- 25 Jun, 2019 7 commits
-
-
Geoff Lang authored
BUG=angleproject:3559 Change-Id: Ib57b40d6f6b488549c5746a4e2d40e6fba6bcd4f Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1676306Reviewed-by:
Jonah Ryan-Davis <jonahr@google.com> Commit-Queue: Geoff Lang <geofflang@chromium.org>
-
Shahbaz Youssefi authored
Previously, only read-only layouts skipped same-layout transition. This is extended to color attachment and depth/stencil attachment layouts as well. On Nvidia+Linux, this does reduce the number of barriers in the Draw* benchmarks, though there's no visible CPU-side performance difference. GPU time is not measured in these tests, but it's possible that the driver already skips such transitions (i.e. this only saves us from recording an ineffective transition). Additionally, transfer dst and shader write same-layout transitions are turned into an execution barrier instead of a memory barrier to avoid an unnecessary flush. Currently, this particularly affects texture upload, and shows a 10% improvement in TextureUploadSubImageBenchmark. Note that this optimization is temporarily disabled due to a bug in the windows AMD driver. Bug: angleproject:3347 Change-Id: I7dc9d0b5dd2ad87ec19ae13277b330438038519f Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1659149Reviewed-by:
Tobin Ehlis <tobine@google.com> Reviewed-by:
Jamie Madill <jmadill@chromium.org> Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
-
shrekshao authored
HLSL adds a unique id suffix to user defined variable and these numbers are currently hardcoded in tests. It could easily fail if new tokens are added to angle builtin variables. Introduce C++11 regex feature into the foundInHLSLCode to avoid over strict tests. Bug: angleproject:3551 Change-Id: Ia3052afec667a7ac11198be31b5c1d03c856a723 Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1670581Reviewed-by:
Geoff Lang <geofflang@chromium.org> Commit-Queue: Shrek Shao <shrekshao@google.com>
-
Geoff Lang authored
Chrome uses regular texture uploads on all platforms except OpenGL on Mac to do robust resource init for textures because multiple bugs have been observed with texture corruption. BUG=883276 BUG=882580 BUG=941620 Change-Id: I5fcd6862147822a08c7533e6a6a9277223034ebd Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1669104Reviewed-by:
-
Geoff Lang authored
ValidateSamplerParameterBase is called with both const and non-const parameter pointers which caused our explicit template instantiations to not cover all cases resulting in link errors. Force the parameters to always be const so that our instantiations cover everything. BUG=angleproject:3555 Change-Id: Ibf92572ba80a689c75dcdc70e5153d9941da76f3 Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1676305Reviewed-by:
-
Qin Jiajia authored
Bug: angleproject:3548 Change-Id: Idd273cdfff5f175399dcc46bd40db52d6f4163b0 Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1666909Reviewed-by:
-
Qin Jiajia authored
The error happens when a RWTexture is used in shader. However, there is no resource binding to it. We should clear the corresponding UAV in case the previous view type is a bufer not a texture. Meanwhile, this patch removes clearSRVs/clearUAVs since we use unsetConflictingSRVs/unsetConflictingUAVs to do the similar thing. Bug: angleproject:3512, angleproject:3548 Change-Id: I01752bb9bc6aca5b767599639c4dc613b4e2e2d2 Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1662017Reviewed-by:
-
- 24 Jun, 2019 2 commits
-
-
Tibor Goldschwendt authored
Bug: 891996 Change-Id: I64381bea1bcbddb76748dd03c7caaf2f5360e1ed Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1674258Reviewed-by:
Cody Northrop <cnorthrop@google.com> Reviewed-by:
Yuly Novikov <ynovikov@chromium.org> Commit-Queue: Yuly Novikov <ynovikov@chromium.org>
-
angle-autoroll authored
https://chromium.googlesource.com/external/github.com/KhronosGroup/SPIRV-Tools.git/+log/7c294608ca19..df86bb44fe47 Created with: gclient setdep -r ./third_party/spirv-tools/src@df86bb44fe47 The AutoRoll server is located here: https://autoroll.skia.org/r/spirv-tools-angle-autoroll Documentation for the AutoRoller is here: https://skia.googlesource.com/buildbot/+/master/autoroll/README.md If the roll is causing failures, please contact the current sheriff, who should be CC'd on the roll, and stop the roller if necessary. TBR=jonahr@google.com Change-Id: I04e0b7ee52f1c981eb42d20dcd045eff4812c1f1 Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1673826Reviewed-by:
angle-autoroll <angle-autoroll@skia-public.iam.gserviceaccount.com> Commit-Queue: angle-autoroll <angle-autoroll@skia-public.iam.gserviceaccount.com>
-
- 21 Jun, 2019 10 commits
-
-
Jamie Madill authored
We could expand this functionality in the future to only update the pipeline when a signficiant bit of pipeline state changes. For now this is mostly a proof of concept. Significantly improves performance in the textures microbenchmark. Bug: angleproject:3117 Change-Id: I3957dd2f08d7d89694f94ddbe45eeeeb620bac0a Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1650790Reviewed-by:
Shahbaz Youssefi <syoussefi@chromium.org> Reviewed-by:
-
Tim Van Patten authored
The call to abi::__cxa_demangle() with too small of a char[] can cause segfaults or hangs. This change increases the array from 256 to 4096. Bug: angleproject:3553 Test: Verify validation layers don't segfault/hang Change-Id: I6704ff00bfab62c99eb288f803ccda35a037dd9d Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1670580Reviewed-by:
-
James Dong authored
In GLES 3.0, depth textures where changed to behave like RED textures, but in GLES 2.0 with ARB/OES_depth_texture, they were treated as luminance textures. This change produces the expected behavior on GLES 2.0 for the GL backend and on GLES 3.0 for the Vulkan backend. Bug: angleproject:3540 Change-Id: Iad6b4e03cf947b27eb97dbb10419bc8bcdb11024 Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1666363Reviewed-by:
-
Geoff Lang authored
Android devices will often fail to allocate textures this large and it is prefereable to return an error from ANGLE instead of losing the context due to an out of memory error. This mirrors the limits Chrome applies with the max_texture_size_limit_4096 and max_msaa_sample_count_4 workarounds that are broadly applied to Android. BUG=882580 Change-Id: I63890baa8712f13b37c607fa475432e67e9384a4 Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1351357 Commit-Queue: Geoff Lang <geofflang@chromium.org> Reviewed-by:
-
Geoff Lang authored
This workaround is always enabled in Chrome, do the same in ANGLE. BUG=882580 Change-Id: I2c01f34a589b07bd0035d7408be884f7a51b1706 Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1666699Reviewed-by:
-
Tobin Ehlis authored
Bug: chromium:977330 Change-Id: I25b2ac11b8ee9b5c084bd828a4d0ddf2b8c1b3eb Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1670092 Commit-Queue: Tobin Ehlis <tobine@google.com> Reviewed-by:
-
Shahbaz Youssefi authored
Issue exposed by a transform feedback test. Since we already create a scissor to match the viewport (be it 0-sized or not), we can create the Vulkan viewport with a small width/height if they are 0, as that's not allowed by Vulkan. Bug: angleproject:3205 Change-Id: Ib23d8b9f5e84447be2c4b7e8aff805e7d9e57323 Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1665352 Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org> Reviewed-by:
-
Geoff Lang authored
The ANGLE Vulkan backend is now thread safe for non-share group contexts. This means that a global GL lock only adds overhead for most use cases. Remove the angle_force_thread_safety gn argument. BUG=angleproject:2464 Change-Id: Ic6ba89e18b46e5dd72aa83d0f409097441fcca3a Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1635749 Commit-Queue: Geoff Lang <geofflang@chromium.org> Reviewed-by:
-
Shahbaz Youssefi authored
Implemented as part of an experiment. Allows easier buffer-related testing by providing a quick way of filling the buffer with dummy data. Bug: angleproject:3205 Change-Id: Ice8cfd0c2566c91a5fb10aaea57985d671d0e7b7 Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1665351 Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org> Reviewed-by:
-
angle-autoroll authored
https://chromium.googlesource.com/external/github.com/KhronosGroup/SPIRV-Tools.git/+log/2090d7a2d26c..7c294608ca19 Created with: gclient setdep -r ./third_party/spirv-tools/src@7c294608ca19 The AutoRoll server is located here: https://autoroll.skia.org/r/spirv-tools-angle-autoroll Documentation for the AutoRoller is here: https://skia.googlesource.com/buildbot/+/master/autoroll/README.md If the roll is causing failures, please contact the current sheriff, who should be CC'd on the roll, and stop the roller if necessary. TBR=jonahr@google.com Change-Id: I92f99e50eaa01782b74f311b1f0d77d4580ab9fa Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1670005Reviewed-by:
-
- 20 Jun, 2019 11 commits
-
-
Yuly Novikov authored
Bug: chromium:973992 Change-Id: Iee5cf6a4b6b0469b81cd0c1a7cd0ebea05060580 Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1670009Reviewed-by:
-
Courtney Goeltzenleuchter authored
The issue occurs that binds the same buffer and calls glDrawElements with the same indices. The offset has been reset in the VertexArrayVk::syncState(), but it doesn't check the actual value in the ContextVk::setupIndexedDraw(). Also corrected case where update via BufferSubData wasn't being sent to the HW. Bug: angleproject:3362 Change-Id: I0f7d2fc162bc8f1c36cb09ba689fd27b482b9035 Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1666345 Commit-Queue: Courtney Goeltzenleuchter <courtneygo@google.com> Reviewed-by:
-
James Dong authored
They can be re-enabled with the angle_enable_trace option. Prevents validation errors from being output in NoFixture tests, as the default platform methods aren't overridden. Bug: angleproject:3525 Change-Id: I77342230b6c7cd167cfc3169f09d4e74b4579244 Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1669848Reviewed-by:
-
Michael Spang authored
Since a6242e4d ("Vulkan: Support submitting multiple semaphores"), we can submit up to 2 semaphores at a time, but pWaitDstStageMask is still a single value. Change it to an array of two elements to prevent an out of bounds access. Bug: angleproject:3289 Change-Id: I5147802ce350af7b78dbf54cfa4a9519dd495f01 Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1666347 Commit-Queue: Michael Spang <spang@chromium.org> Reviewed-by:
-
Tim Van Patten authored
Implement BufferVk::copySubData(). Bug: angleproject:3194 Test: dEQP-GLES3.functional.buffer.copy.* Test: dEQP-GLES3.functional.negative_api.buffer.copy_buffer_sub_data Test: angle_end2end_tests BufferDataTestES3 Change-Id: I220cd490eb1eb799604e217b3e377af3fd431d97 Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1648669 Commit-Queue: Tim Van Patten <timvp@google.com> Reviewed-by:
-
Jamie Madill authored
This captures a bit more information about where the call time goes during a flush. It can show at least on desktop NVIDIA that we spend a fair amount of time in vkQueuePresentKHR. Bug: angleproject:3117 Change-Id: I2d0195b9338bcac80e8dd8cfb550402271f286f9 Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1650787 Commit-Queue: Jamie Madill <jmadill@chromium.org> Reviewed-by:
-
Jamie Madill authored
We can use the DIRTY_BIT_IMPLEMENTATION internal dirty bit in the gl::Texture class to force calls to ContextVk::syncState. In syncState we can ensure we call ensureImageInitialized before we get to the ContextVk. This in turn means we can remove some of the command graph breaks from TextureVk. We need to make sure the dirty bits are propagated to EGL Image siblings with this method. This fixes a potential implementation issue with EGL images with the GL back-end. Also makes a state change test a little better by removing some of the VAO and program init calls before the draw. Improves perf on the texture change microbenchmark by 12.5%. Bug: angleproject:3539 Bug: angleproject:3117 Change-Id: I2b5481690801fa98f859a6c02e3f4b974590cd3d Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1663839 Commit-Queue: Jamie Madill <jmadill@chromium.org> Reviewed-by:
-
Jamie Madill authored
Requires that we update the TRACE_EVENT macros to accept a platform as an argument. The refactor isn't complete. In order to finish we'd need to ensure we have the Display's PlatformMethods available at all sites. Unblocks adding trace events directly in the perf tests. Bug: angleproject:1892 Bug: angleproject:3117 Change-Id: Iee0ca086ccfe23acab3fc186fb042f018711a94c Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1664794 Commit-Queue: Jamie Madill <jmadill@chromium.org> Reviewed-by:
-
Jiacheng Lu authored
Bug: angleproject:3547 Change-Id: I83e491fb0121f0a8473630ead6e48b062ae06da5 Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1668234Reviewed-by:
-
Shahbaz Youssefi authored
The keys in vk_format_map.json were incorrect. The format table generation script is enhanced to give an error in such a case. This revealed a number of entries with no corresponding angle format, which are also removed. Bug: angleproject:2898 Bug: angleproject:3190 Change-Id: I32caf2d2a8abb6f76b25436725670a8e84d576fb Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1666700 Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org> Reviewed-by:
-
Shahbaz Youssefi authored
By using the helper isTransformFeedbackActive* functions. Bug: angleproject:3205 Change-Id: I50b3e9db427da9f65c3e2b50d7847464e8c28202 Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1665353 Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org> Reviewed-by:
-
- 19 Jun, 2019 6 commits
-
-
Geoff Lang authored
BUG=angleproject:1944 Change-Id: I123bb4f97f947272364c765b18d11dba476eee01 Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1666696 Commit-Queue: Geoff Lang <geofflang@chromium.org> Reviewed-by:
-
James Dong authored
Corrects texture swizzle in Vulkan backend. Previously user-supplied swizzle was applied before internal swizzle; this change applies the internal swizzle first, causing the correct behavior. Also recreates image views when swizzle is dirty. Test: ./angle_deqp_gles3_no_gtest --deqp-egl-display-type=angle-vulkan -n 'dEQP-GLES3.functional.texture.swizzle.*' Test: ./angle_end2end_tests --gtest_filter='SwizzleTest.*/ES3_Vulkan' Bug: angleproject:3212 Change-Id: Ie2c44b479da5c19ba744ace7562a73c944f97a49 Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1660909 Commit-Queue: Courtney Goeltzenleuchter <courtneygo@google.com> Reviewed-by:
-
angle-autoroll authored
https://chromium.googlesource.com/external/github.com/KhronosGroup/glslang/+log/11805e41d992..f9d08a25fbe1 Created with: gclient setdep -r ./third_party/glslang/src@f9d08a25fbe1 The AutoRoll server is located here: https://autoroll.skia.org/r/glslang-angle-autoroll Documentation for the AutoRoller is here: https://skia.googlesource.com/buildbot/+/master/autoroll/README.md If the roll is causing failures, please contact the current sheriff, who should be CC'd on the roll, and stop the roller if necessary. TBR=jonahr@google.com Change-Id: Iaa82e2e9fadbbd02b8dae865d61c7f1217746a8e Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1666931Reviewed-by:
-
Stephen Martinis authored
D3D11 cannot allow the same (sub)-resource bound as both a SRV and an UAV at the same time. Unset conflicting SRVs and UVAs between render pipeline and compute pipeline. Bug: angleproject:3152 TEST=angle_end2end_tests.ComputeShaderTest.* Change-Id: I9cb8b902edbf987166a57af314af6b21a6874998 Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1576504Reviewed-by:
-
Michael Spang authored
This this was missed because it doesn't add its own enum value and instead reuses GL_NONE. Bug: chromium:976374 Test: chrome --enable-features=UseSkiaRenderer,UiGpuRasterization \ --enable-gpu-rasterization \ --enable-oop-rasterization \ --use-vulkan=native \ --use-gl=angle \ --use-angle=vulkan Change-Id: I73437211fef3253801e899cb6e50a8e18865cc65 Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1665329 Commit-Queue: Michael Spang <spang@chromium.org> Reviewed-by:
-
Jiawei Shao authored
This patch adds the support of overwriting angle_root to the one defined outside ANGLE, which is useful for other projects which intend to configure ANGLE in their GN builds. Bug: angleproject:3263 Change-Id: I70a56a3d0afdca2eb66fb9619f0dd4d3efc58dee Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1522912 Commit-Queue: Jiawei Shao <jiawei.shao@intel.com> Reviewed-by:
-
- 18 Jun, 2019 4 commits
-
-
Jamie Madill authored
This reverts commit 9aa865a6. Reason for revert: Seems to be causing crashes on Android in glmark2. Might also be responsible for a perf regression. Bug: angleproject:3542 Original change's description: > Vulkan: Add missing command pool reset. > > We were still calling destroy in checkCompletedCommands. We can > instead call reset. > > May not significantly impact performance. > > Bug: angleproject:3489 > Change-Id: I61a2cb154faac56dc78a49678dd822d81ce16ad2 > Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1647029 > Reviewed-by: Jamie Madill <jmadill@chromium.org> > Reviewed-by: Tobin Ehlis <tobine@google.com> > Commit-Queue: Jamie Madill <jmadill@chromium.org> TBR=tobine@google.com,jmadill@chromium.org # Not skipping CQ checks because original CL landed > 1 day ago. Bug: angleproject:3489 Change-Id: Icb0e99f18ecab09798654605973958bd2b263002 Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1665748Reviewed-by:
-
Yuly Novikov authored
Also update doc. Bug: chromium:973992 Change-Id: I59a27935d2bcd438df025945b885384b1d5b0792 Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1665328 Commit-Queue: Yuly Novikov <ynovikov@chromium.org> Reviewed-by:
-
Geoff Lang authored
The Android SurfaceTexture API has to be initialized with a texture id which Chrome has to query from an ANGLE external texture. It also rebinds and sets the texture dimensions on calls to SurfaceTexture.updateTexImage so ANGLE must be notified about these changes so that state tracking and validation continue to function. BUG=967410 Change-Id: I92e9077f75835b088da3a8caffb3ff40e9ad0361 Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1630293 Commit-Queue: Geoff Lang <geofflang@chromium.org> Reviewed-by:
-
Jonah Ryan-Davis authored
NVIDIA Shield do not implement Vulkan, which is used to populate vendor IDs in SystemInfo. The vendorIDs are used to check IsNVIDIA(). This CL adds a workaround so IsNVIDIA() returns true for NVIDIA Shield. Bug: angleproject:3541 Change-Id: Ib36dd376207163acfc8d3a0e341f07a7d6aa3122 Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1663080 Commit-Queue: Jonah Ryan-Davis <jonahr@google.com> Reviewed-by:
-
