Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

CUDA compute capability version check problem #3026

Closed
makortel opened this issue May 11, 2020 · 0 comments
Closed

CUDA compute capability version check problem #3026

makortel opened this issue May 11, 2020 · 0 comments
Labels
Bug Broken / incorrect code; it could be Kokkos' responsibility, or others’ (e.g., Trilinos)

Comments

@makortel
Copy link

I'm trying out Kokkos 3.1.1 on a machine with a Turing GPU, but I use the Volta as the target architecture. When running I get an error

ERROR: running kernels compiled for compute capability 7.0 on device with compute capability 7.5 is not supported by CUDA!

The Turing Compatibility Guide says

any binary that runs on Volta will be able to run on Turing (forward compatibility), but a Turing binary will not be able to run on Volta.

It seems to me that the compute capability check

if (compiled_major != cudaProp.major || compiled_minor < cudaProp.minor) {

the comparison of the minor version should be reversed.

@crtrott crtrott added the Bug Broken / incorrect code; it could be Kokkos' responsibility, or others’ (e.g., Trilinos) label May 11, 2020
@crtrott crtrott added this to To do in Milestone: Release 3.2 via automation May 11, 2020
@crtrott crtrott added this to the Tentative 3.2 Release milestone May 11, 2020
crtrott added a commit to crtrott/kokkos that referenced this issue May 14, 2020
@masterleinad masterleinad moved this from To do to Done in Milestone: Release 3.2 Jun 8, 2020
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
Bug Broken / incorrect code; it could be Kokkos' responsibility, or others’ (e.g., Trilinos)
Projects
No open projects
Development

No branches or pull requests

3 participants