Benchmarks

These numbers come from the in-repo benches under bench/, measured on an Apple M4. Absolute times vary by hardware, Studio build, and load; use the ratios and relative ordering as the main takeaway.

Each operation is timed with BenchHelper: 50 warm-up calls, then 10 timed batches. Reported Avg is mean microseconds per call across those batches.

VoidSentryUltimate throughput

Source: benchresults/benchresult.txt
Harness: LuauBench + RobloxBench
Iterations: 10,000 per timed batch

Values are average microseconds per call (μs). Lower is faster.

Scalars and strings

Type	Serialize	Deserialize	RoundTrip
U8	0.05	0.02	0.06
U40	0.05	0.02	0.07
I40	0.05	0.02	0.07
U48	0.05	0.02	0.07
I48	0.05	0.02	0.07
UInt (1 byte)	0.05	0.02	0.06
Int (1 byte)	0.05	0.02	0.07
UInt (4 byte)	0.05	0.02	0.07
Int (4 byte)	0.05	0.02	0.08
F32	0.05	0.02	0.06
F64	0.05	0.02	0.06
Bool	0.05	0.02	0.06
String (100 chars)	0.08	0.06	0.14

Vectors, buffers, and Roblox types

Type	Serialize	Deserialize	RoundTrip
Vector	0.05	0.02	0.06
VectorF16	0.07	0.03	0.09
BufferFixed (32b)	0.07	0.05	0.12
Buffer8 (33b payload)	0.08	0.05	0.13
Color3	0.09	0.05	0.08
UDim2Quant	0.07	0.04	0.10
CFrame	0.09	0.05	0.14
QCFrame	0.10	0.05	0.16
CFrameF16	0.14	0.08	0.23
CFrameQuantF16	0.11	0.09	0.21
CFrameQuant8F16	0.11	0.10	0.22
Instance	0.05	0.02	0.07
SerInstance	0.19	0.88	1.10

Collections

Type	Serialize	Deserialize	RoundTrip
Array (10 × U8)	0.14	0.14	0.29
Array (100 × U8)	0.90	1.02	1.93
Map (10 × U8→U8)	0.23	0.33	0.56
Map (100 × U8→U8)	1.60	2.20	3.93
Schema (10 × U8)	0.18	0.32	0.51
Schema (100 × U8)	1.53	2.83	4.47

Primitive nodes stay near 0.05 μs serialize and 0.02 μs deserialize. The 40-bit and 48-bit integer nodes, and both the 1-byte and 4-byte UInt / Int paths, remain in that range. Cost grows mainly with collection size and with compressed / property-driven types such as CFrameF16 and SerInstance.

Versus Sera

Source: benchresults/seravssentryresult.txt
Harness: bench/SeraVsVoidSentry/Comparator.luau
Iterations: 10,000 per timed batch

Sera only serializes through schemas. For primitives, vectors, and CFrames the comparator wraps Sera values in a single-field schema, while VoidSentryUltimate uses bare nodes. Schema rows compare Sera.Schema vs VoidSentryUltimate.Schema. Delta rows compare Sera.DeltaSerialize / DeltaDeserialize vs VoidSentryUltimate.Schema.DeltaSerialize / DeltaDeserialize. Push compares Sera.Push / DeltaPush vs VoidSentryUltimate.Schema.Push / Schema.DeltaPush.

Speedup is Sera Avg ÷ VoidSentry Avg. Values above 1.00x mean VoidSentry was faster.

Overall (suite averages)

Category	Serialize	Deserialize	Total / RoundTrip
Primitives	1.51x VS	2.13x VS	1.77x VS
Schemas	1.25x VS	1.01x VS	1.09x VS
Deltas	1.38x VS	1.14x VS	1.24x VS
Push (avg μs)	VS 1.169 / Sera 1.086	—	—

Floats, vectors, and CFrames

Operation	Sera (μs)	VoidSentry (μs)	Speedup
F32 Serialize	0.09	0.05	1.80x
F32 Deserialize	0.05	0.02	2.50x
F32 RoundTrip	0.14	0.06	2.33x
Vector3 Serialize	0.09	0.05	1.80x
Vector3 Deserialize	0.05	0.02	2.50x
Vector3 RoundTrip	0.14	0.07	2.00x
CFrame Serialize	0.14	0.10	1.40x
CFrame Deserialize	0.09	0.05	1.80x
CFrame RoundTrip	0.23	0.15	1.53x
LossyCFrame vs QCFrame Serialize	0.14	0.11	1.27x
LossyCFrame vs QCFrame Deserialize	0.11	0.05	2.20x

Schemas (U8 fields)

Operation	Sera (μs)	VoidSentry (μs)	Speedup
1 field Serialize	0.09	0.06	1.50x
1 field Deserialize	0.05	0.05	1.00x
10 fields Serialize	0.23	0.18	1.28x
10 fields Deserialize	0.32	0.32	1.00x
100 fields Serialize	1.82	1.47	1.24x
100 fields Deserialize	2.82	2.80	1.01x
100 fields RoundTrip	4.73	4.39	1.08x

Deltas (Schema.DeltaSerialize vs Sera delta)

Operation	Sera (μs)	VoidSentry (μs)	Speedup
10 fields, all present Serialize	0.26	0.19	1.37x
10 fields, all present Deserialize	0.36	0.33	1.09x
100 fields, all present Serialize	2.20	1.60	1.38x
100 fields, only 1 present Serialize	0.09	0.07	1.29x
255 fields, only 1 present Serialize	0.09	0.06	1.50x

Push

Operation	Sera (μs)	VoidSentry (μs)	Speedup
Schema Push (100 fields)	1.70	1.45	1.17x
DeltaPush (100 fields, only 1 present)	0.06	0.05	1.20x

Summary

• Primitives are VoidSentry’s clearest win (~1.5× serialize, ~2.1× deserialize, ~1.8× round-trip on the suite averages; bare F32 / Vector3 deserialize is 2.5×).
• Schemas stay close: VoidSentry leads serialize (1.25×) and total (1.09×); deserialize is even.
• Deltas favor VoidSentry (~1.38× serialize, ~1.14× deserialize, ~1.24× total), including sparse “only 1 present” cases.
• Push suite average slightly favors Sera (1.086 μs vs 1.169 μs). Individual rows still vary: 100-field Schema.Push and sparse DeltaPush favor VoidSentry.
• Prefer schema tables for schema-to-schema comparisons; prefer primitive tables when comparing each library’s natural single-value API.

LuauBench also includes a VoidSentryUltimate.Schema section for the pure-Luau package.

Reproducing

-- VoidSentry-only throughput (Luau subset + Roblox types)
local LuauBench = require(path.to.bench.LuauBench)
local RobloxBench = require(path.to.bench.RobloxBench)
LuauBench(true, 10_000)
RobloxBench(true, 10_000)

-- VoidSentry vs Sera
local Comparator = require(path.to.bench.SeraVsVoidSentry.Comparator)
Comparator(true, 10_000)

Raw logs used for this page live in benchresults/.