-
Notifications
You must be signed in to change notification settings - Fork 274
Pull requests: JuliaGPU/CUDA.jl
Author
Label
Projects
Milestones
Reviews
Assignee
Sort
Pull requests list
CompatHelper: bump compat for CUDACore to 6, (keep existing compat)
#3144
opened May 17, 2026 by
github-actions
Bot
Loading…
Add support for family- and architecture-specific features
#3124
opened Apr 30, 2026 by
AntonOresten
Contributor
Loading…
Added Base.similar methods for CuSparseMatrixCOO and BSR
#3114
opened Apr 21, 2026 by
rainerrodrigues
Contributor
Loading…
Use Republic.jl to forward subpackage APIs
#3103
opened Apr 15, 2026 by
AntonOresten
Contributor
•
Draft
Add AnyCuDeviceArray variations and CuScalar
cuda array
Stuff about CuArray.
speculative
Not sure about this one yet.
#2849
opened Aug 22, 2025 by
moukle
Loading…
fixes the
kron implementation for sparse + diagonal matrix
#2804
opened Jun 27, 2025 by
tam724
Contributor
Loading…
Try fast linear indexes for KA
enhancement
New feature or request
needs changes
Changes are needed.
performance
How fast can we go?
Allow disabling the linking of libdevice in CUDACompilerParams
enhancement
New feature or request
needs changes
Changes are needed.
speculative
Not sure about this one yet.
make CUDA randn work with Zygote
enhancement
New feature or request
needs changes
Changes are needed.
WIP: Native I/O.
cuda kernels
Stuff about writing CUDA kernels.
speculative
Not sure about this one yet.
High Level Wrapper for Fused Matmul + Bias + Activation
cuda libraries
Stuff about CUDA library wrappers.
enhancement
New feature or request
Support FFT adjoint plans and test
cuda libraries
Stuff about CUDA library wrappers.
enhancement
New feature or request
#2073
opened Sep 4, 2023 by
gaurav-arya
•
Draft
Add contract through FastmathOverlays.jl
cuda kernels
Stuff about writing CUDA kernels.
enhancement
New feature or request
WIP: Add an index typevar to CuDeviceArray.
enhancement
New feature or request
help wanted
Extra attention is needed
performance
How fast can we go?
Add an experimental opaque closure type.
cuda kernels
Stuff about writing CUDA kernels.
enhancement
New feature or request
speculative
Not sure about this one yet.
Previous Next
ProTip!
Exclude everything labeled
bug with -label:bug.