I am not sure why adding CUDA.@atomic in the below code causes an error? # works fine without CUDA.@atomic function f(test) broadcast(CuVector(1:2), transpose(CuVector(1:3))) do i,j v::Float32 = j test[i] += v end end f(CUDA.zeros(Float32,2)) # ERROR with CUDA.@atomic func…

CUDA.@atomic only works in kernel context. This usage looks OK, so try in a regular kernel where you can use @device_code_warntype to look at what’s wrong instead of our broadcast implementation here just bailing out.

CUDA.@atomic causes type instability?

Specific Domains GPU

guyvdb November 29, 2021, 6:27pm 6

Done: https://github.com/JuliaGPU/CUDA.jl/issues/1253

Topic		Replies	Views
How to use `@atomic` with CUDA? New to Julia cuda	1	559	October 5, 2020
Cannot manage to use CUDA.atomic_add! GPU cuda , atomic	4	110	June 30, 2025
Atomic operations stop working when upgrading to CUDA.jl version 3.3 GPU question	4	1096	June 17, 2021
Adding at specific CuArray position GPU question	6	239	May 6, 2024
Type instability with CuVector inside struct GPU	2	235	April 14, 2023

CUDA.@atomic causes type instability?

Related topics