Don't use a full warp for the final cross-team reduction if the reduction's thread limit is below the warp size. Claude assisted with the test. |
||
|---|---|---|
| .. | ||
| cmake | ||
| device | ||
| docs | ||
| libompd | ||
| module | ||
| runtime | ||
| tools | ||
| .gitignore | ||
| CMakeLists.txt | ||
| CREDITS.txt | ||
| LICENSE.TXT | ||
| Maintainers.md | ||