diff --git a/mabel/data/internals/group_by.py b/mabel/data/internals/group_by.py
index 995a95c..a6ac668 100644
--- a/mabel/data/internals/group_by.py
+++ b/mabel/data/internals/group_by.py
@@ -7,7 +7,7 @@
 """
 from collections import defaultdict
 
-from siphashc import siphash
+from xxhash import xxh3_64_intdigest
 
 
 def summer(x, y):
@@ -25,7 +25,7 @@ def summer(x, y):
     "AVG": lambda x, y: 1,
 }
 
-HASH_SEED = b"Anakin Skywalker"
+HASH_SEED = 42
 
 
 class TooManyGroups(Exception):
@@ -72,14 +72,14 @@ def _map(self, collect_columns):
 
         for record in self._dictset:
             try:
-                group_key: int = siphash(
-                    HASH_SEED,
+                group_key: int = xxh3_64_intdigest(
                     "".join([str(record[column]) for column in self._columns]),
+                    HASH_SEED
                 )
             except KeyError:
-                group_key: int = siphash(
-                    HASH_SEED,
+                group_key: int = xxh3_64_intdigest(
                     "".join([f"{record.get(column, '')}" for column in self._columns]),
+                    HASH_SEED,
                 )
             if group_key not in self._group_keys.keys():
                 self._group_keys[group_key] = [
diff --git a/mabel/data/readers/internals/cursor.py b/mabel/data/readers/internals/cursor.py
index eb1268f..8d6ab80 100644
--- a/mabel/data/readers/internals/cursor.py
+++ b/mabel/data/readers/internals/cursor.py
@@ -3,7 +3,7 @@
 
 Cursor is made of three parts:
 - map      : a bit array representing all of the blobs in the set - unread blobs
-             are 0s and read blobs are 1s. This allows for blobs to be read in 
+             are 0s and read blobs are 1s. This allows for blobs to be read in
              an arbitrary order - although currently only implemented linearly.
 - partition: the active parition (blob) that is being read
 - location : the record in the active partition (blob), so we can resume reading
@@ -11,7 +11,7 @@
 """
 
 import orjson
-from orso.cityhash import CityHash64
+from xxhash import xxh3_64_intdigest
 
 
 class InvalidCursor(Exception):
@@ -47,7 +47,9 @@ def load_cursor(self, cursor):
 
         self.location = cursor["location"]
         find_partition = [
-            blob for blob in self.readable_blobs if CityHash64(blob) == cursor["partition"]
+            blob
+            for blob in self.readable_blobs
+            if xxh3_64_intdigest(blob, 0) == cursor["partition"]
         ]
         if len(find_partition) == 1:
             self.partition = find_partition[0]
@@ -67,7 +69,7 @@ def next_blob(self, previous_blob=None):
             if self.partition in self.readable_blobs:
                 return self.partition
             partition_finder = [
-                blob for blob in self.readable_blobs if CityHash64(blob) == self.partition
+                blob for blob in self.readable_blobs if xxh3_64_intdigest(blob, 0) == self.partition
             ]
             if len(partition_finder) != 1:
                 raise ValueError(f"Unable to determine current partition ({self.partition})")
@@ -103,7 +105,7 @@ def __getitem__(self, item):
             )
             return blob_map.tobytes().hex()
         if item == "partition":
-            return CityHash64(self.partition)
+            return xxh3_64_intdigest(self.partition, 0)
         if item == "location":
             return self.location
         return None
diff --git a/mabel/data/readers/internals/threaded_wrapper.py b/mabel/data/readers/internals/threaded_wrapper.py
index 4aa5f7d..05dc9be 100644
--- a/mabel/data/readers/internals/threaded_wrapper.py
+++ b/mabel/data/readers/internals/threaded_wrapper.py
@@ -1,6 +1,4 @@
-"""
-
-"""
+""" """
 
 import logging
 import threading
diff --git a/mabel/data/validator/__init__.py b/mabel/data/validator/__init__.py
index e9c4d87..a1ad905 100644
--- a/mabel/data/validator/__init__.py
+++ b/mabel/data/validator/__init__.py
@@ -11,7 +11,7 @@
 
 
 def schema_loader(
-    definition: Union[str, List[Dict[str, Any]], dict, RelationSchema, bool]
+    definition: Union[str, List[Dict[str, Any]], dict, RelationSchema, bool],
 ) -> Union[RelationSchema, bool]:
     if definition is None:
         raise ValueError(
diff --git a/mabel/data/writers/internals/blob_writer.py b/mabel/data/writers/internals/blob_writer.py
index 4adece5..2ad72a9 100644
--- a/mabel/data/writers/internals/blob_writer.py
+++ b/mabel/data/writers/internals/blob_writer.py
@@ -165,7 +165,7 @@ def commit(self):
                         pytable = self._normalize_arrow_schema(pytable, self.schema)
 
                     tempfile = io.BytesIO()
-                    pyarrow.parquet.write_table(pytable, where=tempfile, compression="zstd")
+                    pyarrow.parquet.write_table(pytable, where=tempfile)
 
                     tempfile.seek(0)
                     write_buffer = tempfile.read()
diff --git a/mabel/utils/dates.py b/mabel/utils/dates.py
index 1e35645..61ffe07 100644
--- a/mabel/utils/dates.py
+++ b/mabel/utils/dates.py
@@ -41,7 +41,7 @@ def parse_delta(delta: str) -> datetime.timedelta:
 
 
 def parse_iso(
-    value: Union[str, int, float, datetime.datetime, datetime.date]
+    value: Union[str, int, float, datetime.datetime, datetime.date],
 ) -> Optional[datetime.datetime]:
     """
     Parses an ISO date string into a datetime object, with an emphasis on speed.
diff --git a/mabel/version.py b/mabel/version.py
index a98c243..41ac715 100644
--- a/mabel/version.py
+++ b/mabel/version.py
@@ -1,6 +1,6 @@
 # Store the version here so:
 # 1) we don't load dependencies by storing it in __init__.py
 # 2) we can import it in setup.py for the same reason
-__version__ = "0.6.24"
+__version__ = "0.6.25"
 
 # nodoc - don't add to the documentation wiki
diff --git a/requirements.txt b/requirements.txt
index ce8d42d..d7aa106 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -2,5 +2,5 @@ bitarray
 lz4
 orjson
 orso>=0.0.147
-siphashc
+xxhash
 zstandard
\ No newline at end of file
diff --git a/tests/performance/index_performance.py b/tests/performance/index_performance.py
index a872997..442cb6a 100644
--- a/tests/performance/index_performance.py
+++ b/tests/performance/index_performance.py
@@ -1,11 +1,11 @@
 """
 Results (seconds to search for a username in 65,500 rows):
 
- indexed | row exists | time  
+ indexed | row exists | time
 -------------------------------
  yes     |    yes     |   0.094   <- about 3.5x faster when is match
  yes     |    no      |   0.006   <- over 50x faster when no match
- no      |    yes     |   0.357 
+ no      |    yes     |   0.357
  no      |    no      |   0.332
 -------------------------------
 """
diff --git a/tests/performance/indexing.py b/tests/performance/indexing.py
index d6d8dc7..a4cc644 100644
--- a/tests/performance/indexing.py
+++ b/tests/performance/indexing.py
@@ -1,6 +1,4 @@
-"""
-
-"""
+""" """
 
 import sys
 import os
diff --git a/tests/performance/json_performance.py b/tests/performance/json_performance.py
index c90af1e..9d161ef 100644
--- a/tests/performance/json_performance.py
+++ b/tests/performance/json_performance.py
@@ -1,13 +1,13 @@
 """
-JSON parsing and serialization performance tests so a decision on 
+JSON parsing and serialization performance tests so a decision on
 which library(s) to use can be made - previously the selection was
 inconsistent.
 
 Results (seconds to process 10m rows):
 
- library | parsing | serialize  
+ library | parsing | serialize
 -------------------------------
- json    |    36.6 |      1.74 
+ json    |    36.6 |      1.74
  ujson   |    16.5 |      0.86
  orjson  |    10.4 |      0.66   <- lower is better
  simd    |     2.8 |       N/A   <- lower is better
diff --git a/tests/test_data_dictset.py b/tests/test_data_dictset.py
index d0807af..f44306b 100644
--- a/tests/test_data_dictset.py
+++ b/tests/test_data_dictset.py
@@ -2,6 +2,7 @@
 import sys
 
 sys.path.insert(1, os.path.join(sys.path[0], ".."))
+
 from mabel import Reader, DictSet
 from mabel.data import STORAGE_CLASS
 from mabel.adapters.disk import DiskReader
@@ -189,7 +190,7 @@ def test_hash():
     ]
     ds = DictSet(data, storage_class=STORAGE_CLASS.MEMORY)
     hashval = hash(ds)
-    assert hashval == 8826660322425604498, hashval
+    assert hashval == 386528107484878589, hashval
 
 
 def test_sort():
@@ -210,6 +211,6 @@ def test_sort():
 
 
 if __name__ == "__main__":  # pragma: no cover
-    from tests.helpers.runner import run_tests
+    from helpers.runner import run_tests
 
     run_tests()
diff --git a/tests/test_data_group_by.py b/tests/test_data_group_by.py
index 3d95c6c..28013ce 100644
--- a/tests/test_data_group_by.py
+++ b/tests/test_data_group_by.py
@@ -38,17 +38,24 @@ def test_group_by():
     gb = GroupBy(ds, "user")
     cn = gb.count()
     ls = list(cn)
-    assert ls == [{'COUNT(*)': 6, 'user': 'alice'}, {'COUNT(*)': 5, 'user': 'bob'}, {'COUNT(*)': 2, 'user': 'eve'}], ls
+    expected = [{'COUNT(*)': 6, 'user': 'alice'}, {'COUNT(*)': 5, 'user': 'bob'}, {'COUNT(*)': 2, 'user': 'eve'}]
+    assert set(tuple(sorted(d.items())) for d in ls) == set(tuple(sorted(d.items())) for d in expected)
 
     ds = DictSet(data, storage_class=STORAGE_CLASS.MEMORY)
     gb = list(GroupBy(ds, "user").average("value"))
-    assert gb == [{'AVG(value)': Decimal("4.0"), 'user': 'alice'}, {'AVG(value)': Decimal("1.4"), 'user': 'bob'}, {'AVG(value)': Decimal("6.5"), 'user': 'eve'}], gb
+    expected = [{'AVG(value)': Decimal("4.0"), 'user': 'alice'}, {'AVG(value)': Decimal("1.4"), 'user': 'bob'}, {'AVG(value)': Decimal("6.5"), 'user': 'eve'}]
+    assert set(tuple(sorted(d.items())) for d in gb) == set(tuple(sorted(d.items())) for d in expected)
 
     ds = DictSet(data, storage_class=STORAGE_CLASS.MEMORY)
-    assert list(GroupBy(ds, "user").max("value")) == [{'MAX(value)': 5, 'user': 'alice'}, {'MAX(value)': 2, 'user': 'bob'}, {'MAX(value)': 7, 'user': 'eve'}]
+    gs = list(GroupBy(ds, "user").max("value"))
+    expected = [{'MAX(value)': 5, 'user': 'alice'}, {'MAX(value)': 2, 'user': 'bob'}, {'MAX(value)': 7, 'user': 'eve'}]
+    assert set(tuple(sorted(d.items())) for d in gs) == set(tuple(sorted(d.items())) for d in expected)
 
     ds = DictSet(data, storage_class=STORAGE_CLASS.MEMORY)
-    assert list(GroupBy(ds, "user").min("value")) == [{'MIN(value)': 3, 'user': 'alice'}, {'MIN(value)': 1, 'user': 'bob'}, {'MIN(value)': 6, 'user': 'eve'}]
+    gs = list(GroupBy(ds, "user").min("value"))
+    expected = [{'MIN(value)': 3, 'user': 'alice'}, {'MIN(value)': 1, 'user': 'bob'}, {'MIN(value)': 6, 'user': 'eve'}]
+    assert set(tuple(sorted(d.items())) for d in gs) == set(tuple(sorted(d.items())) for d in expected)
+
     # fmt:on
 
 
@@ -75,27 +82,34 @@ def test_combined_group_by():
     # fmt:off
     ds = DictSet(data, storage_class=STORAGE_CLASS.MEMORY)
     gs = list(GroupBy(ds, ("fname", "sname")).groups())
-    assert gs == [{'fname': 'bob', 'sname': 'smith'}, {'fname': 'bob', 'sname': 'jones'}, {'fname': 'alice', 'sname': 'jones'}, {'fname': 'alice', 'sname': 'smith'}, {'fname': 'eve', 'sname': 'jones'}, {'fname': 'eve', 'sname': 'smith'}], gs
+    expected = [{'fname': 'bob', 'sname': 'smith'}, {'fname': 'bob', 'sname': 'jones'}, {'fname': 'alice', 'sname': 'jones'}, {'fname': 'alice', 'sname': 'smith'}, {'fname': 'eve', 'sname': 'jones'}, {'fname': 'eve', 'sname': 'smith'}]
+    assert set(tuple(sorted(d.items())) for d in gs) == set(tuple(sorted(d.items())) for d in expected)
 
     ds = DictSet(data, storage_class=STORAGE_CLASS.MEMORY)
     gs = list(GroupBy(ds, ("fname", "sname")).count())
-    assert gs ==  [{'COUNT(*)': 2, 'fname': 'bob', 'sname': 'jones'}, {'COUNT(*)': 3, 'fname': 'bob', 'sname': 'smith'}, {'COUNT(*)': 3, 'fname': 'alice', 'sname': 'smith'}, {'COUNT(*)': 1, 'fname': 'eve', 'sname': 'smith'}, {'COUNT(*)': 1, 'fname': 'eve', 'sname': 'jones'}, {'COUNT(*)': 3, 'fname': 'alice', 'sname': 'jones'}], gs
+    expected = [{'COUNT(*)': 2, 'fname': 'bob', 'sname': 'jones'}, {'COUNT(*)': 3, 'fname': 'bob', 'sname': 'smith'}, {'COUNT(*)': 3, 'fname': 'alice', 'sname': 'smith'}, {'COUNT(*)': 1, 'fname': 'eve', 'sname': 'smith'}, {'COUNT(*)': 1, 'fname': 'eve', 'sname': 'jones'}, {'COUNT(*)': 3, 'fname': 'alice', 'sname': 'jones'}]
+    assert set(tuple(sorted(d.items())) for d in gs) == set(tuple(sorted(d.items())) for d in expected)
 
     ds = DictSet(data, storage_class=STORAGE_CLASS.MEMORY)
     gs = list(GroupBy(ds, ("fname", "sname")).average('value'))
-    assert gs == [{'AVG(value)': Decimal('2.0'), 'fname': 'bob', 'sname': 'jones'}, {'AVG(value)': Decimal('1.0'), 'fname': 'bob', 'sname': 'smith'}, {'AVG(value)': Decimal('4.333333333333333333333333333'), 'fname': 'alice', 'sname': 'smith'}, {'AVG(value)': Decimal('7.0'), 'fname': 'eve', 'sname': 'smith'}, {'AVG(value)': Decimal('6.0'), 'fname': 'eve', 'sname': 'jones'}, {'AVG(value)': Decimal('3.666666666666666666666666667'), 'fname': 'alice', 'sname': 'jones'}], gs
+    expected = [{'AVG(value)': Decimal('2.0'), 'fname': 'bob', 'sname': 'jones'}, {'AVG(value)': Decimal('1.0'), 'fname': 'bob', 'sname': 'smith'}, {'AVG(value)': Decimal('4.333333333333333333333333333'), 'fname': 'alice', 'sname': 'smith'}, {'AVG(value)': Decimal('7.0'), 'fname': 'eve', 'sname': 'smith'}, {'AVG(value)': Decimal('6.0'), 'fname': 'eve', 'sname': 'jones'}, {'AVG(value)': Decimal('3.666666666666666666666666667'), 'fname': 'alice', 'sname': 'jones'}]
+    assert set(tuple(sorted(d.items())) for d in gs) == set(tuple(sorted(d.items())) for d in expected)
+
 
     ds = DictSet(data, storage_class=STORAGE_CLASS.MEMORY)
     gs = list(GroupBy(ds, ("fname", "sname")).average('cost'))
-    assert gs == [{'AVG(cost)': Decimal('3.0'), 'fname': 'bob', 'sname': 'jones'}, {'AVG(cost)': Decimal('1.666666666666666666666666667'), 'fname': 'bob', 'sname': 'smith'}, {'AVG(cost)': Decimal('2.333333333333333333333333333'), 'fname': 'alice', 'sname': 'smith'}, {'AVG(cost)': Decimal('1.0'), 'fname': 'eve', 'sname': 'smith'}, {'AVG(cost)': Decimal('2.0'), 'fname': 'eve', 'sname': 'jones'}, {'AVG(cost)': Decimal('3.333333333333333333333333333'), 'fname': 'alice', 'sname': 'jones'}], gs
+    expected = [{'AVG(cost)': Decimal('3.0'), 'fname': 'bob', 'sname': 'jones'}, {'AVG(cost)': Decimal('1.666666666666666666666666667'), 'fname': 'bob', 'sname': 'smith'}, {'AVG(cost)': Decimal('2.333333333333333333333333333'), 'fname': 'alice', 'sname': 'smith'}, {'AVG(cost)': Decimal('1.0'), 'fname': 'eve', 'sname': 'smith'}, {'AVG(cost)': Decimal('2.0'), 'fname': 'eve', 'sname': 'jones'}, {'AVG(cost)': Decimal('3.333333333333333333333333333'), 'fname': 'alice', 'sname': 'jones'}]
+    assert set(tuple(sorted(d.items())) for d in gs) == set(tuple(sorted(d.items())) for d in expected)
 
     ds = DictSet(data, storage_class=STORAGE_CLASS.MEMORY)
     gs = list(GroupBy(ds, ("fname", "sname")).average(('cost', 'value',)))
-    assert gs == [{'AVG(cost)': Decimal('3'), 'AVG(value)': Decimal('2'), 'fname': 'bob', 'sname': 'jones'}, {'AVG(cost)': Decimal('1.666666666666666666666666667'), 'AVG(value)': Decimal('1'), 'fname': 'bob', 'sname': 'smith'}, {'AVG(cost)': Decimal('2.333333333333333333333333333'), 'AVG(value)': Decimal('4.333333333333333333333333333'), 'fname': 'alice', 'sname': 'smith'}, {'AVG(cost)': 1.0, 'AVG(value)': 7.0, 'fname': 'eve', 'sname': 'smith'}, {'AVG(cost)': 2.0, 'AVG(value)': 6.0, 'fname': 'eve', 'sname': 'jones'}, {'AVG(cost)': Decimal('3.333333333333333333333333333'), 'AVG(value)': Decimal('3.666666666666666666666666667'), 'fname': 'alice', 'sname': 'jones'}]
+    expected = [{'AVG(cost)': Decimal('3'), 'AVG(value)': Decimal('2'), 'fname': 'bob', 'sname': 'jones'}, {'AVG(cost)': Decimal('1.666666666666666666666666667'), 'AVG(value)': Decimal('1'), 'fname': 'bob', 'sname': 'smith'}, {'AVG(cost)': Decimal('2.333333333333333333333333333'), 'AVG(value)': Decimal('4.333333333333333333333333333'), 'fname': 'alice', 'sname': 'smith'}, {'AVG(cost)': 1.0, 'AVG(value)': 7.0, 'fname': 'eve', 'sname': 'smith'}, {'AVG(cost)': 2.0, 'AVG(value)': 6.0, 'fname': 'eve', 'sname': 'jones'}, {'AVG(cost)': Decimal('3.333333333333333333333333333'), 'AVG(value)': Decimal('3.666666666666666666666666667'), 'fname': 'alice', 'sname': 'jones'}]
+    assert set(tuple(sorted(d.items())) for d in gs) == set(tuple(sorted(d.items())) for d in expected)
 
     ds = DictSet(data, storage_class=STORAGE_CLASS.MEMORY)
     gs = list(GroupBy(ds, ("fname", "sname")).aggregate([('MAX', 'value'),('MIN', 'cost')]))
-    assert gs == [{'MAX(value)': 2, 'MIN(cost)': 2, 'fname': 'bob', 'sname': 'jones'}, {'MAX(value)': 1, 'MIN(cost)': 1, 'fname': 'bob', 'sname': 'smith'}, {'MAX(value)': 5, 'MIN(cost)': 1, 'fname': 'alice', 'sname': 'smith'}, {'MAX(value)': 7, 'MIN(cost)': 1, 'fname': 'eve', 'sname': 'smith'}, {'MAX(value)': 6, 'MIN(cost)': 2, 'fname': 'eve', 'sname': 'jones'}, {'MAX(value)': 5, 'MIN(cost)': 2, 'fname': 'alice', 'sname': 'jones'}], gs
+    expected = [{'MAX(value)': 2, 'MIN(cost)': 2, 'fname': 'bob', 'sname': 'jones'}, {'MAX(value)': 1, 'MIN(cost)': 1, 'fname': 'bob', 'sname': 'smith'}, {'MAX(value)': 5, 'MIN(cost)': 1, 'fname': 'alice', 'sname': 'smith'}, {'MAX(value)': 7, 'MIN(cost)': 1, 'fname': 'eve', 'sname': 'smith'}, {'MAX(value)': 6, 'MIN(cost)': 2, 'fname': 'eve', 'sname': 'jones'}, {'MAX(value)': 5, 'MIN(cost)': 2, 'fname': 'alice', 'sname': 'jones'}]
+    assert set(tuple(sorted(d.items())) for d in gs) == set(tuple(sorted(d.items())) for d in expected)
 
     # fmt:on
 
@@ -109,15 +123,17 @@ def test_gappy_set():
         {"key": 4, "value": None, "plus1": 5},
     ]
     ds = DictSet(data, storage_class=STORAGE_CLASS.MEMORY)
-    g = list(ds.group_by("value").average("key"))
-    assert g == [
+    gs = list(ds.group_by("value").average("key"))
+    expected = [
         {"AVG(key)": 4.0, "value": None},
         {"AVG(key)": 3.0, "value": "two"},
         {"AVG(key)": 1.0, "value": "one"},
-    ], g
+    ]
+
+    assert set(tuple(sorted(d.items())) for d in gs) == set(tuple(sorted(d.items())) for d in expected)
 
 
 if __name__ == "__main__":  # pragma: no cover
-    from tests.helpers.runner import run_tests
+    from helpers.runner import run_tests
 
     run_tests()
diff --git a/tests/test_reader_cursor.py b/tests/test_reader_cursor.py
index e824357..8f81fa4 100644
--- a/tests/test_reader_cursor.py
+++ b/tests/test_reader_cursor.py
@@ -37,8 +37,6 @@ def test_cursor():
     for i in range(number_of_records):
         data.append({"one": 1, "index": i})
 
-    reader = Reader(inner_reader=NullReader, dataset="none", partitions=None, data=data)
-
     # create random offsets for testing - it's illogical to have a 0 cursor
     offsets = (entropy.random_range(1, number_of_records) for i in range(20))
 
@@ -135,7 +133,7 @@ def test_cursor_as_text():
     offsets = (entropy.random_range(1, number_of_records - 1) for i in range(5))
 
     for offset in offsets:
-        cursor = {"location": offset, "map": "00", "partition": 1983839293359648136}
+        cursor = {"location": offset, "map": "00", "partition": 13429097919052166063}
         reader = Reader(
             inner_reader=DiskReader,
             dataset="tests/data/formats/jsonl",
@@ -164,7 +162,7 @@ def test_move_to_cursor():
             inner_reader=DiskReader,
             dataset="tests/data/formats/jsonl",
             partitions=None,
-            cursor={"location": offset, "map": "00", "partition": 1983839293359648136},
+            cursor={"location": offset, "map": "00", "partition": 13429097919052166063},
             persistence=STORAGE_CLASS.NO_PERSISTANCE,
         )
 
@@ -229,6 +227,6 @@ def test_multiple_files():
 
 
 if __name__ == "__main__":  # pragma: no cover
-    from tests.helpers.runner import run_tests
+    from helpers.runner import run_tests
 
     run_tests()
diff --git a/tests/test_utils_common.py b/tests/test_utils_common.py
index 9e2e835..624519b 100644
--- a/tests/test_utils_common.py
+++ b/tests/test_utils_common.py
@@ -1,6 +1,4 @@
-"""
-
-"""
+""" """
 
 import os
 import sys