348cb9688e928f4
Reviewed-on: https://chromium-review.googlesource.com/c/1369286
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>
|
|
e4109f27
|
2018-12-13T16:25:53
|
|
WebGL: validate texture format matches sampler type
WebGL requires that drawing produces INVALID_OPERATION if a texture's
format doesn't match the sampler type it is bound to. This is a little
confusing because samplers have two attributes that could be called
"type": addressing mode (2D/3D/Cube), and component format
(float/signed/unsigned/shadow). ANGLE already handled checking the
addressing mode; this change adds checking for the component format.
Fixes WebGL conformance test
conformance2/uniforms/incompatible-texture-type-for-sampler.html
Bug: chromium:809237
Change-Id: I52ebfecd92625e3ee10274cb5f548d7e53de72dd
Reviewed-on: https://chromium-review.googlesource.com/c/1377611
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
Commit-Queue: James Darpinian <jdarpinian@chromium.org>
|
|
|
1f56ed2a
|
2019-01-03T15:24:22
|
|
Add WGLWindow and WGL test configs.
WGLWindow lets us use a Windows driver's bindings instead of ANGLE.
This only works if the underlying driver supports OpenGL ES
compatibility.
Also adds the WGL headers, WGL XML, and a specialized WGL loader.
Because of a small driver issue with NVIDIA I added a retry for the WGL
Window initialization.
Bug: angleproject:2995
Change-Id: Ie5148ece470dd03df33015f4919ad1fa79a859ec
Reviewed-on: https://chromium-review.googlesource.com/c/1366021
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
|
|
|
4e712be2
|
2019-01-03T13:53:59
|
|
Refactor BlockLayoutEncoder APIs for std430.
This splits HLSL SSBO access into two steps.
First we compute a mapping from the collected SSBO variable names to
TField pointers. Then during tree traversal we use a block encoding
visitor class that uses the shader names to store BlockMemberInfo
structures for the structures and variables. Each nested structure
is traversed separately so that the BlockMemberInfo offsets are
relative to the structure start rather than the enclosing block. The
array stride for a structure is the size of the struct after all the
alignment is included.
This gives the correct results for the SSBO access chain in the HLSL
code. It also will allow us to use the same encoding and visiting logic
for SSBOs on the API side.
Bug: angleproject:3024
Change-Id: I42b1db0e7547782ae77fe5f64a797f803f203f45
Reviewed-on: https://chromium-review.googlesource.com/c/1352731
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
|
|
|
f83a28a6
|
2018-12-09T03:48:34
|
|
Vulkan: Shader path for framebuffer-to-texture copy
Part 1 in a series of changes to perform image copies on the GPU.
Bug: angleproject:2958
Change-Id: I6264a880865c4738c0866f2dc71af63425fc4118
Reviewed-on: https://chromium-review.googlesource.com/c/1370724
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
|
|
|
d7969cdd
|
2019-01-02T11:53:18
|
|
Vulkan: optimize image layout transitions
Removes dstAccessMask being redundantly set where
`GetBasicLayoutAccessFlags()` sets the same bits. Additionally, removes
setting HOST_WRITE_BIT and TRANSFER_WRITE_BIT as src access mask if new
layout is SHADER_READ_ONLY_OPTIMAL. The correct src access mask will be
set based on the previous layout.
Additionally, specializes `GetBasicLayoutAccessFlags()` in
`GetSrcLayoutAccessFlags()` and `GetDstLayoutAccessFlags()` where the
access mask is set correctly (and optimally) based on whether the layout
is the old or the new one. This removes a few unnecessary access masks
from src (for WAR hazards), and adds a few missing access masks to dst
(for RAW hazards), such as VK_ACCESS_COLOR_ATTACHMENT_READ_BIT necessary
for blending.
Bug: angleproject:2958
Change-Id: I5870bc99c755f0444332418f998032850825aca5
Reviewed-on: https://chromium-review.googlesource.com/c/1392397
Reviewed-by: Tobin Ehlis <tobine@google.com>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
9fa54eab
|
2019-01-02T18:38:33
|
|
Optimize and inline more validation.
Previously on each draw call we were checking both the instanced and
non-instanced attributes for overflow. On on-instanced draw calls we
don't need to do any special checks for instanced attributes. Also
we can inline more into the headers. This improves draw call
performance slightly.
Bug: angleproject:2966
Change-Id: Idf5861d2d9daf9fffd5c84f6a1ea5b23ac8ab713
Reviewed-on: https://chromium-review.googlesource.com/c/1390357
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Markus Tavenrath <matavenrath@nvidia.com>
|
|
|
4627b370
|
2019-01-02T18:38:32
|
|
Add gl null driver tests to DrawElements.
Requires a couple tweaks to the back-end to be compatible with the
null driver.
Bug: angleproject:2966
Change-Id: Ia83651aabb0dd14b7d6f64152c276463a8fb28fc
Reviewed-on: https://chromium-review.googlesource.com/c/1392392
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>
|
|
48bbc53e
|
2018-12-03T16:17:00
|
|
ES31: Fix for reading back initialized SSBO buffers in D3D11
If a SSBO is initialized with data, a staging buffer was created. When the
data was read back, it was read from the staging buffer instead of the GPU.
This patch fixes that by making the UAV buffer the latest buffer.
Bug: angleproject:2990
Test: angle_end2end_tests
Change-Id: I43c8c85144c74cc9b317577b7c834e3f7e347f72
Reviewed-on: https://chromium-review.googlesource.com/c/1359712
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Bryan Bernhart <bryan.bernhart@intel.com>
Commit-Queue: Bryan Bernhart <bryan.bernhart@intel.com>
Commit-Queue: Jamie Madill <jmadill@chromium.org>
|
|
|
04ea03e4
|
2019-01-01T15:14:25
|
|
Make default block member info a constant.
Instead of generating a struct each time we can use a constexpr
constructor to use a single representation.
Bug: angleproject:3024
Change-Id: I14dec65a4f6ac9ab2f7e7af444862e4ceab88d8c
Reviewed-on: https://chromium-review.googlesource.com/c/1392395
Reviewed-by: Jiajia Qin <jiajia.qin@intel.com>
Commit-Queue: Jamie Madill <jmadill@chromium.org>
|
|
|
76f66954
|
2018-12-18T17:59:00
|
|
ES31: Fix the bug that SSBO is active but not statically used
If a SSBO is active but not statically used, there will be no UAV slot is
bound. So we should skip this kind of block.
Bug: angleproject:1951
Change-Id: I4d813ddefcce6c31fa02701f26148eb21c00f380
Reviewed-on: https://chromium-review.googlesource.com/c/1381847
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Jiajia Qin <jiajia.qin@intel.com>
|
|
|
0c667215
|
2019-01-01T14:40:36
|
|
Pass ErrorSet to ContextImpl constructor.
This removes the need for the setErrorSet method.
Also update some egl::Error TODO bugs.
Bug: angleproject:2491
Change-Id: I0aba07c4a53b579835a88c3dacae294f752e6b17
Reviewed-on: https://chromium-review.googlesource.com/c/1392393
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>
|
|
|
c3dc5d48
|
2018-12-30T12:12:04
|
|
Merge gl::Context and gl::ContextState.
This reduces the number of indrections when accessing the Extensions
or Caps structures. It will provide a small speed-up to some methods.
It also cleans up the code.
Bug: angleproject:2966
Change-Id: Idddac70758c42c1c2b75c885d0cacc8a5c458685
Reviewed-on: https://chromium-review.googlesource.com/c/1392391
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Markus Tavenrath <matavenrath@nvidia.com>
|
|
957eb07f
|
2019-01-01T12:36:48
|
|
Remove redundant masking of the bits in angle::BitSet when using ternary operators.
Bug: angleproject:3040
Change-Id: I59e69ddb27ec9c5723aff04858ac5c052e2775da
Reviewed-on: https://chromium-review.googlesource.com/c/1392806
Commit-Queue: Markus Tavenrath <matavenrath@nvidia.com>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
|
|
|
cc37cbf2
|
2018-12-30T23:35:25
|
|
Apply several small micro-optimizations to ValidateDrawElementsCommon
* Remove indirection when getting CotextState from Context
* Reduce number of branches from 2 to 1 when checking for an InvalidEnum in FromGLEnum<DrawElementsType>
* Provide IndexRange constructor which doesn't initialize the member variables
* Reduce number of branches to 1 when checking for BasicDrawStateErrors for the fast path
* Remove a few branches and reduce math cost during the buffer range check in ValidateDrawElementsCommon
Bug: angleproject:2966
Change-Id: Ibdbed8a59ad9d7abce76622e9f507498e67ab997
Reviewed-on: https://chromium-review.googlesource.com/c/1392805
Commit-Queue: Markus Tavenrath <matavenrath@nvidia.com>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
|
|
|
60a50cfc
|
2018-12-29T16:04:05
|
|
Inline Context::DrawArrays and Context::BindBuffer.
This reduces the call depth of both of these two hotspots.
Reduces time spent in the CPU overhead benchmarks by about 10%.
Bug: angleproject:2966
Change-Id: I5052e56dcc1dfb80274326a7f0891fafba7d6655
Reviewed-on: https://chromium-review.googlesource.com/c/1392389
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Markus Tavenrath <matavenrath@nvidia.com>
|
|
|
8c78ce4b
|
2018-12-16T19:14:58
|
|
Use visitor pattern for Shader Variable APIs.
In many places in ANGLE we need to traverse a ShaderVariable tree. We
do this to store uniform offset and other information, to flatten the
tree of uniforms for the front-end, or to produce active variable lists
for uniform and shader storage blocks. In each case, we would write
separate tree traversal code.
This patch introduces a shared visitor pattern for all of the shader
variable tree traversal instances. With that get more common code. Also
it is easier to write new variable traversals. ProgramD3D and
ProgramLinkedResources in particular get nice simplificiations.
The visitor object recieves callbacks from the traversal when entering
structs, array elements, and new variables. The visitor can treat these
differently depending on the use case. A common visitor that constructs
full variable names is used as a base class in several places.
Also moves the 'isRowMajorLayout' from sh::InterfaceBlockField to
sh::ShaderVariable. This allows us to forgo using templates in several
call sites.
Bug: angleproject:3024
Change-Id: I472d81ec775e2eee92fb3d2eb0ca83860221ba2e
Reviewed-on: https://chromium-review.googlesource.com/c/1358722
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
|
|
|
e321940c
|
2018-12-08T16:54:14
|
|
Vulkan: Move image clear functionality to UtilsVk
DispatchUtilsVk is renamed to UtilsVk and the functionality in
framebuffer's clearWithDraw() is moved to that class. Eventually, more
fragment-shader-based internal algorithms will be added to this class as
well.
Bug: angleproject:2958
Change-Id: I4753c9cb3288b59cd1ed60fe7a57b9f189704322
Reviewed-on: https://chromium-review.googlesource.com/c/1369284
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
|
|
|
3f01f53c
|
2018-12-11T15:13:51
|
|
ES31: Add state change tests for image
This change adds dirty bit state change tests for image textures.
Meanwhile, syncImages related codes are removed since syncTextures will do
all texture states sync.
Bug: angleproject:3015
Change-Id: I9b299c86af1d589e72c08c5d7c55ac74cc7833aa
Reviewed-on: https://chromium-review.googlesource.com/c/1390596
Commit-Queue: Jiajia Qin <jiajia.qin@intel.com>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
|
|
|
cb9609fe
|
2018-12-26T00:52:44
|
|
Optimize glDrawElements performance
A call to glDrawElements results in a calling depth of up to 4
* glDrawElements
* gl::Context::DrawElements
* rx::ContextGL::DrawElements
* VertexArrayGL::syncDrawState.
Each function call has to save/restore a lot of registers which
results in a stall in the prologue of rx::ContextGL::DrawElements
due to memory bandwidth limitations.
The main change is the function gl::Context::DrawElements being
inlined to reduce the calling depth by one. In addition the call
to ContextGL::syncDrawElementsState is now protected so that it
gets called only if it's required. Finally a few small getter
functions have been inlined where the calling code was bigger
than the actual function.
In total this change improves performance of the
DrawElementsPerfBenchmark.Run/gl benchmark by 16%.
Bug: angleproject:2966
Change-Id: I423d18452f2f5b520ab52850fda2054e1da86991
Reviewed-on: https://chromium-review.googlesource.com/c/1389988
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Markus Tavenrath <matavenrath@nvidia.com>
|
|
|
84412860
|
2018-12-21T17:41:24
|
|
Set dirty for uniform on D3D backend
When uniform's value had been updated, set dirty and
update its content on D3D backend.
Bug: angleproject:2814
TEST=angle_end2end_tests.ComputeShaderTest
.UniformDirty/ES3_1_D3D11
dEQP.GLES31/functional_image_load_store_cube*
dEQP.GLES31/functional_image_load_store_3d*
dEQP.GLES31/functional_image_load_store_2d_array*
Change-Id: Ic135c140559925fe33790475d03dc608afa92384
Reviewed-on: https://chromium-review.googlesource.com/c/1387967
Reviewed-by: Jiajia Qin <jiajia.qin@intel.com>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Xinghua Cao <xinghua.cao@intel.com>
|
|
|
ae6ba9f8
|
2018-12-21T23:00:04
|
|
Optimize check for active not paused XFB.
Local testing showed an improved score of about 2% in the most
sensitive CPU overhead benchmark. Likely because of improved
caching from fewer indirections.
Bug: angleproject:2966
Change-Id: I5d9a4b4bcf624ab0b430bb696c4227e589cdb7a6
Reviewed-on: https://chromium-review.googlesource.com/c/1359518
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
1e853266
|
2018-12-21T09:07:38
|
|
Cache common DrawElements states.
Similar to how we cache the base common draw states. This will improve
DrawElements performance. Several state checks are optimized into a
single 'if' check of a cached value.
Also includes a regression test for mapping the element array buffer.
Bug: angleproject:2966
Change-Id: Ia6e524a58ad6b7df2e455d67733e15d324b1b893
Reviewed-on: https://chromium-review.googlesource.com/c/1357150
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>
|
|
52f5da43
|
2018-12-21T09:02:09
|
|
Also enable flipY for Vulkan 1.1
The necessary support for flipY is included in core Vulkan 1.1
so enable the feature if that's what's available.
Bug: angleproject:2968
Change-Id: I6808a1a0b1ace74976a7476f77d686e137a1e747
Reviewed-on: https://chromium-review.googlesource.com/c/1387758
Commit-Queue: Ian Elliott <ianelliott@google.com>
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
|
|
|
899c5d26
|
2018-12-21T13:12:50
|
|
Appropriately set VkApplicationInfo::apiVersion to 1.1
Check if the Vulkan instance version is 1.1+, and if so, set
VkApplicationInfo::apiVersion to 1.1 (which signals to the validation
layers that ANGLE will use Vulkan 1.1 functionality.
Bug: angleproject:2968
Change-Id: I6808a1a0b1ace74976a7476f77d686e137a1e746
Reviewed-on: https://chromium-review.googlesource.com/c/1387757
Commit-Queue: Ian Elliott <ianelliott@google.com>
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
|
|
|
d50521fd
|
2018-12-20T12:05:14
|
|
Only enable VK_KHR_maintenance1 when available.
This prevents an error in case ANGLE is run on an older Vulkan driver
that does not support the VK_KHR_maintenance1 extension.
Bug: angleproject:3035
Change-Id: I7e2b74bcb88bffdfc2a525a1d332eb3ef7d23d7c
Reviewed-on: https://chromium-review.googlesource.com/c/1387347
Reviewed-by: Ian Elliott <ianelliott@google.com>
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
Reviewed-by: Courtney Goeltzenleuchter <courtneygo@google.com>
Commit-Queue: Ian Elliott <ianelliott@google.com>
|
|
|
bcb78908
|
2018-12-19T11:46:29
|
|
Implement eglSwapBuffersWithDamage on top of VK_KHR_incremental_present
Bug: angleproject:2510
Change-Id: I12b0877f787dbcb48e2890f54ba4bc8ebe8294b4
Reviewed-on: https://chromium-review.googlesource.com/c/1383373
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
Commit-Queue: Ian Elliott <ianelliott@google.com>
|
|
|
2db5082b
|
2018-12-17T14:27:18
|
|
Add UAV format for GL_RGBA8I
GL_RGBA8I is a supported image unit format, add
its corresponding UAV format on D3D backend.
Bug: angleproject:1987
TEST=dEQP.GLES31/functional_image_load_store_2d_store_rgba8i
dEQP.GLES31/functional_image_load_store_2d_load_store_rgba8i
Change-Id: Ib759f873c8629a644f85784a57ce141d3018e0c1
Reviewed-on: https://chromium-review.googlesource.com/c/1378888
Reviewed-by: Jiajia Qin <jiajia.qin@intel.com>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Xinghua Cao <xinghua.cao@intel.com>
|
|
|
7139b434
|
2018-12-14T16:57:28
|
|
Print a warning on unexpected internal errors.
The error message can be dropped when converting angle::Result to egl::Error.
This makes sure something is always printed when we get an unexpected error.
BUG=914911
Change-Id: Icc8b06b1c13e3ea83287da5a217f4c8bc218deec
Reviewed-on: https://chromium-review.googlesource.com/c/1378689
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Geoff Lang <geofflang@chromium.org>
|
|
|
a7af56be
|
2018-12-14T14:20:28
|
|
Vulkan: Print the device id in the renderer string.
Add a end2end test that prints all strings to help debug on the bots.
BUG=angleproject:3026
Change-Id: Ia524a0d7dac88e55e0aa67412339e646a5564e64
Reviewed-on: https://chromium-review.googlesource.com/c/1378686
Commit-Queue: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
|
|
|
e1f742af
|
2018-12-14T14:17:59
|
|
GL: Clean up workaround detection.
Use functions instead of macros for platform detection so compilation can be
checked from any platform. Also makes the workaround detection more readable.
BUG=angleproject:3026
Change-Id: I4153f0a2a6a6d5860c7b37f7cc67561895165ed1
Reviewed-on: https://chromium-review.googlesource.com/c/1378685
Commit-Queue: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
|
|
|
91002266
|
2018-12-12T16:05:24
|
|
Use angle::Result and pass a context pointer to TransformFeedbackImpl
BUG=angleproject:3020
Change-Id: Ib0877dd33f9a8c6ea57976642f1b904258cb6a86
Reviewed-on: https://chromium-review.googlesource.com/c/1374273
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Geoff Lang <geofflang@chromium.org>
|
|
|
9b02506c
|
2018-12-12T15:44:12
|
|
Cache valid draw modes with transform feedback.
Enabling transform feedback can affect which draw modes are valid. We
can use the exiting draw modes cache to save having to check the draw
modes twice. We update the cached draw modes on any change to the
transform feedback activity state. e.g. when transform feedback is
started, or resumed.
There are also spec changes that comes into effect in ES 3.2 or when
EXT_geometry_shader is enabled. Again we cache these draw modes in the
packed valid draw modes map.
Will allow for faster validation for draw calls once the other checks
for transform feedback are optimized. Also adds a new regression test.
Bug: angleproject:2966
Change-Id: Iab901e45aab70980b9e631ec8383fdeadbd32368
Reviewed-on: https://chromium-review.googlesource.com/c/1357149
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
|
|
|
b08457df
|
2018-12-11T15:13:54
|
|
Vulkan: Make flushAfterVertexConversion workaround Nexus5X-specific
Workaround added in 611bbaabb28cf4231ad0a138cd7bc21fe559ea02 is likely
Nexus5X-specific. This commit detects the phone and enables the
workaround only on that.
Bug: angleproject:2958, angleproject:3009
Change-Id: I9ab230d4aa690fd92e3d2d84ad98c159128c1093
Reviewed-on: https://chromium-review.googlesource.com/c/1372445
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
3a256228
|
2018-12-08T09:56:39
|
|
Update ANGLE_multiview validation.
Multiview transform feedback now passes if the transform feedback is
paused. Also updates the relevant validation. This aligns the
ANGLE_multiview validation with OVR_multivew and with the WebGL
extension spec.
This change allow us to combine multiple draw call validation updates
into one update function.
Bug: angleproject:3012
Change-Id: I2ce04edc386039134198afa1792201e03cc20490
Reviewed-on: https://chromium-review.googlesource.com/c/1357148
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
611bbaab
|
2018-12-06T01:59:53
|
|
Vulkan: Convert vertex attributes in compute
In this commit, VertexArrayVk::convertVertexBuffer() is renamed to
VertexArrayVk::convertVertexBufferCpu() to explicitly show it does a CPU
readback. A new VertexArrayVk::convertVertexBuffer() function is added
that has the same functionality in gpu (with some assumptions, where the
CPU fallback is used should those assumptions fail). Currently, the
only requirement is that buffer offset/stride are divided by the
component size.
ConvertVertex.comp is the shader responsible for this conversion, and it
implements the functionality in renderer/copyvertex.inc, minus a few
functions that are not used in the Vulkan backend.
Bug: angleproject:2958, angleproject:3009
Change-Id: I8ec9a5f4672509bcf7b9e352cd27663970ad4653
Reviewed-on: https://chromium-review.googlesource.com/c/1364451
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
|
|
|
be607ad6
|
2018-11-29T10:14:22
|
|
GL: Implement GL_OES_EGL_image_external and GL_OES_EGL_image_external_essl3
Chrome uses external textures as part of its hardware video decode paths
on Android.
BUG=angleproject:2507
Change-Id: I2af608f84a99843c99a16dcfb9fb2fa28cc8fbb6
Reviewed-on: https://chromium-review.googlesource.com/c/1361480
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
Commit-Queue: Geoff Lang <geofflang@chromium.org>
|
|
|
f5c88e7e
|
2018-12-08T09:56:38
|
|
Refactor and inline DrawElements validation.
This moves ValidateDrawBase into a common inline function. It also
moves some checks in ValidateDrawElementsCommon into
ValidateDrawElementsBase. "Base" is called from DrawElementsIndirect
while "Common" is only called from non-indirect entry points. But this
improves conformance because the missing checks in "Base" apply to
DrawIndirect as well.
In a follow-up patch many checks in "Base" will be cached into a single
value. Much like what we did for ValidateDrawBase.
Also inlines ValidateDrawElements which just calls through to "Common".
Bug: angleproject:2966
Change-Id: Ibe0c5db25f403dd52a50dc73373ebb50cedad003
Reviewed-on: https://chromium-review.googlesource.com/c/1357147
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
|
|
|
149eb33b
|
2018-12-08T14:53:49
|
|
Vulkan: add possibility to build subset of shaders
As the number of internal shaders grow, it's helpful in development time
to be able to build variations of a specific shader, instead of building
everything every time.
With this commit, one can do:
$ python gen_vk_internal_shaders.py SubString
which will result in compilation of shaders whose name matches the glob
*SubString*.
Bug: angleproject:2958
Change-Id: I3b18d7e32e2ec4317531a7655eb2c6148cea177b
Reviewed-on: https://chromium-review.googlesource.com/c/1369285
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
|
|
|
5c4af666
|
2018-12-07T06:31:00
|
|
Vulkan: unix line endings in generated shaders
Bug: angleproject:3007
Change-Id: I466caa92ae337508ae284b6fb8d38f5fa3ea1042
Reviewed-on: https://chromium-review.googlesource.com/c/1366796
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
0c2dc8c3
|
2018-12-07T08:00:49
|
|
Vulkan: Fix dirty default attributes handling
When invalidating default attributes, the mask to invalidate was set as
the current dirty mask, but it should have been or'ed with it.
The following test triggered a validation error on Linux:
dEQP.GLES2/functional_default_vertex_attrib_mat4_vertex_attrib_4fv
In this test, Vertex Array Bindings and Program dirty bits would set the
default attrib dirty mask to all the default attributes, but the
handling of Current Values dirty bits would mistakenly remove some of
those dirty bits.
Bug: angleproject:2786
Change-Id: Ife96621da7646102c1ce3b24e45f405e34a08b40
Reviewed-on: https://chromium-review.googlesource.com/c/1366797
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
7c985f5c
|
2018-11-29T18:16:17
|
|
Make angle::Result an enum.
This moves away from a class type to a value type. This should improve
performance when using angle::Result as a return value. Previously the
generated code would return a pointer instead of a value.
Improves performance in the most targeted microbenchmark by 10%. In
more realistic scanarios it will have a smaller improvement. Also
simplifies the class implementation and usage.
Includes some unrelated code generation changes.
Bug: angleproject:2491
Change-Id: Ifcf86870bf1c00a2f73c39ea6e4f05ca705050aa
Reviewed-on: https://chromium-review.googlesource.com/c/1356139
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
|
|
|
c3e3731c
|
2018-11-30T15:25:39
|
|
Give every validation error a constant.
Makes the code more consistent and organized.
Also reduces binary size by about 4k.
Also refactors validation for GetBufferPointerv.
Bug: angleproject:2957
Change-Id: Ib076ce936193e9840eef2499bf815489cdb48479
Reviewed-on: https://chromium-review.googlesource.com/c/1350489
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
|
|
|
ec06305b
|
2018-12-06T01:58:06
|
|
Vulkan: Add dynamic vertex buffers to graph
With DynamicBuffer outputting BufferHelper objects, these objects can
participate in the command graph, i.e. record commands. This means they
need appropriate dependencies in the graph as well as pipeline barriers.
There are a few users of DynamicBuffer for which this change should be
applied to. This change covers vertex attribute buffers.
Bug: angleproject:2958
Change-Id: Ic3f8e187639919aba23102bba238de9ce70e6394
Reviewed-on: https://chromium-review.googlesource.com/c/1364450
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
132d15c2
|
2018-11-30T15:25:38
|
|
Clear uncleared textures as a dirty object.
This removes the check for robust resource init from prepareForDraw.
It should improve performance more when WebGL compatiblity is enabled.
Also inlines Context::syncDirtyBits and Context::prepareForDraw.
Reduces CPU overhead in all draw and dispatch calls up to 5%.
Bug: angleproject:2966
Change-Id: I7e9009a501c0331a6140bc78678aef27e38c3df2
Reviewed-on: https://chromium-review.googlesource.com/c/1347453
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
|
|
|
8f1b7a66
|
2018-11-14T16:02:54
|
|
Vulkan: Add DispatchUtilsVK
This class provides a set of compute-based internal utilities.
Currently, buffer clear and copy are implemented. Other possibilities
include more efficient mip map generation, or specialized texture
operations.
VertexArrayVk::updateIndexTranslation() is updated to convert the
GL_UNSIGNED_BYTE index buffer to a GL_UNSIGNED_SHORT one using this
class to avoid a CPU readback.
The vk::Format class is augmented with a few flags (IsInt, IsUnsigned)
to be able to select the appropriate shader based on the format (float,
int or uint).
Bug: angleproject:2958,angleproject:3003
Change-Id: Ie35519deb3c32a3da5ccf74080c70092c9287f0a
Reviewed-on: https://chromium-review.googlesource.com/c/1336307
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
|
|
|
eae081c6
|
2018-12-03T22:32:32
|
|
Vulkan: Fix ASTC_12x12_SRGB_BLOCK support
A typo in the .json file made this format unavailable.
Bug: angleproject:2357
Change-Id: I3214c6d8a00d9fa731573347b06bded8fb57a918
Reviewed-on: https://chromium-review.googlesource.com/c/1359619
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
|
|
|
96bd8fdf
|
2018-11-30T14:30:18
|
|
Vulkan: Fix format properties queries
When querying format properties (in vk::GetFormatProperties), the
mandatory feature support table was consulted to check whether a number
of texture features are present. If so, the entry from that table was
returned. The goal had been to speed up initialization by not issuing
device queries if possible.
That is, when vk::GetFormatProperties was called on a format, if it
supported that select few texture features, the VkFormatProperties entry
from the mandatory table would be returned.
However, that function found its way to other uses (such as querying
buffer format properties, or other image properties beyond the select
few). As a result, when the VkFormatProperties from the mandatory table
was returned, actual support for these other features was often not
tested and assumed false (unless they happened to be mandatory as well).
This commit reworks the format feature query functions such that the
specific features to be tested are provided when querying the format
properties. The mandatory table is consulted as before, and if the
entry doesn't contain those features, the device is queried and the
results cached.
Bug: angleproject:2958
Change-Id: I28d046eb63c3bd5173468aa4cb3e4c63c83e67b1
Reviewed-on: https://chromium-review.googlesource.com/c/1357152
Reviewed-by: Tobin Ehlis <tobine@google.com>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
654b3c56
|
2018-12-03T21:59:35
|
|
Vulkan: Fix multi-layer image readback
The buffer size did not take layer count into account.
angleproject: 2828
Change-Id: I94b9294503d785d3e727422550d7e4643eed1917
Reviewed-on: https://chromium-review.googlesource.com/c/1359618
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
|
|
|
e3e680ca
|
2018-12-03T17:49:08
|
|
Remove State::syncProgramTextures.
Removes the concept of the program textures dirty object. Instead we
use a set of dirty bits to represent dirty texture samples. We mark
certain textures dirty and update state structures whenever there is
a new Texture/Program/Sampler bound, or when Texture/Program/Sampler
state changes.
This is in preparation for making clearing the uncleared active
textures into a dirty bit as well.
Also includes new dirty bit handling for texture image units. These are
a GLES 3.1 feature.
Bug: angleproject:2966
Change-Id: Ibb8619dd2669bb39fdbcd75e3685be9a8aeeee91
Reviewed-on: https://chromium-review.googlesource.com/c/1346649
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
66f0d2c1
|
2018-11-30T15:25:36
|
|
Make Framebuffer attachments angle::Subjects.
Now that there's storage change notifications in the GL front-end we
no longer need to give the back-end access to the angle::Subject. The
Texture object is a special case where it has mirrored dirty bits. To
keep the gl::Texture class notified of when the Impl has dirty bits we
make the TextureImpl class an angle::Subject that is observed by the
gl::Texture class.
This will enable further dirty bits improvements.
Bug: angleproject:2966
Change-Id: Id22da0926f51ff4679e58af3e62903f4d7948915
Reviewed-on: https://chromium-review.googlesource.com/c/1347670
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>
|
|
|
4abdf74f
|
2018-11-28T14:41:10
|
|
Vulkan: Add dynamic index buffers to graph
With DynamicBuffer outputting BufferHelper objects, these objects can
participate in the command graph, i.e. record commands. This means they
need appropriate dependencies in the graph as well as pipeline barriers.
There are a few users of DynamicBuffer for which this change should be
applied to. This change covers index buffers.
This commit includes a fix to BufferHelper::copyFromBuffer for WaW
hazards.
It also includes a fix for a missing pipeline barrier after
BufferVk::copyToBuffer.
Bug: angleproject:2958
Change-Id: I3e61af56936580b2da20c28c45defece552d9a39
Reviewed-on: https://chromium-review.googlesource.com/c/1352732
Reviewed-by: Tobin Ehlis <tobine@google.com>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
b51eb09f
|
2018-11-29T10:48:36
|
|
Vulkan: Allow bindings to compute stage
Includes both binding of resources and updating push constants on the
compute shader stage.
Bug: angleproject:2958
Change-Id: I5686cbac81e0dd83d0e938cb40f9f9ac7d2ef48a
Reviewed-on: https://chromium-review.googlesource.com/c/1355500
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
|
|
|
df8c1053
|
2018-11-29T11:49:45
|
|
Vulkan: Uber-shader generated code optimization
Refactors code out of generated functions into a common function and
simplifies assertions.
Bug: angleproject:2958
Change-Id: I896c2304c3ac1e043c16f9ecf81fa8b72a6b87c0
Reviewed-on: https://chromium-review.googlesource.com/c/1355501
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
|
|
|
e90d4ee9
|
2018-11-28T14:04:00
|
|
Pass Context to setLabel.
This is useful for triggering a dirty state notification for Textures.
It will lead to improvements for program and texture dirty bits.
Bug: angleproject:2966
Change-Id: Iaba625da8a970a558f7d158bfa2f09c964f6761a
Reviewed-on: https://chromium-review.googlesource.com/c/1347669
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
|
|
|
ef9fcd91
|
2018-11-28T14:03:59
|
|
Clear draw attachments as a dirty object.
This allows us to skip 1/2 of the robust resource init check in
Context::setupDraw. The plan is to remove the other half in a follow-up
CL.
Most of the work of this CL was already handled. We just need to add
the right dirty object mask to the Context's draw dirty objects list.
We can mask out this check when robust resource is not enabled.
Bug: angleproject:2966
Change-Id: I97ec2497c95e5cdf52988e1ce85f7602206935f0
Reviewed-on: https://chromium-review.googlesource.com/c/1343140
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
|
|
|
8dc27f99
|
2018-11-29T11:45:44
|
|
Use packed enum for DrawElementsType.
The packing and unpacking take a few extra instructions. But it
completely obviates the need for any switches in the validation code.
Speed is slightly faster or the similar depending on the back-end.
Also add gl_angle_ext.xml to GL entry point generator inputs. This was
missing and would cause the code generation to miss certain changes.
Bug: angleproject:2985
Change-Id: I1ea41a71db71135000166ead8305ec42d22ff7b3
Reviewed-on: https://chromium-review.googlesource.com/c/1351729
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
471358f3
|
2018-11-28T15:21:55
|
|
Vulkan: fix pipeline stages of memory barriers
VK_PIPELINE_STAGE_TOP_OF_PIPE_BIT was used for barriers, but TOP and
BOTTOM bits can only be used for execution barriers. These stages don't
access memory, so they don't participate in memory barriers at all. A
validation error is generated as our current barriers violate this:
https://www.khronos.org/registry/vulkan/specs/1.1-extensions/html/vkspec.html#VUID-vkCmdPipelineBarrier-pMemoryBarriers-01184
A note from the spec (see second paragraph in particular):
> An execution dependency with only VK_PIPELINE_STAGE_BOTTOM_OF_PIPE_BIT
> in the destination stage mask will only prevent that stage from
> executing in subsequently submitted commands. As this stage does not
> perform any actual execution, this is not observable - in effect, it
> does not delay processing of subsequent commands. Similarly an execution
> dependency with only VK_PIPELINE_STAGE_TOP_OF_PIPE_BIT in the source
> stage mask will effectively not wait for any prior commands to complete.
> When defining a memory dependency, using only
> VK_PIPELINE_STAGE_BOTTOM_OF_PIPE_BIT or
> VK_PIPELINE_STAGE_TOP_OF_PIPE_BIT would never make any accesses
> available and/or visible because these stages do not access memory.
> VK_PIPELINE_STAGE_BOTTOM_OF_PIPE_BIT and
> VK_PIPELINE_STAGE_TOP_OF_PIPE_BIT are useful for accomplishing layout
> transitions and queue ownership operations when the required execution
> dependency is satisfied by other means - for example, semaphore
> operations between queues.
Bug: angleproject:2958
Change-Id: Ic616dcad7583db6b386d7d01a774b3ebd71a7081
Reviewed-on: https://chromium-review.googlesource.com/c/1352733
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
|
|
|
f7f8c518
|
2018-11-18T15:56:45
|
|
Optimize State::syncDirtyObjects
Create a function table for State::syncDirtyObjects to avoid the
costly switch and inline calls along the Context::syncDirtyObjects
calling hierarchy.
Bug: angleproject:2975
Change-Id: I1ec797452af41bc767578e4017c8eccb7d83628b
Reviewed-on: https://chromium-review.googlesource.com/c/1340222
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
|
|
|
a1442ecd
|
2018-11-26T12:48:10
|
|
Vulkan: Uber-shader support
For each internal shader, there is now possibly a companion .json file
that defines shader variations. The variations consist of a set of
flags and a set of enums. Each enum defines a variation that takes one
of a few values. Flags are shorthands for 2-value enums.
In the shader code, #if can be used to change the shader based on
variations derived from flags and enums. On the C++ side, those
variations are combined into an index and the appropriate shader is
retrieved from a table.
Bug: angleproject:2958
Change-Id: Ic4fc7061adb20c047c26624305285e3005092aab
Reviewed-on: https://chromium-review.googlesource.com/c/1351117
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
|
|
|
5f01324f
|
2018-11-26T12:41:12
|
|
Vulkan: Allow descriptor pool to allocate different types
The internal shaders know exactly what mix of resources they use, so
with this change they can make one allocation of multiple types.
Bug: angleproject:2958
Change-Id: Ia6f03672b95af4be9618a2f5eeb41ef1c027e810
Reviewed-on: https://chromium-review.googlesource.com/c/1351116
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
7310da33
|
2018-11-05T14:40:01
|
|
Implement GetMultismaplefvANGLE and SampleMaskiANGLE of texture multisample extension
This patch adds GetMultismaplefvANGLE and SampleMaskiANGLE api.
TEST=TextureMultisampleTest.CheckSamplePositions*
TEST=NegativeTextureMultisampleTest.Negtive*
BUG=angleproject:2275
Change-Id: Ib1e8b856059229bbe88328a28cf2a892f4673214
Reviewed-on: https://chromium-review.googlesource.com/c/1319252
Commit-Queue: Yizhou Jiang <yizhou.jiang@intel.com>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
|
|
|
2c0d6a94
|
2018-11-28T14:03:58
|
|
WebGL: Fix DrawElements test and remove size check.
Also optimizes the DrawElements validation to remove redundant tests.
Also make some more validation WebGL-specific.
Bug: angleproject:2979
Change-Id: I2d5c3094ed86ebadbc572e46b8ae105431f7ae68
Reviewed-on: https://chromium-review.googlesource.com/c/1343139
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>
|
|
|
5829c03d
|
2018-11-28T10:39:41
|
|
Vulkan: fix ProgramVk never releasing uniform buffers
Bug: angleproject:2958
Change-Id: I7fd50e8b2029aad59a35050f401e1cf70d047444
Reviewed-on: https://chromium-review.googlesource.com/c/1354098
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
|
|
|
254b32cb
|
2018-11-26T11:58:03
|
|
Vulkan: Make DynamicBuffer use BufferHelper
This is so that the resulting buffers can be written to by the GPU.
Additionally, the class is given the ability to create host-visible or
device-local buffers, making map()-on-init() optional.
This is in preparation for vertex/index transformations in compute.
Bug: angleproject:2958
Change-Id: Ib8f5829e33a1e49fa8f80c70dbde74f313ae49ec
Reviewed-on: https://chromium-review.googlesource.com/c/1351113
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
|
|
|
83b2f0e8
|
2018-11-27T15:56:01
|
|
WebGL: Forbid GenerateMipmap on zero-size textures
This is a WebGL spec change:
https://github.com/KhronosGroup/WebGL/commit/bfbe124a7bb92eed34ea7cc18694167ed66b1030
Fixes conformance2/textures/misc/tex-mipmap-levels.html
Bug: chromium:898351
Change-Id: Ib20409e10bec598611be273577115a380a63b523
Reviewed-on: https://chromium-review.googlesource.com/c/1352385
Commit-Queue: James Darpinian <jdarpinian@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
|
|
3b7c9d09
|
2018-11-21T18:09:05
|
|
Change offsets in MultiDrawElements* entrypoints to match glMultiDrawElements
This patch changes arguments for MultiDrawElements and
MultiDrawElementsInstanced from from GLsizei *offsets to
const* GLvoid* indices.
Bug: chromium:890539
Change-Id: I2124cf2daeaa6af0f889f3c46aa2b8c77be200a6
Reviewed-on: https://chromium-review.googlesource.com/c/1352591
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Kai Ninomiya <kainino@chromium.org>
Commit-Queue: Austin Eng <enga@chromium.org>
|
|
|
4f6592fa
|
2018-11-27T16:37:45
|
|
Remove gl::Error.
Removes several TODOs. Only egl::Error remains.
Also slightly decreases binary size.
Bug: angleproject:2491
Change-Id: I3a9d1c22eb0884ca9e37362463fddd0083faf826
Reviewed-on: https://chromium-review.googlesource.com/c/1337462
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
|
|
|
e0472f3d
|
2018-11-27T16:32:45
|
|
Slight refactor to ErrorStrings.h.
Allows for better readability. Also sort the error strings by ID.
Bug: angleproject:2957
Change-Id: Ib3900a4c90507ac84496fd5ff916417eb768667b
Reviewed-on: https://chromium-review.googlesource.com/c/1337461
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
|
|
|
752d220a
|
2018-11-27T13:29:48
|
|
Use flat enum for PrimitiveMode.
The GLenum was already mostly packed. There were just a few missing
values because of the exclusion of things like GL_QUADS and GL_POLYGON
from GLES.
Also update the PackedEnumMap initialization to take an intializer.
The initializer is a list of key/values which is much more robust to
changes in the packed map.
Improves draw call speed slightly as there is no conversion needed any
more for the mode enum.
Bug: angleproject:2966
Change-Id: Icae658272c6234f29335f6a57a63cf341cf5b2a0
Reviewed-on: https://chromium-review.googlesource.com/c/1346529
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
b980c563
|
2018-11-27T11:34:27
|
|
Reformat all cpp and h files.
This applies git cl format --full to all ANGLE sources.
Bug: angleproject:2986
Change-Id: Ib504e618c1589332a37e97696cdc3515d739308f
Reviewed-on: https://chromium-review.googlesource.com/c/1351367
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
4f2b94cb
|
2018-11-26T11:38:02
|
|
Expose EXT_color_buffer_half_float if RGBA16F is renderable.
The EXT_color_buffer_half_float extension spec issue #2 states that none
of the added color buffer formats are required to be renderable and the
client must check for framebuffer completeness for using them. Chrome
exposes the extension if at least RGBA16F is renderable for WebGL.
Replicate Chrome's behaviour by loosening our requirements to only
expose this extension if an explicit check for RGBA16F renderability
succeeds.
BUG=angleproject:2984
Change-Id: Id97f3043ebf3fd11b5e9e2d505e57b76baba9716
Reviewed-on: https://chromium-review.googlesource.com/c/1351350
Commit-Queue: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
|
|
|
471b8d4c
|
2018-11-21T15:37:47
|
|
WebGL, D3D: Forbid multiple TF outputs in one buffer
Implementing a WebGL spec change:
https://github.com/KhronosGroup/WebGL/pull/2658
Fixes WebGL 2 conformance test:
conformance2/transform_feedback/same-buffer-two-binding-points.html
Also applying the change to D3D backends because they can't support
multiple transform feedback outputs in one buffer.
Bug: chromium:866089
Change-Id: I8d7eda14225c13efb7ca1ed974239332be4e79a6
Reviewed-on: https://chromium-review.googlesource.com/c/1347749
Commit-Queue: James Darpinian <jdarpinian@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
|
|
|
5552cdf0
|
2018-11-26T12:19:26
|
|
Vulkan: Create buffer view for BufferHelper
Bug: angleproject:2958
Change-Id: Ifa40537bc6003486b37d5d2ac9a2b7f1f5c3d1c2
Reviewed-on: https://chromium-review.googlesource.com/c/1351115
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
1b17f90b
|
2018-11-26T14:02:02
|
|
Vulkan: Implement ANGLE_translated_shader_source
BUG=angleproject:2909
Change-Id: Ie59c1206e2ce4940e27fed7887dbee665721049b
Reviewed-on: https://chromium-review.googlesource.com/c/1351354
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Geoff Lang <geofflang@chromium.org>
|
|
|
bc8cc5af
|
2018-11-26T12:15:14
|
|
Vulkan: Implement vk::BufferView
Bug: angleproject:2958
Change-Id: I3f77e694638b4dc060d20a149c3f9c8aed3692d1
Reviewed-on: https://chromium-review.googlesource.com/c/1351114
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
7198ebc4
|
2018-11-22T14:36:06
|
|
GL: Work around Qualcomm sRGB ReadPixels driver bug.
When calling glReadPixels on unsized sRGB textures, the data returned is
not linearized. Using sized sRGB formats works around this problem.
Fix SRGBTextureTest assuming that unsized SRGB formats work in ES3.
BUG=angleproject:2977
Change-Id: I413c888ca0c3564ecdc364f3ea693e75ed4f8d6d
Reviewed-on: https://chromium-review.googlesource.com/c/1343138
Reviewed-by: Tobin Ehlis <tobine@google.com>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Commit-Queue: Geoff Lang <geofflang@chromium.org>
|
|
|
617103ed
|
2018-11-22T13:18:05
|
|
GL: Implement GL_NV_fence on top of GLsync objects.
When ANGLE exposes an ES2 context, there is often no way to create fence
objects because GL_NV_fence is not available.
BUG=882580
Change-Id: Ic18539e0243317a188545f3945f6b33d3401f1a8
Reviewed-on: https://chromium-review.googlesource.com/c/1348769
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Commit-Queue: Geoff Lang <geofflang@chromium.org>
|
|
|
0d665139
|
2018-11-18T15:47:02
|
|
Don't generate TypeInfo objects as static objects within GetTypeInfo function
Generating static objects within a function results in multithread safe
code. This code generates the static objects upon the first execution of the
line which declares the object. This results in high runtime cost for
synchronization and a bigger code size.
Instead introduce a new function uint32_t GetPackedTypeInfo(GLenum type)
which returns a packed representation for the Type class. This
representation is usually returned in a register on the assembly level.
As a result we save constant storage for the TypeInfo object and one
indirection when reading any value from this object. The Type constructor
accepts the packed representation and unpacks it an inline function. For
fields which are not used the compiler also applies dead code
elimination which reduces the cost furthermore.
As a result of this change the cost of GetTypeInfo is reduced by a
factor of 4-5.
Bug: angleproject:2974
Change-Id: I8ed0bf2f09d087fa4cffa04f82e3b7f8c183fe30
Reviewed-on: https://chromium-review.googlesource.com/c/1340221
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>
|