-
Notifications
You must be signed in to change notification settings - Fork 14
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
Matrix Data Reshuffling (matrix re-distribute) #933
Conversation
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Just a couple of initial comments. Sorry if you were already planning to handle them.
b02a104
to
50f89dc
Compare
After #935 got merged, it was a quick change to move from retiled to matrix and I was also able to fix const-correctness 😉 Apart small details, this should be ready for final review. |
cscs-ci run |
There are deadlocks in the CUDA tests. |
cscs-ci run |
…block which is not currently supported
4afd7f2
to
beea7e1
Compare
cscs-ci run |
Codecov Report
❗ Your organization is not using the GitHub App Integration. As a result you may experience degraded service beginning May 15th. Please install the Github App Integration for your organization. Read more. @@ Coverage Diff @@
## master #933 +/- ##
==========================================
+ Coverage 93.30% 93.34% +0.03%
==========================================
Files 135 135
Lines 8397 8440 +43
Branches 1085 1089 +4
==========================================
+ Hits 7835 7878 +43
Misses 386 386
Partials 176 176
|
Close #919
Simple implementation of a copy function that allows to re-distribute a matrix with a different tile size.
Currently, we concentrated on a small subset of configurations by adding some constraints, that should not represent a problem for the use-cases we'd like to try, but simplify the first implementation.
Constraints:
ranks(grid_in) == ranks(grid_out)
Note:
Current implementation uses
RetiledMatrix
(which simplifies the work a lot!), which we know it is a temporary solution waiting for the refactoring. For this reason, I had to relax the const-correctness of the input arguments of the newcopy
function, because currentlyRetiledMatrix<const T>
is not supported.