labath added inline comments.
================
Comment at: lldb/source/Plugins/SymbolFile/DWARF/ManualDWARFIndex.cpp:24-30
+static void TaskMapOverInt(size_t end,
+ const llvm::function_ref<void(size_t)> &func) {
+ llvm::ThreadPool pool;
+ for (size_t i = 0; i < end; i++)
+ pool.async(func, i);
+ pool.wait();
+}
----------------
labath wrote:
> This *is* simpler than the previous implementation, but it may have different
> performance characteristics. The previous implementation was optimized for
> small workloads where the overhead of enqueuing a task was not negligible
> compared to the size of the job itself.
(I'm not saying this should be changed back. llvm's thread pool creates threads
differently, and that is likely to have been the main cause of overhead, so
this optimization may not matter anymore. However, it would be good to verify
that.)
Repository:
rLLDB LLDB
CHANGES SINCE LAST ACTION
https://reviews.llvm.org/D78337/new/
https://reviews.llvm.org/D78337
_______________________________________________
lldb-commits mailing list
lldb-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/lldb-commits
