Min Si
To better coordinate student internships, it is better to have a page list all small R&D projects (e.g., a project can be done by a student in 3 months)
- [ ] OSHMPI with MPICH on Tulip - [ ] ROC_SHMEM on Tulip Dependes on #50
- [ ] Prototype code for supporting AMD GPU memkind - [ ] Merge into OSHMPI/master
Description: Performance analysis of team-based collectives. Depends on #90 TODO: - [ ] Compare team-based collective with original active-set-based version - [ ] Compare OSHMPI team-based collectives with [SOS](https://github.com/Sandia-OpenSHMEM/SOS) on...
**Description**: Analyze overheads of strided datatype creation and decoding in the strided RMA path. Need to configure with `--disable-strided-cache` to disable the datatype cache optimization in OSHMPI. **Starting point**: -...
**Description**: For OSHMPI we use only `resized vector with blocklength=1` datatype. Can we do any optimization for such simple datatypes in MPICH to reduce creation overhead and also decoding overhead?...
**Description**: OpenSHMEM 1.5 adds team-based collective. It is straightforward to translate to MPI collectives with the corresponding `MPI comm`. Depend on #36 **Starting point**: - Section 9.9 Collective Routines in...
Current backend defines all hooks as function pointer. Some hooks are accessed at fast-path or accessed multiple times in a single ipack/iunpack call. Compiler cannot optimize much for function pointers....