diff --git a/src/spikeinterface/core/basesorting.py b/src/spikeinterface/core/basesorting.py
index 42f98b3473..e17731c70e 100644
--- a/src/spikeinterface/core/basesorting.py
+++ b/src/spikeinterface/core/basesorting.py
@@ -766,10 +766,6 @@ def _compute_and_cache_spike_vector(self) -> None:
             if len(sample_indices) > 0:
                 sample_indices = np.concatenate(sample_indices, dtype="int64")
                 unit_indices = np.concatenate(unit_indices, dtype="int64")
-                order = np.argsort(sample_indices)
-                sample_indices = sample_indices[order]
-                unit_indices = unit_indices[order]
-
                 n = sample_indices.size
                 segment_slices[segment_index, 0] = seg_pos
                 segment_slices[segment_index, 1] = seg_pos + n
@@ -783,7 +779,9 @@ def _compute_and_cache_spike_vector(self) -> None:
             spikes_in_seg["unit_index"] = unit_indices
             spikes_in_seg["segment_index"] = segment_index
             spikes.append(spikes_in_seg)
+
         spikes = np.concatenate(spikes)
+        spikes = spikes[np.lexsort((spikes["unit_index"], spikes["sample_index"], spikes["segment_index"]))]
 
         self._cached_spike_vector = spikes
         self._cached_spike_vector_segment_slices = segment_slices
diff --git a/src/spikeinterface/core/generate.py b/src/spikeinterface/core/generate.py
index eddbe318a6..5d7ca1917a 100644
--- a/src/spikeinterface/core/generate.py
+++ b/src/spikeinterface/core/generate.py
@@ -174,7 +174,7 @@ def generate_sorting(
             spikes.append(spikes_on_borders)
 
     spikes = np.concatenate(spikes)
-    spikes = spikes[np.lexsort((spikes["sample_index"], spikes["segment_index"]))]
+    spikes = spikes[np.lexsort((spikes["unit_index"], spikes["sample_index"], spikes["segment_index"]))]
 
     sorting = NumpySorting(spikes, sampling_frequency, unit_ids)
 
diff --git a/src/spikeinterface/core/testing.py b/src/spikeinterface/core/testing.py
index 3f311f1bdd..57b6397863 100644
--- a/src/spikeinterface/core/testing.py
+++ b/src/spikeinterface/core/testing.py
@@ -112,30 +112,20 @@ def check_sortings_equal(
 
     max_spike_index = SX1.to_spike_vector()["sample_index"].max()
 
-    # TODO for later  use to_spike_vector() to do this without looping
-    for segment_idx in range(SX1.get_num_segments()):
-        # get_unit_ids
-        ids1 = np.sort(np.array(SX1.get_unit_ids()))
-        ids2 = np.sort(np.array(SX2.get_unit_ids()))
-        assert_array_equal(ids1, ids2)
-        for id in ids1:
-            train1 = np.sort(SX1.get_unit_spike_train(id, segment_index=segment_idx))
-            train2 = np.sort(SX2.get_unit_spike_train(id, segment_index=segment_idx))
-            assert np.array_equal(train1, train2)
-            train1 = np.sort(SX1.get_unit_spike_train(id, segment_index=segment_idx, start_frame=30))
-            train2 = np.sort(SX2.get_unit_spike_train(id, segment_index=segment_idx, start_frame=30))
-            assert np.array_equal(train1, train2)
-            # test that slicing works correctly
-            train1 = np.sort(SX1.get_unit_spike_train(id, segment_index=segment_idx, end_frame=max_spike_index - 30))
-            train2 = np.sort(SX2.get_unit_spike_train(id, segment_index=segment_idx, end_frame=max_spike_index - 30))
-            assert np.array_equal(train1, train2)
-            train1 = np.sort(
-                SX1.get_unit_spike_train(id, segment_index=segment_idx, start_frame=30, end_frame=max_spike_index - 30)
-            )
-            train2 = np.sort(
-                SX2.get_unit_spike_train(id, segment_index=segment_idx, start_frame=30, end_frame=max_spike_index - 30)
-            )
-            assert np.array_equal(train1, train2)
+    s1 = SX1.to_spike_vector()
+    s2 = SX2.to_spike_vector()
+    assert_array_equal(s1, s2)
+
+    for start_frame, end_frame in [
+        (None, None),
+        (30, None),
+        (None, max_spike_index - 30),
+        (30, max_spike_index - 30),
+    ]:
+
+        slice1 = _slice_spikes(s1, start_frame, end_frame)
+        slice2 = _slice_spikes(s2, start_frame, end_frame)
+        assert np.array_equal(slice1, slice2)
 
     if check_annotations:
         check_extractor_annotations_equal(SX1, SX2)
@@ -155,3 +145,12 @@ def check_extractor_properties_equal(EX1, EX2) -> None:
 
     for property_name in EX1.get_property_keys():
         assert_array_equal(EX1.get_property(property_name), EX2.get_property(property_name))
+
+
+def _slice_spikes(spikes, start_frame=None, end_frame=None):
+    mask = np.ones(spikes.size, dtype=bool)
+    if start_frame is not None:
+        mask &= spikes["sample_index"] >= start_frame
+    if end_frame is not None:
+        mask &= spikes["sample_index"] <= end_frame
+    return spikes[mask]
diff --git a/src/spikeinterface/core/zarrextractors.py b/src/spikeinterface/core/zarrextractors.py
index c0c9f9b5d7..a13a54bf48 100644
--- a/src/spikeinterface/core/zarrextractors.py
+++ b/src/spikeinterface/core/zarrextractors.py
@@ -295,6 +295,8 @@ def __init__(self, folder_path: Path | str, storage_options: dict | None = None,
         spikes["unit_index"] = spikes_group["unit_index"][:]
         for i, (start, end) in enumerate(segment_slices_list):
             spikes["segment_index"][start:end] = i
+        spikes = spikes[np.lexsort((spikes["unit_index"], spikes["sample_index"], spikes["segment_index"]))]
+        self._cached_spike_vector = spikes
 
         for segment_index in range(num_segments):
             soring_segment = SpikeVectorSortingSegment(spikes, segment_index, unit_ids)