Sync the __builtin_expects for our 3 quadratically probed hash table implementations.

This assumes that
  a) finding the bucket containing the value is LIKELY
  b) finding an empty bucket is LIKELY
  c) growing the table is UNLIKELY

I also switched the a) and b) cases for SmallPtrSet as we seem to use
the set mostly more for insertion than for checking existence.

In a simple benchmark consisting of 2^21 insertions of 2^20 unique
pointers into a DenseMap or SmallPtrSet a few percent speedup on average,
but nothing statistically significant.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@230232 91177308-0d34-0410-b5e6-96231b3b80d8
This commit is contained in:
Benjamin Kramer
2015-02-23 16:41:36 +00:00
parent 01312dd0b4
commit 42f72e2c16
3 changed files with 18 additions and 15 deletions

View File

@@ -188,9 +188,10 @@ unsigned StringMapImpl::RehashTable(unsigned BucketNo) {
// If the hash table is now more than 3/4 full, or if fewer than 1/8 of
// the buckets are empty (meaning that many are filled with tombstones),
// grow/rehash the table.
if (NumItems*4 > NumBuckets*3) {
if (LLVM_UNLIKELY(NumItems * 4 > NumBuckets * 3)) {
NewSize = NumBuckets*2;
} else if (NumBuckets-(NumItems+NumTombstones) <= NumBuckets/8) {
} else if (LLVM_UNLIKELY(NumBuckets - (NumItems + NumTombstones) <=
NumBuckets / 8)) {
NewSize = NumBuckets;
} else {
return BucketNo;