Skip to content

feat: Adding Base.reverse / reverse! support using AcceleratedKernels.jl - #786

Closed
shreyas-omkar wants to merge 1 commit into
JuliaGPU:mainfrom
shreyas-omkar:sh/reverse
Closed

shreyas-omkar wants to merge 1 commit into
JuliaGPU:mainfrom
shreyas-omkar:sh/reverse

Conversation

@shreyas-omkar

Copy link
Copy Markdown
Member

Base.reverse / reverse! delegated to AcceleratedKernels.

Device time (ms), current (vendor) vs AK:

CUDA RTX 5080

size vendor (CUDA.jl) AK
1K 0.0016 0.0016
64K 0.0017 0.0017
1M 0.0037 0.0037
16M 0.0840 0.0797
dims=1 1024² 0.0042 0.0044

AMD RX 9060 XT

size vendor (AMDGPU.jl) AK
1K 0.013 0.015
64K 0.016 0.024
1M 0.036 0.026
16M 0.461 0.460
dims=1 1024² 0.043 0.041

@shreyas-omkar shreyas-omkar changed the title Wire reverse to AcceleratedKernels feat: Adding Base.reverse / reverse! support using AcceleratedKernels.jl Sep 23, 2026
@shreyas-omkar
shreyas-omkar marked this pull request as ready for review October 1, 2026 11:20
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant