Skip to content

Navigation Menu

Appearance settings

- AI CODE CREATION
  GitHub CopilotWrite better code with AI
  GitHub Copilot appDirect agents from issue to merge
  MCP Registry^NewIntegrate external tools
- DEVELOPER WORKFLOWS
  ActionsAutomate any workflow
  CodespacesInstant dev environments
  IssuesPlan and track work
  Code ReviewManage code changes
- APPLICATION SECURITY
  GitHub Advanced SecurityFind and fix vulnerabilities
  Code securitySecure your code as you build
  Secret protectionStop leaks before they start
- EXPLORE
  Why GitHub
  Documentation
  Blog
  Changelog
  Marketplace
View all features
- BY COMPANY SIZE
  Enterprises
  Small and medium teams
  Startups
  Nonprofits
- BY USE CASE
  App Modernization
  DevSecOps
  DevOps
  CI/CD
  View all use cases
- BY INDUSTRY
  Healthcare
  Financial services
  Manufacturing
  Government
  View all industries
View all solutions
- EXPLORE BY TOPIC
  AI
  Software Development
  DevOps
  Security
  View all topics
- EXPLORE BY TYPE
  Customer stories
  Events & webinars
  Ebooks & reports
  Business insights
  GitHub Skills
- SUPPORT & SERVICES
  Documentation
  Customer support
  Community forum
  Trust center
  Partners
View all resources
- COMMUNITY
  GitHub SponsorsFund open source developers
- PROGRAMS
  Security Lab
  Maintainer Community
  Accelerator
  GitHub Stars
  Archive Program
- REPOSITORIES
  Topics
  Trending
  Collections
- ENTERPRISE SOLUTIONS
  Enterprise platformAI-powered developer platform
- AVAILABLE ADD-ONS
  GitHub Advanced SecurityEnterprise-grade security features
  Copilot for BusinessEnterprise-grade AI features
  Premium SupportEnterprise-grade 24/7 support
Pricing

Search code, repositories, users, issues, pull requests...

Search

Clear

Search syntax tips

Provide feedback

We read every piece of feedback, and take your input very seriously.

Include my email address so I can be contacted

Saved searches

Use saved searches to filter your results more quickly

Name

Query

To see all available qualifiers, see our documentation.

Appearance settings

You signed in with another tab or window. Reload to refresh your session. You signed out in another tab or window. Reload to refresh your session. You switched accounts on another tab or window. Reload to refresh your session.

Dismiss alert

mratsim / constantine Public

Notifications You must be signed in to change notification settings
Fork 65
Star 495

Code
Issues 80
Pull requests 13
Discussions
Actions
Projects
Wiki
Security and quality
Insights

Additional navigation options

Code
Issues
Pull requests
Discussions
Actions
Projects
Wiki
Security and quality
Insights

WebGPU GPU compiler backend improvements (and small CUDA)#565

Merged

mratsim merged 87 commits into

mastermratsim/constantine:masterfrom

webgpu-improvementsmratsim/constantine:webgpu-improvementsCopy head branch name to clipboard

Sep 11, 2025

Conversation Commits87 (87)Checks Files changed

Merged

WebGPU GPU compiler backend improvements (and small CUDA)#565
mratsim merged 87 commits into
masterfrom
webgpu-improvements

Commits

Commits on Sep 9, 2025

handle `nnkSym` in `getInnerPointerType`
Vindaar
committed
fix `constructPtrSignature` after change from nil -> gtVoid for iTyp
Vindaar
committed
handle `gpuCast` in determineSymKind/Mutability/Ident
Vindaar
committed
allow determination of GPU type to fail in `nimToGpuType`
Vindaar
committed
handle `UncheckedArray` in `gpuTypeToString` on CUDA backend
Vindaar
committed
rename constant to set workgroup size
Vindaar
committed
minor cleanup
Vindaar
committed
correctly handle `gtUA` (UncheckedArray) on CUDA backend
Vindaar
committed
fix access of type for left ident in assignment
Vindaar
committed
rewrite compound assignment operators in all functions
Vindaar
committed
extend `GpuFieldInit` by a type field
Vindaar
committed
make sure to update symbols in global functions too
Vindaar
committed
remove local `determineIdent` from `genWebGpu`
Vindaar
committed
fix tree representation of GpuAst
Vindaar
committed
add `mpairs`, `pairs` iterator for `GpuAst`
Vindaar
committed
support default initialization in obj constr fields
Vindaar
committed
correctly handle `var foo {.constant.}` variables by ⇒ globals
Vindaar
committed
move rewriting of compound assignment out of `storagePass`
Vindaar
committed
implement lifting of struct pointer fields
Vindaar
committed
add helper to check code for validity
Vindaar
committed
add support for full Nim generics in the context of GPU code
Vindaar
committed
implement support for type aliases
Vindaar
committed
allow "pulling in" procs from outside `cuda` scope
Vindaar
committed
add pretty printer for GpuType
Vindaar
committed
store all types in `ctx.types` table, support 'pulling' types...
Vindaar
committed
support generic instantiations, producing unique WGSL types
Vindaar
committed
add `{.builtin.}` pragma intended to specify builtin procs in CUDA/WGSL
Vindaar
committed
fix context version for `cuModuleGetGlobal`
Vindaar
committed
allow to compile with `-d:debugCuda` to compile in debug mode
Vindaar
committed
ignore `varargs` pragma in procs
Vindaar
committed
make sure statically sized arrays are copied
Vindaar
committed
make sure to catch `CUdeviceptr` arguments
Vindaar
committed
change CUDA codegen to similar style as WGSL
Vindaar
committed
improve error message for `Dot` node if not ident/deref in WGSL
Vindaar
committed
always also inject `num_workgroups` argument into WGSL globals
Vindaar
committed
rename WGSL `storagePass` to `preprocess`
Vindaar
committed
remove `gpuTypeToString`, `genFunctionType` from `backends`
Vindaar
committed
[CUDA] fix indentation for variable declarations
Vindaar
committed
improve type deduction for types that look generic
Vindaar
committed
clone `forwardDeclare` in GpuAst
Vindaar
committed
support tuples by mapping to object types
Vindaar
committed
support Nim's implicit `result` variable
Vindaar
committed
first steps towards supporting expressions
Vindaar
committed
move `cIsExpr` in clone to correct branch
Vindaar
committed
also catch tuple types from `nnkBracketExpr`
Vindaar
committed
handle `inline` pragma same as `forceinline`
Vindaar
committed
fix detection of custom `result` by looking into `gpuBlock`
Vindaar
committed
fix handling of `skipSemicolon` for nested usage
Vindaar
committed
if last expression _is_ return, we don't need `result` variable
Vindaar
committed
remove debug output
Vindaar
committed
doc comment updates for procs in gpu_field_ops
Vindaar
committed
do not emit semicolon in binary operand child nodes
Vindaar
committed
handle function overloads by using `iSym` if encountered
Vindaar
committed
ignore `magic` pragma procs
Vindaar
committed
ignore forward declarations in Nim -> GpuAst
Vindaar
committed
add BigInt comparison and `toCanonical` conversion
Vindaar
committed
add FIPS Montgomery multiplication for fields without spare bits
Vindaar
committed
[cuda] for now emit `constexpr` for a `gpuConstexpr` (i.e. Nim `const`)
Vindaar
committed
[cuda] handle generic instantiations and `UncheckedArray`
Vindaar
committed
[wgsl] do not suffix generic inst types if they have no args
Vindaar
committed
handle `raises`, `noinit` pragmas
Vindaar
committed
use `getTypeName` for tuple type fields
Vindaar
committed
map `ntyString` to string explicitly in gpu type kinds
Vindaar
committed
map ntyUnused2 (lent T) to ptr T
Vindaar
committed
generate type names for bracket expressions e.g. generics
Vindaar
committed
improve generic inst -> gpu type by handling nnkSym
Vindaar
committed
overwrite function names of problematic identifiers
Vindaar
committed
refactor proc signature parsing & unify adding types to `ctx.types`
Vindaar
committed
handle recursive calls in GPU code
Vindaar
committed
fix `modadd` in debug builds on CUDA
Vindaar
committed
fix reassignment of `types` in RT codegen for aliases
Vindaar
committed
handle `nnkObjConstr` correctly when encountering gtGenericInst
Vindaar
committed
do not emit `f` suffix for float literals
Vindaar
committed
[wgsl] append `;` for aliases
Vindaar
committed
rewrite infix as `gpuCall` if arguments not basic types
Vindaar
committed
support arbitrary values in array literals
Vindaar
committed
handle generic instantiation with actual initializations
Vindaar
committed
fix top level type definitions
Vindaar
committed
[cuda] add pass to strip `deref` if found inside `index` for pointers
Vindaar
committed
remove old comment
Vindaar
committed
better handle replacement of derefs
Vindaar
committed
remove Nim gensym'd suffix from `tmpTuple` variables
Vindaar
committed

Commits on Sep 10, 2025

replace single letter strings by characters
Vindaar
committed
extend `maybeAddType` to find types behind Ptr/UA/Array
Vindaar
committed
move all `builtins` into separate files, one for each backend
Vindaar
committed
update doc comment of `cuda` macro
Vindaar
committed
improve explanation of `typeOfIndex` in CUDA's `getType` helper
Vindaar
committed

Footer

© 2026 GitHub, Inc.

Footer navigation

Terms
Privacy
Security
Status
Community
Docs
Contact

You can’t perform that action at this time.