| Age | Commit message (Collapse) | Author |
|
* Fix linking issue in slangpy + no mask param for kernels.
* add cuda header changes
* fix
* More correct change of active mask insertion.
---------
Co-authored-by: Yong He <yhe@nvidia.com>
|
|
of registered types. (#2777)
* Create high-order-forward-diff-struct.slang
* Add a small pass to register differentials for differentials of registered types
* Remove type reg pass and apply logic in differentiable type loading instead
* Update slang-ir-autodiff.cpp
* Fix comments.
---------
Co-authored-by: Yong He <yonghe@outlook.com>
|
|
Co-authored-by: Yong He <yhe@nvidia.com>
|
|
* #include an absolute path didn't work - because paths were taken to always be relative.
* Initial support for realtime clock.
* Add realtime-clock render feature where seems appropriate.
* Fixes to make NVAPI compile properly.
Change realtime-clock.slang check to use maths that can't overflow.
|
|
* #include an absolute path didn't work - because paths were taken to always be relative.
* WIP in handling artifacts/SourceFiles.
* Add getUniqueIdentity to IPathArtifactRepresentation
* Simplification because around using the representation to provide the uniqueIdentity.
|
|
attributes (#2773)
* Fix compiler crash if referencing undefined method in the deriavative decorator
* Delete sample.comp
---------
Co-authored-by: Yong He <yonghe@outlook.com>
|
|
* Emit simpler vector element access code
* Fix.
---------
Co-authored-by: Yong He <yhe@nvidia.com>
|
|
Co-authored-by: Yong He <yonghe@outlook.com>
|
|
|
|
* Detect when calls to execv fail
* Use pipe+fcntl instead of pipe2
pipe2 is not available on macOS
* remove outdated comment
* Use _exit in child if execv fails
* Prevent Process::create leaks on some failures, and more robust in setting std streams for child
---------
Co-authored-by: Yong He <yonghe@outlook.com>
|
|
Co-authored-by: Yong He <yhe@nvidia.com>
|
|
Co-authored-by: Yong He <yhe@nvidia.com>
|
|
* Remove extra qualification in cuda device impl
Only MSVC accepts this illegal code
* Enable CUDA render api on unix
---------
Co-authored-by: Yong He <yonghe@outlook.com>
|
|
Co-authored-by: Yong He <yonghe@outlook.com>
|
|
Co-authored-by: Yong He <yhe@nvidia.com>
|
|
Co-authored-by: Yong He <yhe@nvidia.com>
|
|
* Fix IRArrayType emit logic.
* Fix test.
* Fix ast constant folding.
---------
Co-authored-by: Yong He <yhe@nvidia.com>
|
|
* Update checkpoint policy to make obvious recompute decisions.
Also adds an optimization to fold updateElement chains on the same array or struct into a single makeArray or makeStruct.
* Bug fixes around array types with different int typed count.
* change test.
* Fix.
---------
Co-authored-by: Yong He <yhe@nvidia.com>
|
|
|
|
* Add slangpy doc, fix cuda prelude.
* more bug fix.
* fix.
* fix.
* More fix.
* fix.
* f
* fix prelude.
* update prelude.
* update doc
* Update prelude.
* add zeros_like
* update doc.
---------
Co-authored-by: Yong He <yhe@nvidia.com>
|
|
* Create render.slang
* Added higher-order differentiability decorators for built-ins + preliminary tests
* Update diff.meta.slang
* Copy over conformance synthesis code to `DifferentiableTypeConformanceContext`
* Update render.slang
* Fixed 1D warped-area sampling test
* Update warped-sampling-1d.slang
* Remove commented line.
* Change WAS test to use fixed point
* Replaced InterlockedCmpExchange with InterlockedAdd
* Increase fixed point precision
* Reduce floating-point precision by 2 digits to avoid platform-specific problems
* Dropped another digit (just to be safe)
---------
Co-authored-by: Yong He <yonghe@outlook.com>
|
|
* Small fixes and cleanups on CUDA/CPP codegen.
* Disable `legalizeEmptyTypes` for now.
---------
Co-authored-by: Yong He <yhe@nvidia.com>
|
|
* Translate all composed types into tuple types in pyBind.
* Delete temp file.
* Fix get tuple element code emit logic.
---------
Co-authored-by: Yong He <yhe@nvidia.com>
|
|
* #include an absolute path didn't work - because paths were taken to always be relative.
* Fix SlangCompileTarget to keep ordering.
* Add test.
Remove V2 version of interface to access IDownstreamCompiler
Update to slang-llvm which has _chkstk support.
* Update slang.h
Co-authored-by: Ellie Hermaszewska <github@sub.monoid.al>
---------
Co-authored-by: Yong He <yonghe@outlook.com>
Co-authored-by: Ellie Hermaszewska <github@sub.monoid.al>
|
|
(#2739)
* Apply IR simplifcation immediately after specialization to avoid duplicates.
* Update source/slang/slang-ir-specialize.cpp
Co-authored-by: Ellie Hermaszewska <github@sub.monoid.al>
---------
Co-authored-by: Yong He <yhe@nvidia.com>
Co-authored-by: Ellie Hermaszewska <github@sub.monoid.al>
|
|
Co-authored-by: Yong He <yhe@nvidia.com>
|
|
* #include an absolute path didn't work - because paths were taken to always be relative.
* WIP using SourceMap with SourceManager.
* Add a test to check obfuscation map is working.
---------
Co-authored-by: Yong He <yonghe@outlook.com>
|
|
Co-authored-by: Yong He <yhe@nvidia.com>
|
|
* Use c++17 standard
* regenerate VS projects
* Switch to using std::invoke
|
|
|
|
* Add PyTorch C++ binding generation.
* fix
---------
Co-authored-by: Yong He <yhe@nvidia.com>
|
|
|
|
* Added higher-order differentiability decorators for built-ins + preliminary tests
* Update diff.meta.slang
|
|
* #include an absolute path didn't work - because paths were taken to always be relative.
* WIP produce obfuscated source map and write when container is specified.
* Make the sourcemap generated name stable.
|
|
Co-authored-by: Yong He <yhe@nvidia.com>
|
|
* Fix crash.
* Fix `[ForwradDerivative]` on member functions.
* Update comments.
* Fix crash when [BackwardDerivative] is provided but not [ForwardDerivative].
* Allow calling dynamic dispatched generic method from differentiable func.
* Fix.
---------
Co-authored-by: Yong He <yhe@nvidia.com>
|
|
* Fixed type coercion issue with higher-order user defined methods
* Placed associated type lookup method in a loop
* Update high-order-user-defined-derivative.slang
* Revert changes to associated type lookup method
|
|
* Update diff-bwd material test
* Minor update
* Hack handling of primal insts that has a function type.
---------
Co-authored-by: winmad <winmad.wlf@gmail.com>
Co-authored-by: Yong He <yhe@nvidia.com>
|
|
* Fix optimization pass not converging.
* Fix.
* Fix tests.
---------
Co-authored-by: Yong He <yhe@nvidia.com>
|
|
Co-authored-by: Yong He <yhe@nvidia.com>
|
|
* Fix generic lowering.
* Fix generic lowering regression due to IR deduplication.
---------
Co-authored-by: Yong He <yhe@nvidia.com>
|
|
* Bug fixes.
* Fix.
* Only perform autodiff for functions whose derivative is actually used.
* Fix loop optimize bug.
* Fix high order diff.
* Fix trivial diff func generation.
* Fixes.
* Cleanup.
---------
Co-authored-by: Yong He <yhe@nvidia.com>
|
|
* #include an absolute path didn't work - because paths were taken to always be relative.
* WIP source map.
* Split out handling of RttiTypeFuncs to a map type.
* Make RttiTypeFuncsMap hold default impls.
* Slightly more sophisticated RttiTypeFuncsMap
* Source map decoding.
* Fix tabs.
* Fix asserts due to negative values.
* Use less obscure mechanisms in SourceMap.
* Source map decoding.
Simplifying SourceMap usage.
* First attempt at ouputting a source map as part of emit.
* Added support for -source-map option. SourceMap is added to the artifact.
* Small improvements around column calculation in SourceWriter.
* Source Loc obuscation WIP.
* Fix some issues around SourceMap obfuscation.
* Split out obfuscation into its own file.
* Keep obfuscated SourceMap even through serialization bottleneck.
|
|
|
|
* Fix associated type resolution bug.
* Fix.
* Fix language server hinting messed up by breadcrumb nodes.
---------
Co-authored-by: Yong He <yhe@nvidia.com>
|
|
* Fix IntVal type check.
* add test case result.
---------
Co-authored-by: Yong He <yhe@nvidia.com>
|
|
* Add quad texture example.
* delete output image
* remove irrelavent files
* update project files
* fix
* Update example.
* Fix.
* remove out-texture
---------
Co-authored-by: Yong He <yhe@nvidia.com>
|
|
* Add support for emitting cuda kernel and host functions.
* Update test.
* Fix cuda preamble emit.
---------
Co-authored-by: Yong He <yhe@nvidia.com>
|
|
Co-authored-by: Yong He <yhe@nvidia.com>
|
|
* #include an absolute path didn't work - because paths were taken to always be relative.
* WIP source map.
* Split out handling of RttiTypeFuncs to a map type.
* Make RttiTypeFuncsMap hold default impls.
* Slightly more sophisticated RttiTypeFuncsMap
* Source map decoding.
* Fix tabs.
* Fix asserts due to negative values.
* Use less obscure mechanisms in SourceMap.
* Source map decoding.
Simplifying SourceMap usage.
* First attempt at ouputting a source map as part of emit.
* Added support for -source-map option. SourceMap is added to the artifact.
|