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

ILP64 support #85

Open
AidanGG opened this issue Feb 13, 2020 · 2 comments
Open

ILP64 support #85

AidanGG opened this issue Feb 13, 2020 · 2 comments

Comments

@AidanGG
Copy link

AidanGG commented Feb 13, 2020

Hello, I was wondering if CTF supported the 64-bit integer interface to MKL BLAS, LAPACK and ScaLAPACK for arrays with more than 2^31 - 1 elements. For extremely large tensors, this is even more important because a global ScaLAPACK descriptor using 32-bit ints limits the size of the global matrix, not just the local ones.

From what I can tell from the source, the wrappers make use of hard-coded int. Would it be possible to add a build directive to enable the ILP64 interface when building with MKL?

@solomonik
Copy link
Collaborator

Thanks for the suggestion, it would be possible to add this support, but will require a bit of work. We will try to add this soon.

@AidanGG
Copy link
Author

AidanGG commented Mar 27, 2020

I believe ILP64 support has recently been merged into Reference ScaLAPACK (Reference-ScaLAPACK/scalapack#19) so in principle CTF would be able to use 64-bit ints for ScaLAPACK regardless of implementation (MKL or reference).

@AidanGG AidanGG changed the title ILP64 support with MKL ILP64 support Mar 27, 2020
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Development

No branches or pull requests

2 participants