vs Grafeo

LoraDB vs Grafeo

LoraDB leads by 2.30× overall. Grafeo is an in-process Rust property-graph crate. Numbers from the same suite as the overview — identical fixtures, identical seed, identical iteration count.

80 comparable workloads2 omissions12 groups

2.30×slower

Grafeo vs LoraDB overall

Geometric-mean slowdown across the full suite.

52/ 80

Workloads won by LoraDB

Grafeo wins 9; 19 go to a third engine.

0close calls

Within ~10% on both sides

Workloads where neither engine ran away with the row.

Same suite as the overview — only the comparison engine differs.

Overall

LoraDB sets the row; Grafeo sits at 2.30× slower.

Geometric-mean slowdown across every workload both engines run. Empty rows happen when the language can’t express the workload — they’re called out below, not hidden.

LoraDBfastest

Grafeo2.30× slower

Per group

Wins by group.

For each group, the workload count, who wins it, and how the per-row tally breaks down between LoraDB, Grafeo, and any third engine that took a row.

setup1 workload · winner Grafeo
LoraDB2.70× slower
Grafeofastest
Rows · 10·1
writes9 workloads · winner Grafeo
LoraDB0.61× slower
Grafeofastest
Rows · 93·5·1
scans6 workloads · winner LoraDB
LoraDBfastest
Grafeo5.43× slower
Rows · 65·0·1
predicates12 workloads · winner LoraDB
LoraDBfastest
Grafeo1.83× slower
Rows · 128·1·3
strings5 workloads · winner Kuzu
LoraDB1.22× slower
Grafeo1.84× slower
Rows · 50·0·5
numerics6 workloads · winner Kuzu
LoraDB1.08× slower
Grafeo1.87× slower
Rows · 61·0·5
aggregates9 workloads · winner LoraDB
LoraDBfastest
Grafeo1.89× slower
Rows · 98·1
pipeline9 workloads · winner LoraDB
LoraDBfastest
Grafeo1.30× slower
Rows · 95·1·3
lists3 workloads · winner LoraDB
LoraDBfastest
Grafeo3.21× slower
Rows · 32·0·1
sort3 workloads · winner LoraDB
LoraDBfastest
Grafeo1.60× slower
Rows · 33·0
traversals14 workloads · winner LoraDB
LoraDBfastest
Grafeo5.88× slower
Rows · 1414·0
patterns3 workloads · winner LoraDB
LoraDBfastest
Grafeo2.45× slower
Rows · 33·0

Per workload

Raw timings, workload by workload.

Two columns — LoraDB and Grafeo — across every workload they share. Rows with no comparable Grafeorun are hidden here; they’re listed in the omissions block at the bottom.

setup(1)

WinnerGrafeo

Workload	Size	LoraDB	Grafeo	Winner
construct_empty	—	4.06 µs2.70× slower	1.50 µsfastest	Grafeo

writes(9)

WinnerGrafeo

Workload	Size	LoraDB	Grafeo	Winner
bulk_edges	200	606.63 µsfastest	19.86 ms32.7× slower	LoraDB
bulk_set_match	1000	552.39 µs1.84× slower	315.88 µs1.05× slower	Kuzu
delete_node	1000	355.70 µs1.47× slower	241.43 µsfastest	Grafeo
merge_create	1000	112.57 µs1.25× slower	90.02 µsfastest	Grafeo
merge_existing	1000	23.95 µs2.89× slower	8.29 µsfastest	Grafeo
set_multiple_props	1000	21.15 µsfastest	182.77 µs8.64× slower	LoraDB
update_set	1000	18.01 µsfastest	208.61 µs11.6× slower	LoraDB
write_bulk	1000	1.46 ms1.93× slower	756.21 µsfastest	Grafeo
write_single	1000	14.90 µs2.20× slower	6.78 µsfastest	Grafeo

scans(6)

WinnerLoraDB

Workload	Size	LoraDB	Grafeo	Winner
distinct	1000	198.38 µsfastest	246.37 µs1.24× slower	LoraDB
lookup_by_id	1000	716.64 nsfastest	170.03 µs237× slower	LoraDB
lookup_by_id_indexed	1000	684.00 nsfastest	22.73 µs33.2× slower	LoraDB
range_filter	1000	199.70 µs1.04× slower	214.22 µs1.11× slower	Kuzu
scan_filtered	1000	149.08 µsfastest	211.95 µs1.42× slower	LoraDB
scan_label	1000	124.96 µsfastest	213.71 µs1.71× slower	LoraDB

predicates(12)

WinnerLoraDB

Workload	Size	LoraDB	Grafeo	Winner
where_compound_and_or	1000	213.07 µsfastest	646.50 µs3.03× slower	LoraDB
where_contains	1000	145.04 µsfastest	255.53 µs1.76× slower	LoraDB
where_ends_with	1000	143.60 µsfastest	236.22 µs1.64× slower	LoraDB
where_id_in_range	1000	142.71 µs2.02× slower	70.55 µsfastest	Grafeo
where_in_list	1000	162.42 µsfastest	270.30 µs1.66× slower	LoraDB
where_modulo_eq	1000	127.32 µsfastest	250.88 µs1.97× slower	LoraDB
where_not	1000	166.05 µs1.01× slower	340.52 µs2.08× slower	Kuzu
where_or	1000	147.16 µsfastest	450.08 µs3.06× slower	LoraDB
where_starts_with	1000	146.91 µsfastest	249.22 µs1.70× slower	LoraDB
where_string_gte	1000	180.97 µs1.07× slower	247.34 µs1.47× slower	Kuzu
where_subexpr	1000	227.59 µs1.69× slower	589.01 µs4.37× slower	SurrealDB
where_two_props	1000	152.69 µsfastest	404.15 µs2.65× slower	LoraDB

strings(5)

WinnerKuzu

Workload	Size	LoraDB	Grafeo	Winner
string_concat	1000	186.55 µs1.27× slower	283.23 µs1.92× slower	Kuzu
string_size	1000	172.17 µs1.10× slower	251.12 µs1.61× slower	Kuzu
string_substring	1000	210.20 µs1.21× slower	314.23 µs1.81× slower	Kuzu
string_to_lower	1000	201.00 µs1.33× slower	297.66 µs1.97× slower	Kuzu
string_to_upper	1000	185.41 µs1.22× slower	292.95 µs1.93× slower	Kuzu

numerics(6)

WinnerKuzu

Workload	Size	LoraDB	Grafeo	Winner
numeric_abs	1000	174.85 µs1.13× slower	272.48 µs1.76× slower	Kuzu
numeric_ceil	1000	171.69 µs1.10× slower	269.13 µs1.73× slower	Kuzu
numeric_floor	1000	175.48 µs1.10× slower	326.82 µs2.05× slower	Kuzu
numeric_modulo	1000	144.95 µs1.02× slower	236.39 µs1.66× slower	Kuzu
numeric_pow	1000	166.22 µs1.16× slower	394.61 µs2.75× slower	Kuzu
numeric_round	1000	180.68 µsfastest	269.75 µs1.49× slower	LoraDB

aggregates(9)

WinnerLoraDB

Workload	Size	LoraDB	Grafeo	Winner
aggregate_avg	1000	81.43 µsfastest	202.26 µs2.48× slower	LoraDB
aggregate_collect	1000	78.90 µsfastest	214.58 µs2.72× slower	LoraDB
aggregate_count	1000	59.50 µs2.77× slower	21.44 µsfastest	Grafeo
aggregate_count_distinct	1000	104.05 µsfastest	213.10 µs2.05× slower	LoraDB
aggregate_max	1000	79.97 µsfastest	200.95 µs2.51× slower	LoraDB
aggregate_min	1000	79.67 µsfastest	197.17 µs2.47× slower	LoraDB
aggregate_sum	1000	78.09 µsfastest	199.65 µs2.56× slower	LoraDB
grouped_aggregation	1000	154.46 µsfastest	269.45 µs1.74× slower	LoraDB
top_k	1000	186.50 µsfastest	424.61 µs2.28× slower	LoraDB

pipeline(9)

WinnerLoraDB

Workload	Size	LoraDB	Grafeo	Winner
case_when	1000	173.33 µsfastest	256.32 µs1.48× slower	LoraDB
coalesce_existing	1000	161.92 µsfastest	260.13 µs1.61× slower	LoraDB
computed_in_return	1000	151.15 µs1.07× slower	241.22 µs1.71× slower	Kuzu
distinct_with_order	1000	508.01 µs2.08× slower	243.88 µsfastest	Grafeo
predicate_via_function	1000	238.39 µs1.38× slower	434.70 µs2.52× slower	Kuzu
with_aggregate_then_filter	1000	148.70 µsfastest	267.49 µs1.80× slower	LoraDB
with_distinct_then_count	1000	202.72 µsfastest	244.75 µs1.21× slower	LoraDB
with_pipeline	1000	186.41 µsfastest	218.95 µs1.17× slower	LoraDB
with_two_chained	1000	313.64 µs1.41× slower	388.95 µs1.75× slower	Kuzu

lists(3)

WinnerLoraDB

Workload	Size	LoraDB	Grafeo	Winner
list_in_construction	1000	177.94 µs1.05× slower	450.42 µs2.67× slower	Kuzu
list_unwind_explicit	1000	1.11 µsfastest	7.11 µs6.40× slower	LoraDB
range_function	1000	18.86 µsfastest	38.62 µs2.05× slower	LoraDB

sort(3)

WinnerLoraDB

Workload	Size	LoraDB	Grafeo	Winner
order_by_id_asc	1000	167.95 µsfastest	226.74 µs1.35× slower	LoraDB
order_by_multi_key	1000	211.99 µsfastest	442.71 µs2.09× slower	LoraDB
skip_limit	1000	162.19 µsfastest	233.92 µs1.44× slower	LoraDB

traversals(14)

WinnerLoraDB

Workload	Size	LoraDB	Grafeo	Winner
direct_record_traversal	500	831.70 nsfastest	64.09 µs77.1× slower	LoraDB
recursive_depth2	500	968.69 nsfastest	62.29 µs64.3× slower	LoraDB
recursive_depth3	500	1.04 µsfastest	62.87 µs60.5× slower	LoraDB
recursive_depth5	500	1.16 µsfastest	62.30 µs53.7× slower	LoraDB
traversal_count_one_hop	500	59.36 µsfastest	141.39 µs2.38× slower	LoraDB
traversal_filter_one_hop	500	138.99 µsfastest	259.07 µs1.86× slower	LoraDB
traversal_one_hop	500	125.98 µsfastest	263.65 µs2.09× slower	LoraDB
traversal_reverse	500	123.67 µsfastest	266.69 µs2.16× slower	LoraDB
traversal_three_hop	500	236.10 µsfastest	556.10 µs2.36× slower	LoraDB
traversal_two_hop	500	165.58 µsfastest	411.83 µs2.49× slower	LoraDB
traversal_undirected	500	213.41 µsfastest	494.10 µs2.32× slower	LoraDB
variable_length_path	100	86.24 µsfastest	210.43 µs2.44× slower	LoraDB
varlen_2_to_5	100	123.74 µsfastest	274.33 µs2.22× slower	LoraDB
varlen_exact_5	100	56.89 µsfastest	140.52 µs2.47× slower	LoraDB

patterns(3)

WinnerLoraDB

Workload	Size	LoraDB	Grafeo	Winner
star_fanout	1000	138.96 µsfastest	321.31 µs2.31× slower	LoraDB
star_fanout_count	1000	61.36 µsfastest	193.38 µs3.15× slower	LoraDB
star_fanout_filter	1000	112.53 µsfastest	227.60 µs2.02× slower	LoraDB

Honest omissions

Workloads Grafeo couldn’t run like-for-like.

Omissions

Workloads omitted only where the engine has no like-for-like equivalent — a missing scalar function, a different MERGE semantics, a reserved word. The reason is recorded with the workload, not silently dropped.

relation_filter
Grafeo
- grafeo's `create_edge` facade takes no edge properties, so the chain fixture has no `step` to filter on. memgraph/neo4j/kuzu seed `step` directly, so they do run this workload.
edge_subquery_clause
Grafeo
- grafeo's `create_edge` facade takes no edge properties, so the social fixture has no `strength` to filter on. memgraph/neo4j/kuzu seed `strength` directly, so they do run this workload.