Compact compiled machine runtime

[SandroMaglione]

Summary

• replace the closure-heavy compiled statechart runtime with a compact class-backed process kernel whose operations are shared through the prototype
• materialize terminal waiting and snapshot-observation primitives only when those public capabilities are used, while leaving arbitrary Machine.logic processes on the general runtime
• preserve independent actor scheduling, child cleanup, terminal arbitration, and interruption semantics, with regression coverage for stopping a blocked compiled drain

Changeset

• Added or updated for a library or package-metadata change
• Not required because this PR does not change src/ or package.json

Validation

• pnpm check
• Relevant example checks, when examples changed
• Reviewed the automated type-performance report, when the public TypeScript API or inference changed
• Reviewed the automated runtime-performance report, when runtime behavior changed

The public API and type inference are unchanged. Local validation also included pnpm perf:types, three process-isolated base/candidate benchmark pairs, and matching full-duration runtime reports. The five-process CI comparison confirms the retained-memory reduction and positive runtime direction.

[github-actions]

Type performance

Measured with TypeScript 6.0.3 and skipLibCheck=true.

Scenario	Base	PR	Difference
Effect only	55	55	0 (0.0%)
Import effect-machine	55	55	0 (0.0%)
Machine.defineStates (3 states)	2,965	2,965	0 (0.0%)
Machine.make (3 states, 2 events)	8,715	8,715	0 (0.0%)
machine.handle (3 states, 2 transitions)	23,492	23,492	0 (0.0%)
machine.handle (depth 8)	117,075	117,075	0 (0.0%)
machine.handle (depth 12)	129,713	129,713	0 (0.0%)
machine.handle (depth 16)	145,055	145,055	0 (0.0%)
machine.handle (depth 24)	183,851	183,851	0 (0.0%)
machine.handle (wide depth 16)	214,189	214,189	0 (0.0%)
machine.handle (parallel/history/choice)	132,667	132,667	0 (0.0%)
machine.handle (4 successive calls)	128,674	128,674	0 (0.0%)
machine exact input/output/error/services	110,584	110,584	0 (0.0%)
execution adapter readiness	127,854	127,854	0 (0.0%)

Marginal instantiations are measured against the matching setup without that API call:

Scenario	Base	PR	Difference
Import effect-machine	0	0	0
Machine.defineStates (3 states)	2,910	2,910	0 (0.0%)
Machine.make (3 states, 2 events)	5,742	5,742	0 (0.0%)
machine.handle (3 states, 2 transitions)	14,777	14,777	0 (0.0%)
machine.handle (depth 8)	109,601	109,601	0 (0.0%)
machine.handle (depth 12)	121,039	121,039	0 (0.0%)
machine.handle (depth 16)	135,181	135,181	0 (0.0%)
machine.handle (depth 24)	171,577	171,577	0 (0.0%)
machine.handle (wide depth 16)	202,224	202,224	0 (0.0%)
machine.handle (parallel/history/choice)	114,331	114,331	0 (0.0%)
machine.handle (4 successive calls)	113,703	113,703	0 (0.0%)
machine exact input/output/error/services	100,670	100,670	0 (0.0%)
execution adapter readiness	102,092	102,092	0 (0.0%)

Check times (informational)

Scenario	Base	PR
Effect only	0.03s	0.03s
Import effect-machine	0.03s	0.03s
Machine.defineStates (3 states)	0.09s	0.09s
Machine.make (3 states, 2 events)	0.15s	0.15s
machine.handle (3 states, 2 transitions)	0.21s	0.22s
machine.handle (depth 8)	0.48s	0.48s
machine.handle (depth 12)	0.52s	0.52s
machine.handle (depth 16)	0.56s	0.55s
machine.handle (depth 24)	0.62s	0.62s
machine.handle (wide depth 16)	0.70s	0.68s
machine.handle (parallel/history/choice)	0.58s	0.57s
machine.handle (4 successive calls)	0.54s	0.53s
machine exact input/output/error/services	0.46s	0.47s
execution adapter readiness	0.54s	0.53s

Type instantiations are the comparison metric. Check time varies with runner load and is informational only.

[github-actions]

Runtime performance

Median of 5 independent benchmark processes on AMD EPYC 7763 64-Core Processor with Node v24.18.0.

Pull request baseline

Scenario	Effect Machine	XState 5	XState 6 alpha
Plan counter transitions	101,637 transitions/s	112,360 transitions/s	125,551 transitions/s
Drain burst with terminal fence	92,931 increments/s	414,195 increments/s	113,676 increments/s
Drain burst with a change observer	67,451 increments/s	411,008 increments/s	113,599 increments/s
Lookup and send to one child	86,465 increments/s	408,432 increments/s	108,929 increments/s
Start and stop a machine	21,841 machines/s	190,150 machines/s	74,543 machines/s
Start and stop a parent with one child	8,553 families/s	71,144 families/s	26,977 families/s

Effect runtime reference points

Scenario	Effect Machine	Effect runtime primitives
Start and stop a raw compiled process	35,343 processes/s	—
Start and interrupt a suspended fiber	—	169,176 fibers/s
Start and stop a queue worker	—	117,288 workers/s
Start and stop an actor shell	—	109,206 actors/s
Start and stop two actor shells	—	67,668 families/s
Update an owner-only mutable snapshot	—	317,863,954 updates/s
Update a synchronized snapshot	—	1,556,510 updates/s
Create, resolve, and await a terminal latch	—	1,471,642 latches/s

Memory profile	Effect Machine	XState 5	XState 6 alpha	Effect runtime primitives
Idle machine	2.2 KiB	3.8 KiB	6.1 KiB	—
Raw generic managed process	12.3 KiB	—	—	—
Raw compiled process	2.7 KiB	—	—	—
Two independent idle machines	4.0 KiB	7.3 KiB	11.8 KiB	—
Idle parent with one child	6.3 KiB	5.5 KiB	8.8 KiB	—
Parent with observed child registry	10.5 KiB	—	—	—
Parent with observed invoked child snapshots	6.5 KiB	—	—	—
Suspended Effect fiber	—	—	—	0.7 KiB
Effect queue with waiting fiber	—	—	—	2.8 KiB
Effect mailbox actor shell	—	—	—	3.2 KiB
Two Effect actor shells	—	—	—	6.9 KiB

Effect Machine change from base

Metric	Base	Base variability	PR	PR variability	Difference
Plan counter transitions	101,668 transitions/s	1.7% MAD	101,637 transitions/s	5.1% MAD	-0.0%
Drain burst with terminal fence	90,271 increments/s	3.1% MAD	92,931 increments/s	3.9% MAD	+2.9%
Drain burst with a change observer	66,056 increments/s	1.0% MAD	67,451 increments/s	5.1% MAD	+2.1%
Lookup and send to one child	83,630 increments/s	4.0% MAD	86,465 increments/s	3.8% MAD	+3.4%
Start and stop a machine	20,246 machines/s	1.4% MAD	21,841 machines/s	2.2% MAD	+7.9%
Start and stop a parent with one child	7,675 families/s	2.4% MAD	8,553 families/s	4.4% MAD	+11.4%
Idle machine heap per unit	7.0 KiB	0.0% MAD	2.2 KiB	0.0% MAD	-67.8%
Raw generic managed process heap per unit	12.3 KiB	0.1% MAD	12.3 KiB	0.0% MAD	+0.1%
Raw compiled process heap per unit	7.6 KiB	0.1% MAD	2.7 KiB	0.1% MAD	-64.3%
Two independent idle machines heap per unit	13.5 KiB	0.0% MAD	4.0 KiB	0.0% MAD	-69.9%
Idle parent with one child heap per unit	15.8 KiB	0.0% MAD	6.3 KiB	0.0% MAD	-60.0%
Parent with observed child registry heap per unit	19.9 KiB	0.0% MAD	10.5 KiB	0.0% MAD	-47.4%
Parent with observed invoked child snapshots heap per unit	15.9 KiB	0.0% MAD	6.5 KiB	0.0% MAD	-59.4%

Effect runtime reference change from base

Metric	Base	Base variability	PR	PR variability	Difference
Start and stop a raw compiled process	31,477 processes/s	0.8% MAD	35,343 processes/s	1.1% MAD	+12.3%
Start and interrupt a suspended fiber	168,890 fibers/s	0.0% MAD	169,176 fibers/s	0.5% MAD	+0.2%
Start and stop a queue worker	115,794 workers/s	1.7% MAD	117,288 workers/s	1.4% MAD	+1.3%
Start and stop an actor shell	107,910 actors/s	1.2% MAD	109,206 actors/s	1.1% MAD	+1.2%
Start and stop two actor shells	65,794 families/s	2.4% MAD	67,668 families/s	0.9% MAD	+2.8%
Update an owner-only mutable snapshot	316,856,781 updates/s	0.0% MAD	317,863,954 updates/s	0.0% MAD	+0.3%
Update a synchronized snapshot	1,505,197 updates/s	0.7% MAD	1,556,510 updates/s	1.3% MAD	+3.4%
Create, resolve, and await a terminal latch	1,443,593 latches/s	2.2% MAD	1,471,642 latches/s	1.5% MAD	+1.9%

Versions and interpretation

• Effect Machine: 0.3.0
• XState 5: 5.32.5
• XState 6 alpha: 6.0.0-alpha.27
• Effect runtime primitives: 4.0.0-beta.105

Higher throughput is better; lower heap is better. Variability is the median absolute deviation across independent processes, relative to their median. Runtime measurements on shared GitHub-hosted hardware remain informational, so small differences should be confirmed across multiple workflow runs.