libcudacxx icon indicating copy to clipboard operation
libcudacxx copied to clipboard

Published 20 hours ago • verified publisher icon NVIDIA

`cuda::atomic::fetch_add/sub` should support float/double

Open jrhemstad opened this issue 4 years ago • 1 comments

The cuda::atomic<T> extension type should support atomic addition/subtraction on floating point types.

jrhemstad avatar Jan 12 '21 21:01 jrhemstad

What's the ETA for this?

jeffhammond avatar Feb 04 '22 07:02 jeffhammond

This has been done some time ago.

griwes avatar Feb 23 '23 16:02 griwes