[llvm] [MemDep] Optimize SortNonLocalDepInfoCache sorting strategy for large caches with few unsorted entries (PR #143107)

Qiu Chaofan via llvm-commits llvm-commits at lists.llvm.org
Sun Jun 8 23:04:11 PDT 2025


================
@@ -991,33 +994,19 @@ MemDepResult MemoryDependenceResults::getNonLocalInfoForBlock(
 static void
 SortNonLocalDepInfoCache(MemoryDependenceResults::NonLocalDepInfo &Cache,
                          unsigned NumSortedEntries) {
-  switch (Cache.size() - NumSortedEntries) {
-  case 0:
-    // done, no new entries.
-    break;
-  case 2: {
-    // Two new entries, insert the last one into place.
-    NonLocalDepEntry Val = Cache.back();
-    Cache.pop_back();
-    MemoryDependenceResults::NonLocalDepInfo::iterator Entry =
-        std::upper_bound(Cache.begin(), Cache.end() - 1, Val);
-    Cache.insert(Entry, Val);
-    [[fallthrough]];
-  }
-  case 1:
-    // One new entry, Just insert the new value at the appropriate position.
-    if (Cache.size() != 1) {
+
+  auto s = Cache.size() - NumSortedEntries;
+  if (s < log2(Cache.size()) * ln2) {
----------------
ecnelises wrote:

Why not use `log(Cache.size())` directly? Or, at least, Compiler automatically folds `log(2)`, no need to define a constant.

https://github.com/llvm/llvm-project/pull/143107


More information about the llvm-commits mailing list