Created by: mhoemmen
@trilinos/tpetra @trilinos/belos @vbrunini
Description
This PR supersedes PR #4648. I've extracted out the parts of #4648 that are actual bug fixes rather than optimizations. Here they are:
- Fix #4639 (closed) (an actual correctness bug in
Tpetra::MultiVector::reduce
that hindered solving #4626 (closed)) - Fix #4633 (closed) (a strange thing in
Tpetra::MultiVector
that could lead to bugs, especially withFEMultiVector
) - Clean up a few things in Belos
The part of PR #4648 I did not include here is the fix for #4626 (closed). That triggers a likely Intel compiler bug in PR tests, which is why I'm not including it. I have a multiple-step work-around in progress; see PR #4734 for the start of that.
Related Issues
- Closes #4639 (closed), #4633 (closed)
- Blocks #4626 (closed)
- Related to #4640, #4648, #4734
How Has This Been Tested?
With and without CUDA.