-
Notifications
You must be signed in to change notification settings - Fork 694
SPIR‐V CodeGen
DirectXShaderCompiler also supports translating HLSL into the SPIR-V intermediate language. This wiki page details how to build and invoke SPIR-V CodeGen, and also how to contribute.
For how HLSL features are mapped into SPIR-V, please see the HLSL to SPIR-V mapping doc.
Note: SPIR-V CodeGen is mature-ish now.
Before building, please make sure that you have all the prerequisite tools listed on the Building Sources page.
Assume that you clone the source code into <dxc-src-dir>
and generate the
artifacts into <dxc-bin-dir>
:
git clone [email protected]:Microsoft/DirectXShaderCompiler.git <dxc-src-dir>
cd <dxc-src-dir>
git submodule update --init
# Run the following if Windows Driver Kit is not installed
python .\utils\hct\hctgettaef.py
.\utils\hct\hctstart.cmd <dxc-src-dir> <dxc-bin-dir> # Set up environment
.\utils\hct\hctbuild.cmd -spirv # Configure and build
If the build is successful, the generated dxc.exe
should have SPIR-V CodeGen
built in.
To translate HLSL source file <hlsl-src-file>
into SPIR-V binary
<spirv-bin-file>
:
\path\to\dxc.exe -spirv -T <target-prfile> -E <entry-point>
<hlsl-src-file> -Fo <spirv-bin-file>
Apart from the common command-line options like -O
, -Fo
, -Fh
, there are
a few Vulkan/SPIR-V specific command-line options that
you can use and dxc.exe -help
explains them.
The SPIR-V CodeGen is mainly contributed and maintained by the @google/shaderc team. But contributions to the SPIR-V CodeGen are definitely very welcome! :)
In addition to the DirectXShaderCompiler contributing guidelines, please also make sure to follow the following guidelines.
When configuring, please make sure to enable building SPIR-V tests:
.\utils\hct\hctbuild.cmd -s -spirvtest # Configure with SPIR-V CodeGen & tests
.\utils\hct\hctbuild.cmd -b # Build
For each pull request, please make sure
- Tests are written to cover the modifications.
- The HLSL to SPIR-V mapping doc is updated for newly supported features.
We use GoogleTest as the unit test and CodeGen test framework. Appveyor will be used to check regression of all pull requests.
Read more about the SPIR-V CodeGen internals in the following section.
Various designs are driven by technical considerations together with the following guidelines for good citizenship within DirectXShaderCompiler:
- Conduct minimal changes to existing interfaces and libraries
- Perfer less intrusive solutions
The general approach is to translate frontend AST directly into SPIR-V binary. We choose this approach considering that
- Frontend AST is much more higher-level than DXIL. For example, DXIL scalarized vectors but SPIR-V has native support.
- DXIL has widely different semantics than Vulkan flavor of SPIR-V. For example, structured control flow is not preserved in DXIL but SPIR-V for Vulkan requires it.
- Frontend AST perserves the information in the source code better.
- Also, the right place to generate error messages is in Clang's semantic analysis step, which is when the compiler is still processing the AST.
Therefore, it is easier to go from frontend AST to SPIR-V than from DXIL since we do not need to rediscover certain information.
Translating frontend AST directly into SPIR-V binary precludes the usage of existing LLVM optimization passes. This is expected since there are also subtle semantics differences between SPIR-V and LLVM IR. Certain concepts in SPIR-V do not have direct corresponding representation in LLVM IR and there are no existing translation schemes handling the differences. Using vanilla LLVM optimization passes will likely violate the requirements of SPIR-V and results in invalid SPIR-V modules.
Instead, optimizations are available in the SPIRV-Tools project.
On the library side, this means introducing a new ASTFrontendAction
and a
SPIR-V module builder. The new frontend action will traverse the AST and call
the SPIR-V module builder to construct SPIR-V words. These code should be
placed at tools/clang/lib/SPIRV
and packed into one library (or multiple
libraries in the future).
Detailed design will be revised to accommodate more and more HLSL features. At the moment, we have:
EmitSPIRVAction
|
| creates
V
SPIRVEmitter
|
| contains
|
+-------------+------------+
| |
| |
V references V
SPIRVContext <------------ ModuleBuilder
|
| contains
V
InstBuilder
|
| depends on
V
WordConsumer
-
SPIRVEmitter
: The derivedASTConsumer
which acts on various frontend AST nodes by calling correspondingModuleBuilder
methods to build SPIR-V modules gradually. -
ModuleBuilder
: Exposes API for constructing SPIR-V modules. Internally it has structured representation of SPIR-V modules, functions, basic blocks as well as various SPIR-V specific structs like entry points, debug names, and so on. -
SPIRVContext
: Responsible for allocation and maintaining the lifetime of objects allocated to represent types, decorations, and others. It is used in conjunction withModuleBuilder
. -
InstBuilder
: The low-level interface for generating SPIR-V words for various SPIR-V instructions. All SPIR-V instructions are eventually serialized viaInstBuilder
. -
WordConsumer
: The consumer of generated SPIR-V words.
A new option, -spirv
, will be added into dxc.exe
for invoking the new
SPIR-V CodeGen action.
GoogleTest is used as both the unit test and the SPIR-V CodeGen test framework.
Unit tests are placed under the tools/clang/unittests/SPIRV/
directory,
while SPIR-V CodeGen tests are placed under the tools/clang/test/CodeGenSPIRV/
directory.
For SPIR-V CodeGen tests, there are two test fixtures: one for checking the whole disassembly of the generated SPIR-V code, the other is FileCheck-like, for partial pattern matching.
-
Whole disassembly check: These tests are in files with suffix
.hlsl2spv
. Each file consists of two parts, HLSL source code input and expected SPIR-V disassembly ouput, delimited by// CHECK-WHOLE-SPIR-V:
. The compiler takes in the whole file as input and compile its into SPIR-V binary. The test fixture then disasembles the SPIR-V binary and compares the disassembly with the expected disassembly listed after// CHECK-WHOLE-SPIR-V
. -
Partial disassembly match: These tests are in files with suffix
.hlsl
. Effcee is used for the stateful pattern matching. Effcee itself depends on a regular expression library, RE2. See Effcee for supportedCHECK
syntax. They are largely the same as LLVM FileCheck.
SPIR-V CodeGen functionality will require two external projects:
SPIRV-Headers (for spirv.hpp11
) and
SPIRV-Tools (for SPIR-V disassembling and optimizations).
These two projects should be checked out under the external/
directory.
The three projects for testing, GoogleTest, Effcee, and RE2, should also be
checked out under the external/
directory.
To add these dependencies as submodules, use the command: git submodule add -f <url> <path>
.
The -f
flag is necessary because the submodupdes are located in the external directory,
which is in the default .gitignore file,
SPIR-V CodeGen functionality will structured as an optional feature in DirectXShaderCompiler. Two new CMake options will be introduced to control the configuring and building SPIR-V CodeGen:
-
ENABLE_SPIRV_CODEGEN
: If turned on, enables the SPIR-V CodeGen functionality. (Default: OFF) -
SPIRV_BUILD_TESTS
: If turned on, enables building of SPIR-V related tests. This option will also implicitly turn onENABLE_SPIRV_CODEGEN
. (Default: OFF)
For building, hctbuild
will be extended with two new switches, -spirv
and -spirvtest
, to turn on the above two options, respectively.
For testing, hcttest spirv
will run all existing tests together with SPIR-V
tests, while htctest spirv_only
will only trigger SPIR-V tests.