1 |
Hi Andrew, |
2 |
|
3 |
This week I created an overlay repository and wrote MPI and MPI-based packages (for example, hpl) module files. Based on setting up a virtual environment, they worked as expected. During this period, I investigated and tested several things may influence the module files, e.g. run-time environment variables. And I found the MPI_TARGETS is the most important and most time-consuming part over the project. |
4 |
|
5 |
The progress is on schedule. However, as we talked yesterday, for next-week’s plan, I would like to some next-two-week’s jobs in advance, to leave more time to implement MPI_TARGETS. Next week’s plan would be: |
6 |
|
7 |
|
8 |
1. Finish module files writing. |
9 |
2. Solve some problems like path conflict. |
10 |
3. Implement the test eselect files and interfaces. (next-two-week’s jobs) |
11 |
|
12 |
And these work should be compatible with MPI_TARGETS implementation. |
13 |
|
14 |
Regards, |
15 |
|
16 |
Jian |