Performance Report (latest run)

Environment

• Generated at: 2026-05-14T03:31:14.006Z
• Node.js: v23.11.0
• Platform: darwin arm64
• CPU: Apple M1 Pro
• CPU count: 10
• Commit: 946195e2f2980521974f02bcf415839fd30ad770

Default API note: normal md.parse(src) / md.render(src) calls may auto-activate an internal large-input path for very large finite strings only when no plugin has been installed and parser rulers have not been modified. Explicit chunk-stream APIs such as parseIterable / UnboundedBuffer are advanced tools for sources that already arrive as chunks.

Size (chars)	S1 one	S2 one	S3 one	S4 one	S5 one	M1 one	E1 one	MM1 one	S1 append(par)	S2 append(par)	S3 append(par)	S4 append(par)	S5 append(par)	M1 append(par)	E1 append(par)	MM1 append(par)	S1 append(line)	S2 append(line)	S3 append(line)	S4 append(line)	S5 append(line)	M1 append(line)	E1 append(line)	MM1 append(line)	S1 replace	S2 replace	S3 replace	S4 replace	S5 replace	M1 replace	E1 replace	MM1 replace
5000	0.2208ms	0.1619ms	0.1911ms	0.1768ms	0.1448ms	0.2630ms	0.3136ms	4.2385ms	0.6086ms	0.3279ms	0.2794ms	0.5395ms	0.5131ms	0.5723ms	0.7424ms	12.99ms	1.7556ms	0.5609ms	0.5331ms	1.4633ms	1.5194ms	1.5625ms	2.1217ms	36.92ms	0.2244ms	0.1785ms	0.1841ms	0.2179ms	0.1419ms	0.2341ms	0.3307ms	3.9680ms
20000	0.7819ms	0.6218ms	0.7865ms	0.7438ms	0.5903ms	0.6548ms	0.8680ms	17.81ms	2.5770ms	1.1127ms	1.2151ms	2.5687ms	2.0970ms	2.2415ms	2.8933ms	56.41ms	7.2616ms	1.8999ms	2.1059ms	6.8992ms	5.9544ms	6.2729ms	8.0455ms	159.21ms	0.8149ms	0.6564ms	0.8021ms	0.7207ms	0.5925ms	0.6525ms	0.8827ms	16.68ms
50000	2.0226ms	1.6595ms	2.0423ms	1.9661ms	1.5677ms	1.7021ms	2.2452ms	46.87ms	6.6239ms	2.5873ms	2.6207ms	6.5004ms	5.4719ms	5.6121ms	7.2617ms	158.71ms	17.88ms	4.4156ms	4.5650ms	17.39ms	15.08ms	15.34ms	20.00ms	468.75ms	1.8888ms	1.6933ms	1.9212ms	1.8263ms	1.4929ms	1.6272ms	2.2480ms	44.43ms
100000	5.2168ms	4.2114ms	4.7431ms	4.8337ms	3.7770ms	4.6536ms	5.5021ms	96.96ms	15.05ms	5.4172ms	5.6964ms	14.03ms	12.60ms	11.53ms	14.96ms	319.31ms	37.55ms	9.0520ms	8.9477ms	36.18ms	33.31ms	30.88ms	41.58ms	908.11ms	4.0351ms	3.5826ms	4.0356ms	3.8006ms	3.7302ms	3.6258ms	4.7358ms	95.29ms
200000	10.69ms	10.10ms	10.71ms	10.63ms	10.15ms	10.27ms	12.17ms	199.53ms	31.10ms	11.28ms	12.59ms	29.30ms	27.64ms	24.87ms	34.61ms	688.17ms	77.18ms	19.11ms	25.00ms	74.80ms	79.62ms	70.53ms	95.46ms	1832.38ms	8.0693ms	9.5296ms	8.0764ms	7.8744ms	7.7813ms	6.9167ms	9.5415ms	184.39ms
500000	25.56ms	25.61ms	26.90ms	24.26ms	27.16ms	24.78ms	32.02ms	-	83.00ms	42.00ms	40.49ms	82.31ms	92.10ms	96.89ms	101.18ms	-	238.79ms	68.99ms	63.77ms	234.77ms	235.36ms	238.51ms	311.47ms	-	25.28ms	22.57ms	24.92ms	24.83ms	26.98ms	25.52ms	29.94ms	-
1000000	54.41ms	107.42ms	58.58ms	47.06ms	49.35ms	59.67ms	60.02ms	-	189.76ms	381.58ms	80.26ms	183.58ms	194.09ms	190.48ms	220.42ms	-	691.79ms	141.57ms	131.01ms	488.35ms	492.67ms	500.65ms	605.22ms	-	281.73ms	53.14ms	57.16ms	49.63ms	55.58ms	52.53ms	69.71ms	-

Best (one-shot) per size:

• 5000: S5 0.1448ms (stream OFF, chunk OFF)
• 20000: S5 0.5903ms (stream OFF, chunk OFF)
• 50000: S5 1.5677ms (stream OFF, chunk OFF)
• 100000: S5 3.7770ms (stream OFF, chunk OFF)
• 200000: S2 10.10ms (stream ON, cache ON, chunk OFF)
• 500000: S4 24.26ms (stream OFF, chunk ON)
• 1000000: S4 47.06ms (stream OFF, chunk ON)

Best (append workload) per size:

• 5000: S3 0.2794ms (stream ON, cache ON, chunk ON)
• 20000: S2 1.1127ms (stream ON, cache ON, chunk OFF)
• 50000: S2 2.5873ms (stream ON, cache ON, chunk OFF)
• 100000: S2 5.4172ms (stream ON, cache ON, chunk OFF)
• 200000: S2 11.28ms (stream ON, cache ON, chunk OFF)
• 500000: S3 40.49ms (stream ON, cache ON, chunk ON)
• 1000000: S3 80.26ms (stream ON, cache ON, chunk ON)

Best (line-append workload) per size:

• 5000: S3 0.5331ms (stream ON, cache ON, chunk ON)
• 20000: S2 1.8999ms (stream ON, cache ON, chunk OFF)
• 50000: S2 4.4156ms (stream ON, cache ON, chunk OFF)
• 100000: S3 8.9477ms (stream ON, cache ON, chunk ON)
• 200000: S2 19.11ms (stream ON, cache ON, chunk OFF)
• 500000: S3 63.77ms (stream ON, cache ON, chunk ON)
• 1000000: S3 131.01ms (stream ON, cache ON, chunk ON)

Best (replace-paragraph workload) per size:

• 5000: S5 0.1419ms (stream OFF, chunk OFF)
• 20000: S5 0.5925ms (stream OFF, chunk OFF)
• 50000: S5 1.4929ms (stream OFF, chunk OFF)
• 100000: S2 3.5826ms (stream ON, cache ON, chunk OFF)
• 200000: M1 6.9167ms (markdown-it (baseline))
• 500000: S2 22.57ms (stream ON, cache ON, chunk OFF)
• 1000000: S4 49.63ms (stream OFF, chunk ON)

Recommendations (by majority across sizes):

• One-shot: S5(4), S4(2), S2(1)
• Append-heavy: S2(4), S3(3)

Notes: S2/S3 appendHits should equal 5 when append fast-path triggers (shared env). Large-size rows may show - for especially heavy parse-only or render-only baselines (currently remark/micromark above 200k) so perf:all stays practical.

Render API throughput (markdown → HTML)

This measures end-to-end md.render(markdown) throughput across markdown-it-ts, upstream markdown-it, micromark (CommonMark reference), and remark+rehype (parse + stringify). Lower is better. It is intentionally a full render-API benchmark (parse + render), not a renderer-only hot-path benchmark.

Size (chars)	markdown-it-ts.render	markdown-it.render	micromark	remark+rehype	markdown-exit
5000	0.1793ms	0.2059ms	3.6437ms	4.2929ms	0.2487ms
20000	0.6677ms	0.7787ms	19.91ms	24.47ms	0.9981ms
50000	1.8301ms	2.0047ms	55.10ms	74.42ms	2.6056ms
100000	4.6034ms	5.3503ms	107.22ms	170.78ms	6.3543ms
200000	11.28ms	13.16ms	214.15ms	377.11ms	14.93ms
500000	31.15ms	37.07ms	-	-	42.39ms
1000000	66.77ms	78.51ms	-	-	81.11ms

Render vs markdown-it:

• 5,000 chars: 0.1793ms vs 0.2059ms → 1.15× faster
• 20,000 chars: 0.6677ms vs 0.7787ms → 1.17× faster
• 50,000 chars: 1.8301ms vs 2.0047ms → 1.10× faster
• 100,000 chars: 4.6034ms vs 5.3503ms → 1.16× faster
• 200,000 chars: 11.28ms vs 13.16ms → 1.17× faster
• 500,000 chars: 31.15ms vs 37.07ms → 1.19× faster
• 1,000,000 chars: 66.77ms vs 78.51ms → 1.18× faster

Render vs micromark:

• 5,000 chars: 0.1793ms vs 3.6437ms → 20.32× faster
• 20,000 chars: 0.6677ms vs 19.91ms → 29.82× faster
• 50,000 chars: 1.8301ms vs 55.10ms → 30.11× faster
• 100,000 chars: 4.6034ms vs 107.22ms → 23.29× faster
• 200,000 chars: 11.28ms vs 214.15ms → 18.98× faster

Render vs remark+rehype:

• 5,000 chars: 0.1793ms vs 4.2929ms → 23.94× faster
• 20,000 chars: 0.6677ms vs 24.47ms → 36.64× faster
• 50,000 chars: 1.8301ms vs 74.42ms → 40.67× faster
• 100,000 chars: 4.6034ms vs 170.78ms → 37.10× faster
• 200,000 chars: 11.28ms vs 377.11ms → 33.42× faster

Render vs markdown-exit:

• 5,000 chars: 0.1793ms vs 0.2487ms → 1.39× faster
• 20,000 chars: 0.6677ms vs 0.9981ms → 1.49× faster
• 50,000 chars: 1.8301ms vs 2.6056ms → 1.42× faster
• 100,000 chars: 4.6034ms vs 6.3543ms → 1.38× faster
• 200,000 chars: 11.28ms vs 14.93ms → 1.32× faster
• 500,000 chars: 31.15ms vs 42.39ms → 1.36× faster
• 1,000,000 chars: 66.77ms vs 81.11ms → 1.21× faster

Best-of markdown-it-ts vs markdown-it (baseline)

Size (chars)	TS best one	Baseline one	One comparison	TS best append	Baseline append	Append comparison	TS scenario (one/append)
5000	0.1448ms	0.2630ms	1.82× faster, 45% less time	0.2794ms	0.5723ms	2.05× faster, 51.2% less time	S5/S3
20000	0.5903ms	0.6548ms	1.11× faster, 9.9% less time	1.1127ms	2.2415ms	2.01× faster, 50.4% less time	S5/S2
50000	1.5677ms	1.7021ms	1.09× faster, 7.9% less time	2.5873ms	5.6121ms	2.17× faster, 53.9% less time	S5/S2
100000	3.7770ms	4.6536ms	1.23× faster, 18.8% less time	5.4172ms	11.53ms	2.13× faster, 53% less time	S5/S2
200000	10.10ms	10.27ms	1.02× faster, 1.7% less time	11.28ms	24.87ms	2.2× faster, 54.6% less time	S2/S2
500000	24.26ms	24.78ms	1.02× faster, 2.1% less time	40.49ms	96.89ms	2.39× faster, 58.2% less time	S4/S3
1000000	47.06ms	59.67ms	1.27× faster, 21.1% less time	80.26ms	190.48ms	2.37× faster, 57.9% less time	S4/S3

• Comparison columns are written from markdown-it-ts against the markdown-it baseline.
• faster / less time is better; if a future run regresses, the wording will flip to slower / more time.

Diagnostic: Chunk Info (if chunked)

Size (chars)	S1 one chunks	S3 one chunks	S4 one chunks	S1 append last	S3 append last	S4 append last
5000	4	4	4	4	4	4
20000	8	8	8	8	8	8
50000	8	8	8	8	8	8
100000	8	8	8	8	8	8
200000	8	8	8	8	8	8
500000	8	8	8	8	8	8
1000000	15	8	16	15	8	16

Cold vs Hot (one-shot)

Cold-start parses instantiate a new parser and run once with no warmup. Hot parses use a fresh instance with warmup plus averaged runs. 表格按不同文档大小分别列出 markdown-it 与 remark 对照。

5,000 chars

Impl	Cold	Hot
markdown-exit	0.2338ms	0.2092ms
markdown-it (baseline)	0.1712ms	0.1478ms
markdown-it-ts (stream+chunk)	0.2097ms	0.1829ms
micromark (parse only)	3.6626ms	3.5695ms
remark (parse only)	4.3328ms	3.8400ms

20,000 chars

Impl	Cold	Hot
markdown-exit	0.8468ms	0.8731ms
markdown-it (baseline)	0.6471ms	0.6818ms
markdown-it-ts (stream+chunk)	0.8078ms	0.7828ms
micromark (parse only)	16.58ms	16.51ms
remark (parse only)	21.43ms	21.07ms

50,000 chars

Impl	Cold	Hot
markdown-exit	2.0236ms	2.3363ms
markdown-it (baseline)	1.4881ms	1.7086ms
markdown-it-ts (stream+chunk)	1.8599ms	2.0154ms
micromark (parse only)	40.32ms	44.85ms
remark (parse only)	65.53ms	62.59ms

100,000 chars

Impl	Cold	Hot
markdown-exit	4.1545ms	5.9437ms
markdown-it (baseline)	4.5372ms	6.1906ms
markdown-it-ts (stream+chunk)	3.7872ms	5.4689ms
micromark (parse only)	219.89ms	216.99ms
remark (parse only)	141.53ms	137.41ms