[llvm-branch-commits] [Support] parallelFor: enable caller participation (PR #209690)

[via llvm-branch-commits]

llvmorg-github-actions[bot] wrote:


<!--LLVM PR SUMMARY COMMENT-->

@llvm/pr-subscribers-llvm-support

Author: Fangrui Song (MaskRay)

<details>
<summary>Changes</summary>

`parallelFor` dispatches its work to the thread pool and blocks the calling
thread in the `TaskGroup` destructor until the workers finish. The caller
thus stays idle instead of joining the work as modern parallel libraries do,
and `numactl -C 0-7 ld.lld --threads=8` ends up slower than
`numactl -C 0-8 ld.lld --threads=8`.

Run one worker on the calling thread instead. This requires that no
`parallelFor` body reads `getThreadIndex()`, which is unset on the caller;
the last such user is removed by #<!-- -->209687.

In the jobserver mode: only pool workers acquire job slots, so a
jobserver-limited process may briefly exceed its granted concurrency by
one thread. This is acceptable, better than pessimizing the common
non-jobserver case, and parallelFor is unused by jobserver users.


---
Full diff: https://github.com/llvm/llvm-project/pull/209690.diff


1 Files Affected:

- (modified) llvm/lib/Support/Parallel.cpp (+4-1) 


``````````diff
diff --git a/llvm/lib/Support/Parallel.cpp b/llvm/lib/Support/Parallel.cpp
index 8997cfd40ffe6..d2be7ad07b5b6 100644
--- a/llvm/lib/Support/Parallel.cpp
+++ b/llvm/lib/Support/Parallel.cpp
@@ -284,9 +284,12 @@ void llvm::parallelFor(size_t Begin, size_t End,
       }
     };
 
+    // Run one worker on the calling thread: starts working immediately and
+    // avoids an idle thread.
     TaskGroup TG;
-    for (size_t I = 0; I != NumWorkers; ++I)
+    while (--NumWorkers)
       TG.spawn(Worker);
+    Worker();
     return;
   }
 #endif

``````````

</details>


https://github.com/llvm/llvm-project/pull/209690