Use XXH3 to get a good hash distribution sine it will be very bad when using group/user doc distribution.

2 files changed, 11 insertions, 0 deletions

diff --git a/document/src/vespa/document/bucket/bucketid.cpp b/document/src/vespa/document/bucket/bucketid.cpp
index f01bfb67674..3752eab4810 100644
--- a/document/src/vespa/document/bucket/bucketid.cpp
+++ b/document/src/vespa/document/bucket/bucketid.cpp
@@ -8,6 +8,7 @@
 #include <vespa/vespalib/stllike/hash_set.hpp>
 #include <vespa/vespalib/util/stringfmt.h>
 #include <limits>
+#include <xxh3.h>
 
 using vespalib::nbostream;
 using vespalib::asciistream;
@@ -76,6 +77,12 @@ void BucketId::initialize() noexcept {
     fillStripMasks(BucketId::_stripMasks, BucketId::maxNumBits);
 }
 
+uint64_t
+BucketId::xxhash::operator () (const BucketId& bucketId) const noexcept {
+    const uint64_t raw_id = bucketId.getId();
+    return XXH3_64bits(&raw_id, sizeof(uint64_t));
+}
+
 vespalib::string
 BucketId::toString() const
 {
@@ -152,3 +159,4 @@ operator>>(nbostream &is, BucketId &bucketId)
 } // document
 
 VESPALIB_HASH_SET_INSTANTIATE_H(document::BucketId, document::BucketId::hash);
+VESPALIB_HASH_SET_INSTANTIATE_H(document::BucketId, document::BucketId::xxhash);
diff --git a/document/src/vespa/document/bucket/bucketid.h b/document/src/vespa/document/bucket/bucketid.h
index d54f86e4ae7..a0f9ca4e94f 100644
--- a/document/src/vespa/document/bucket/bucketid.h
+++ b/document/src/vespa/document/bucket/bucketid.h
@@ -41,6 +41,9 @@ public:
             return g.getId();
         }
     };
+    struct xxhash {
+        uint64_t operator () (const BucketId& g) const noexcept;
+    };
 
     /**
      * The primitive type used to store bucket identifiers. If you use the