开放
Naming inconsistencies in generic dense cuSOLVER functions
good first issue
0 条评论0 个反应0 负责人
议题
开放
Support CuDeviceArray construction with non-Int dims
cuda kernelsgood first issue
1 条评论0 个反应0 负责人
开放
Weird bug with matrix multiplication on Array view
cuda librarieshelp wanted
0 条评论0 个反应0 负责人
开放
Adding two cuarrays of different sizes throws no error
cuda arraygood first issue
0 条评论0 个反应0 负责人
开放
Interactive reflection of caught errors is broken
good first issue
1 条评论0 个反应0 负责人
开放
Summing a vector is faster than summing a multidimensional array
cuda arraygood first issueperformance
1 条评论0 个反应0 负责人
开放
@captured with scalar arguments fails with CUDA 5.7
good first issue
4 条评论0 个反应0 负责人
开放
Atomics: configurable scope (for multi-device unified memory)
cuda kernelshelp wanted
3 条评论0 个反应0 负责人
开放
`view(::CuArray, ...)` is sometimes type-unstable:
cuda arraygood first issue
6 条评论0 个反应0 负责人
开放
Improve error message on PPC
enhancementgood first issueinstallation
3 条评论0 个反应0 负责人
开放
Inconsistency between CUDA.jl and Base for sum!
bugcuda arraygood first issue
1 条评论0 个反应0 负责人
开放
Wrong source paths in debug info
bugcuda kernelsgood first issue
0 条评论0 个反应0 负责人
开放
FFT on real array with `plan_fft(x)` is 50% slower than PyTorch
help wantedneeds informationperformance
1 条评论0 个反应0 负责人
开放
CUDA kernel launches should use their own RNG
cuda kernelsenhancementgood first issue
2 条评论0 个反应0 负责人
开放
Adapt + strictly-typed fields can trigger confusing errors
enhancementgood first issue
0 条评论0 个反应0 负责人
开放
CUSPARSE: Float16 sparse arrays fail on arithmetic of opposite transpositions
bugcuda librarieshelp wanted
0 条评论0 个反应0 负责人
开放
cuDNN: Provide wrappers for the declarative API
enhancementhelp wanted
3 条评论0 个反应0 负责人
开放
Power of adjoint of `CuSparseMatrix` does not work anymore
bugcuda librarieshelp wanted
3 条评论0 个反应0 负责人
开放
Base.stack is underperforming.
good first issueperformance
2 条评论1 个反应0 负责人
开放
ComplexF32 eigen can return `NaN` unexpectedly
bughelp wanted
2 条评论0 个反应0 负责人