[vulkan] glsl shaders relaxed precision mode to cmake option #43076
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
[ghstack-poisoned]
Differential Revision: [D23143354](https://our.internmc.facebook.com/intern/diff/D23143354) Introducing cmake option USE_VULKAN_RELAXED_PRECISION that controls which precision will be used in Vullkan shaders. Default: `highp`: 32-bit two’s complement for integers, 32-bit IEEE 754 floating-point for float if USE_VULKAN_RELAXED_PRECISION=ON: `mediump`: RelaxedPrecision, on storage variable and operation All shaders starts from `#define PRECISION $precision` where `$precision` will be substituted on codegen step (build). To be able to use CodeTemplate from `aten/src/ATen/code_template.py` for substitution processing: moving `aten/src/ATen/native/vulkan/gen_glsl.py` to `aten/src/ATen/gen_vulkan_glsl.py` as without it can not import python code from parent directories. `aten/src/ATen/gen_vulkan_glsl.py`, `aten/src/ATen/gen_vulkan_glsl.py` added `env` argument that specifies list of key-value pairs for substitution, by default env is `{"precision": "highp"}` and can be overriden adding `--env precision=mediump` Some fixes in cpp to fix build for `USE_VULKAN_SHADERC_RUNTIME=ON` [ghstack-poisoned]
💊 CI failures summary and remediationsAs of commit c7009c5 (more details on the Dr. CI page):
1 job timed out:
❄️ 2 failures tentatively classified as flakybut reruns have not yet been triggered to confirm:
|
Differential Revision: [D23143354](https://our.internmc.facebook.com/intern/diff/D23143354) Introducing cmake option USE_VULKAN_RELAXED_PRECISION that controls which precision will be used in Vullkan shaders. Default: `highp`: 32-bit two’s complement for integers, 32-bit IEEE 754 floating-point for float if USE_VULKAN_RELAXED_PRECISION=ON: `mediump`: RelaxedPrecision, on storage variable and operation All shaders starts from `#define PRECISION $precision` where `$precision` will be substituted on codegen step (build). To be able to use CodeTemplate from `aten/src/ATen/code_template.py` for substitution processing: moving `aten/src/ATen/native/vulkan/gen_glsl.py` to `aten/src/ATen/gen_vulkan_glsl.py` as without it can not import python code from parent directories. `aten/src/ATen/gen_vulkan_glsl.py`, `aten/src/ATen/gen_vulkan_glsl.py` added `env` argument that specifies list of key-value pairs for substitution, by default env is `{"precision": "highp"}` and can be overriden adding `--env precision=mediump` Some fixes in cpp to fix build for `USE_VULKAN_SHADERC_RUNTIME=ON` [ghstack-poisoned]
Differential Revision: [D23143354](https://our.internmc.facebook.com/intern/diff/D23143354) Introducing cmake option USE_VULKAN_RELAXED_PRECISION that controls which precision will be used in Vullkan shaders. Default: `highp`: 32-bit two’s complement for integers, 32-bit IEEE 754 floating-point for float if USE_VULKAN_RELAXED_PRECISION=ON: `mediump`: RelaxedPrecision, on storage variable and operation All shaders starts from `#define PRECISION $precision` where `$precision` will be substituted on codegen step (build). To be able to use CodeTemplate from `aten/src/ATen/code_template.py` for substitution processing: moving `aten/src/ATen/native/vulkan/gen_glsl.py` to `aten/src/ATen/gen_vulkan_glsl.py` as without it can not import python code from parent directories. `aten/src/ATen/gen_vulkan_glsl.py`, `aten/src/ATen/gen_vulkan_glsl.py` added `env` argument that specifies list of key-value pairs for substitution, by default env is `{"precision": "highp"}` and can be overriden adding `--env precision=mediump` Some fixes in cpp to fix build for `USE_VULKAN_SHADERC_RUNTIME=ON` [ghstack-poisoned]
Differential Revision: [D23143354](https://our.internmc.facebook.com/intern/diff/D23143354) Introducing cmake option USE_VULKAN_RELAXED_PRECISION that controls which precision will be used in Vullkan shaders. Default: `highp`: 32-bit two’s complement for integers, 32-bit IEEE 754 floating-point for float if USE_VULKAN_RELAXED_PRECISION=ON: `mediump`: RelaxedPrecision, on storage variable and operation All shaders starts from `#define PRECISION $precision` where `$precision` will be substituted on codegen step (build). To be able to use CodeTemplate from `aten/src/ATen/code_template.py` for substitution processing: moving `aten/src/ATen/native/vulkan/gen_glsl.py` to `aten/src/ATen/gen_vulkan_glsl.py` as without it can not import python code from parent directories. `aten/src/ATen/gen_vulkan_glsl.py`, `aten/src/ATen/gen_vulkan_glsl.py` added `env` argument that specifies list of key-value pairs for substitution, by default env is `{"precision": "highp"}` and can be overriden adding `--env precision=mediump` Some fixes in cpp to fix build for `USE_VULKAN_SHADERC_RUNTIME=ON` [ghstack-poisoned]
Differential Revision: [D23143354](https://our.internmc.facebook.com/intern/diff/D23143354) Introducing cmake option USE_VULKAN_RELAXED_PRECISION that controls which precision will be used in Vullkan shaders. Default: `highp`: 32-bit two’s complement for integers, 32-bit IEEE 754 floating-point for float if USE_VULKAN_RELAXED_PRECISION=ON: `mediump`: RelaxedPrecision, on storage variable and operation All shaders starts from `#define PRECISION $precision` where `$precision` will be substituted on codegen step (build). To be able to use CodeTemplate from `aten/src/ATen/code_template.py` for substitution processing: moving `aten/src/ATen/native/vulkan/gen_glsl.py` to `aten/src/ATen/gen_vulkan_glsl.py` as without it can not import python code from parent directories. `aten/src/ATen/gen_vulkan_glsl.py`, `aten/src/ATen/gen_vulkan_glsl.py` added `env` argument that specifies list of key-value pairs for substitution, by default env is `{"precision": "highp"}` and can be overriden adding `--env precision=mediump` Some fixes in cpp to fix build for `USE_VULKAN_SHADERC_RUNTIME=ON` [ghstack-poisoned]
Differential Revision: [D23143354](https://our.internmc.facebook.com/intern/diff/D23143354) Introducing cmake option USE_VULKAN_RELAXED_PRECISION that controls which precision will be used in Vullkan shaders. Default: `highp`: 32-bit two’s complement for integers, 32-bit IEEE 754 floating-point for float if USE_VULKAN_RELAXED_PRECISION=ON: `mediump`: RelaxedPrecision, on storage variable and operation All shaders starts from `#define PRECISION $precision` where `$precision` will be substituted on codegen step (build). To be able to use CodeTemplate from `aten/src/ATen/code_template.py` for substitution processing: moving `aten/src/ATen/native/vulkan/gen_glsl.py` to `aten/src/ATen/gen_vulkan_glsl.py` as without it can not import python code from parent directories. `aten/src/ATen/gen_vulkan_glsl.py`, `aten/src/ATen/gen_vulkan_glsl.py` added `env` argument that specifies list of key-value pairs for substitution, by default env is `{"precision": "highp"}` and can be overriden adding `--env precision=mediump` Some fixes in cpp to fix build for `USE_VULKAN_SHADERC_RUNTIME=ON` [ghstack-poisoned]
AshkanAliabadi
approved these changes
Sep 16, 2020
Contributor
|
@IvanKobzarev merged this pull request in 6debe82. |
xuzhao9
pushed a commit
that referenced
this pull request
Sep 18, 2020
Summary: Pull Request resolved: #43076 Test Plan: Imported from OSS Reviewed By: AshkanAliabadi Differential Revision: D23143354 Pulled By: IvanKobzarev fbshipit-source-id: 7b3ead1e63cf8acf6e8e547080a8ead7a2db994b
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Stack from ghstack:
Differential Revision: D23143354
Introducing cmake option USE_VULKAN_RELAXED_PRECISION that controls which precision will be used in Vullkan shaders.
Default:
highp: 32-bit two’s complement for integers, 32-bit IEEE754 floating-point for float
if USE_VULKAN_RELAXED_PRECISION=ON:
mediump: RelaxedPrecision, on storage variable and operationAll shaders starts from
#define PRECISION $precisionwhere$precisionwill be substituted on codegen step (build).To be able to use CodeTemplate from
aten/src/ATen/code_template.pyfor substitution processing:moving
aten/src/ATen/native/vulkan/gen_glsl.pytoaten/src/ATen/gen_vulkan_glsl.pyas without it can not import python code from parent directories.aten/src/ATen/gen_vulkan_glsl.py,aten/src/ATen/gen_vulkan_glsl.pyaddedenvargument that specifies list of key-value pairs for substitution, by default env is{"precision": "highp"}and can be overriden adding--env precision=mediumpSome fixes in cpp to fix build for
USE_VULKAN_SHADERC_RUNTIME=ON