Triangle Count DSL has long datatype which is not supported in CUDA atomicAdd #23
ashwinktpu
started this conversation in
General
Replies: 0 comments
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
Cuda doesn't support long data type with atomicAdd (https://docs.nvidia.com/cuda/cuda-c-programming-guide/#atomicadd)
In our DSL we have long triangle_count = 0. During code generation in CUDA, it will give an error.
There might be two possibilities -
Final Solution
While calling atomicAdd, we can typecast as atomicAdd((unsigned long long*)x,(unsigned long long)1)
This will remove compilation issues and work fine.
Beta Was this translation helpful? Give feedback.
All reactions