Benchmarks

Each case has four logically equivalent implementations. kvlang is currently slower than Python, Rust, and C; these charts track how fast it converges.

Data from v0.3.24 · commit 531e4c7 · measured 2026-10-29 · Intel(R) Xeon(R) Platinum 8336C CPU @ 2.30GHz · raw CSV

Latest results

Full results (30 rows)

Case	Input	shm	fs	redis	Python	Rust	C
binary_search	N=16	199 µs	—	—	14.1 µs	738 ns	451 ns
binary_search	N=32	402 µs	—	—	23.7 µs	1.36 µs	956 ns
binary_search	N=64	851 µs	—	—	45.3 µs	2.94 µs	2.07 µs
binary_trees	depth=4	441 µs	—	—	14.4 µs	2.63 µs	5.29 µs
binary_trees	depth=5	843 µs	—	—	22.6 µs	3.12 µs	6.14 µs
binary_trees	depth=6	1.64 ms	—	—	40.7 µs	4.41 µs	8.75 µs
fib	n=8	56.1 µs	—	—	10.1 µs	321 ns	373 ns
fib	n=9	77.1 µs	—	—	12.0 µs	401 ns	444 ns
fib	n=10	117 µs	—	—	15.2 µs	466 ns	527 ns
hash_table	N=50	305 µs	—	—	33.6 µs	4.94 µs	265 ns
hash_table	N=100	583 µs	—	—	63.1 µs	8.07 µs	481 ns
hash_table	N=200	1.14 ms	—	—	113 µs	15.1 µs	727 ns
iops	N=500	13.8 µs	—	—	23.7 µs	475 ns	284 ns
iops	N=1000	14.3 µs	—	—	45.0 µs	800 ns	449 ns
iops	N=2000	14.8 µs	—	—	95.2 µs	1.47 µs	754 ns
k_nucleotide	rep=3	932 µs	—	—	39.5 µs	3.29 µs	225 ns
k_nucleotide	rep=5	1.59 ms	—	—	65.1 µs	4.97 µs	306 ns
k_nucleotide	rep=8	2.60 ms	—	—	101 µs	7.50 µs	416 ns
matmul	N=4	50.6 µs	—	—	20.1 µs	313 ns	232 ns
matmul	N=6	120 µs	—	—	54.1 µs	467 ns	364 ns
matmul	N=8	242 µs	—	—	116 µs	740 ns	565 ns
nqueens	N=4	39.1 µs	—	—	13.8 µs	299 ns	253 ns
nqueens	N=5	73.6 µs	—	—	23.7 µs	550 ns	526 ns
nqueens	N=6	171 µs	—	—	50.7 µs	1.60 µs	1.44 µs
prime_sieve	N=50	576 µs	—	—	41.6 µs	16.2 µs	19.2 µs
prime_sieve	N=75	762 µs	—	—	61.7 µs	17.7 µs	19.1 µs
prime_sieve	N=100	904 µs	—	—	82.1 µs	21.1 µs	21.7 µs
quicksort	N=32,seed=1	76.4 µs	—	—	18.5 µs	1.21 µs	1.01 µs
quicksort	N=64,seed=1	147 µs	—	—	33.7 µs	2.33 µs	2.19 µs
quicksort	N=128,seed=1	296 µs	—	—	66.0 µs	5.13 µs	4.46 µs

ns/µs/ms are self-adaptive. Missing values are shown as —. benchmark/README.md

Versions

Per version, three rows per case: wall time, ratio to Python, and the headline ratio to C (C = 1). Each cell stacks the case's three input points.

v0.3.24 · Intel(R) Xeon(R) Platinum 8336C CPU @ 2.30GHz · gcc -O1 / rustc opt-level=1 / cpython · measured 2026-10-29

v0.2.16 · Intel(R) Xeon(R) Platinum 8336C CPU @ 2.30GHz · gcc -O1 / rustc opt-level=1 / cpython · measured 2026-09-10

v0.2.14 · Intel(R) Xeon(R) Platinum 8336C CPU @ 2.30GHz · gcc -O1 / rustc opt-level=1 / cpython · measured 2026-09-09

v0.2.12 · Intel(R) Xeon(R) Platinum 8336C CPU @ 2.30GHz · gcc -O1 / rustc opt-level=1 / cpython · measured 2026-09-08

v0.2.9 · Intel(R) Xeon(R) Platinum 8336C CPU @ 2.30GHz · gcc -O1 / rustc opt-level=1 / cpython · measured 2026-09-08

v0.2.5 · Intel(R) Xeon(R) Platinum 8336C CPU @ 2.30GHz · gcc -O1 / rustc opt-level=1 / cpython · measured 2026-09-04

Version	binary_search	binary_trees	fib	hash_table	iops	k_nucleotide	matmul	nqueens	prime_sieve	quicksort
v0.3.24
time	N=16 199 µs N=32 402 µs N=64 851 µs	depth=4 441 µs depth=5 843 µs depth=6 1.64 ms	n=8 56.1 µs n=9 77.1 µs n=10 117 µs	N=50 305 µs N=100 583 µs N=200 1.14 ms	N=500 13.8 µs N=1000 14.3 µs N=2000 14.8 µs	rep=3 932 µs rep=5 1.59 ms rep=8 2.60 ms	N=4 50.6 µs N=6 120 µs N=8 242 µs	N=4 39.1 µs N=5 73.6 µs N=6 171 µs	N=50 576 µs N=75 762 µs N=100 904 µs	N=32,seed=1 76.4 µs N=64,seed=1 147 µs N=128,seed=1 296 µs
vs Python	N=16 14.1× N=32 17.0× N=64 18.8×	depth=4 30.5× depth=5 37.3× depth=6 40.4×	n=8 5.54× n=9 6.45× n=10 7.69×	N=50 9.07× N=100 9.25× N=200 10.1×	N=500 0.583× N=1000 0.317× N=2000 0.156×	rep=3 23.6× rep=5 24.4× rep=8 25.7×	N=4 2.52× N=6 2.22× N=8 2.09×	N=4 2.83× N=5 3.10× N=6 3.37×	N=50 13.8× N=75 12.4× N=100 11.0×	N=32,seed=1 4.14× N=64,seed=1 4.37× N=128,seed=1 4.48×
vs C	N=16 442× N=32 421× N=64 412×	depth=4 83.4× depth=5 137× depth=6 188×	n=8 150× n=9 174× n=10 221×	N=50 1,151× N=100 1,212× N=200 1,574×	N=500 48.7× N=1000 31.8× N=2000 19.7×	rep=3 4,143× rep=5 5,181× rep=8 6,262×	N=4 218× N=6 330× N=8 429×	N=4 155× N=5 140× N=6 119×	N=50 29.9× N=75 39.9× N=100 41.6×	N=32,seed=1 75.5× N=64,seed=1 67.2× N=128,seed=1 66.3×
v0.2.16
time	N=16 6.95 ms N=32 15.0 ms N=64 32.7 ms	depth=4 5.18 ms depth=5 9.60 ms depth=6 19.3 ms	n=8 7.47 ms n=9 12.1 ms n=10 19.8 ms	N=50 4.57 ms N=100 9.04 ms N=200 19.0 ms	N=500 7.45 ms N=1000 14.6 ms N=2000 28.9 ms	rep=3 10.5 ms rep=5 20.5 ms rep=8 39.8 ms	N=4 3.29 ms N=6 9.54 ms N=8 21.4 ms	N=4 5.22 ms N=5 15.8 ms N=6 45.2 ms	N=50 11.9 ms N=75 23.0 ms N=100 32.9 ms	N=32,seed=1 7.76 ms N=64,seed=1 17.0 ms N=128,seed=1 36.5 ms
vs Python	N=16 482× N=32 629× N=64 712×	depth=4 362× depth=5 433× depth=6 475×	n=8 726× n=9 1,016× n=10 1,291×	N=50 140× N=100 137× N=200 165×	N=500 312× N=1000 329× N=2000 320×	rep=3 257× rep=5 324× rep=8 406×	N=4 162× N=6 184× N=8 190×	N=4 374× N=5 670× N=6 890×	N=50 289× N=75 374× N=100 409×	N=32,seed=1 416× N=64,seed=1 503× N=128,seed=1 549×
vs C	N=16 14,640× N=32 15,153× N=64 15,679×	depth=4 830× depth=5 1,212× depth=6 2,176×	n=8 23,045× n=9 29,340× n=10 32,321×	N=50 16,937× N=100 20,172× N=200 24,364×	N=500 28,548× N=1000 33,716× N=2000 37,111×	rep=3 54,449× rep=5 72,402× rep=8 94,208×	N=4 17,123× N=6 27,820× N=8 38,678×	N=4 19,264× N=5 28,998× N=6 31,663×	N=50 691× N=75 1,185× N=100 1,397×	N=32,seed=1 7,393× N=64,seed=1 7,834× N=128,seed=1 7,894×
v0.2.14
time	N=16 10.1 ms N=32 22.0 ms N=64 48.8 ms	depth=4 7.31 ms depth=5 13.7 ms depth=6 27.4 ms	n=8 11.5 ms n=9 18.9 ms n=10 31.2 ms	N=50 6.60 ms N=100 12.9 ms N=200 27.0 ms	N=500 12.9 ms N=1000 25.5 ms N=2000 50.5 ms	rep=3 14.5 ms rep=5 27.0 ms rep=8 50.1 ms	N=4 5.00 ms N=6 14.8 ms N=8 33.5 ms	N=4 7.88 ms N=5 24.2 ms N=6 70.2 ms	N=50 17.3 ms N=75 33.4 ms N=100 48.8 ms	N=32,seed=1 11.8 ms N=64,seed=1 25.8 ms N=128,seed=1 55.2 ms
vs Python	N=16 727× N=32 897× N=64 1,062×	depth=4 518× depth=5 606× depth=6 684×	n=8 1,178× n=9 1,590× n=10 2,110×	N=50 199× N=100 213× N=200 242×	N=500 553× N=1000 568× N=2000 571×	rep=3 365× rep=5 426× rep=8 501×	N=4 254× N=6 287× N=8 296×	N=4 586× N=5 1,028× N=6 1,404×	N=50 439× N=75 552× N=100 615×	N=32,seed=1 631× N=64,seed=1 772× N=128,seed=1 835×
vs C	N=16 22,656× N=32 21,637× N=64 22,241×	depth=4 1,399× depth=5 2,111× depth=6 3,016×	n=8 32,173× n=9 47,317× n=10 53,551×	N=50 21,942× N=100 29,216× N=200 35,336×	N=500 46,173× N=1000 58,729× N=2000 65,883×	rep=3 64,893× rep=5 98,882× rep=8 114,343×	N=4 23,682× N=6 40,952× N=8 57,021×	N=4 33,395× N=5 45,008× N=6 48,696×	N=50 910× N=75 1,851× N=100 2,078×	N=32,seed=1 11,901× N=64,seed=1 11,749× N=128,seed=1 11,961×
v0.2.12
time	N=16 28.2 ms N=32 58.0 ms N=64 120 ms	depth=4 5.52 ms depth=5 10.4 ms depth=6 20.7 ms	n=8 55.4 ms n=9 90.6 ms n=10 147 ms	N=50 4.79 ms N=100 9.56 ms N=200 20.2 ms	N=500 10.0 ms N=1000 19.1 ms N=2000 37.1 ms	rep=3 11.5 ms rep=5 22.1 ms rep=8 42.6 ms	N=4 3.60 ms N=6 10.6 ms N=8 23.9 ms	N=4 31.8 ms N=5 103 ms N=6 297 ms	N=50 13.9 ms N=75 26.3 ms N=100 37.5 ms	N=32,seed=1 8.55 ms N=64,seed=1 18.9 ms N=128,seed=1 37.7 ms
vs Python	N=16 1,953× N=32 2,367× N=64 2,665×	depth=4 380× depth=5 463× depth=6 511×	n=8 5,572× n=9 7,340× n=10 9,653×	N=50 144× N=100 154× N=200 177×	N=500 425× N=1000 412× N=2000 412×	rep=3 297× rep=5 322× rep=8 425×	N=4 180× N=6 199× N=8 214×	N=4 2,344× N=5 4,384× N=6 5,848×	N=50 346× N=75 421× N=100 470×	N=32,seed=1 469× N=64,seed=1 559× N=128,seed=1 568×
vs C	N=16 57,588× N=32 59,280× N=64 56,745×	depth=4 961× depth=5 1,597× depth=6 2,448×	n=8 146,259× n=9 222,157× n=10 281,837×	N=50 16,647× N=100 21,061× N=200 28,109×	N=500 37,686× N=1000 42,300× N=2000 47,611×	rep=3 52,172× rep=5 75,311× rep=8 102,685×	N=4 13,964× N=6 31,191× N=8 41,405×	N=4 115,335× N=5 198,524× N=6 201,314×	N=50 728× N=75 1,137× N=100 1,652×	N=32,seed=1 8,613× N=64,seed=1 8,864× N=128,seed=1 8,094×
v0.2.9
time	N=16 33.3 ms N=32 68.7 ms N=64 144 ms	depth=4 9.75 ms depth=5 18.6 ms depth=6 36.6 ms	n=8 60.5 ms n=9 98.1 ms n=10 160 ms	N=50 8.67 ms N=100 17.1 ms N=200 35.1 ms	N=500 19.5 ms N=1000 38.4 ms N=2000 74.9 ms	rep=3 20.0 ms rep=5 36.0 ms rep=8 65.0 ms	N=4 7.45 ms N=6 22.6 ms N=8 51.1 ms	N=4 35.3 ms N=5 114 ms N=6 326 ms	N=50 25.5 ms N=75 49.2 ms N=100 71.4 ms	N=32,seed=1 16.9 ms N=64,seed=1 37.8 ms N=128,seed=1 81.4 ms
vs Python	N=16 2,331× N=32 2,855× N=64 3,148×	depth=4 677× depth=5 821× depth=6 903×	n=8 6,183× n=9 8,180× n=10 10,434×	N=50 261× N=100 277× N=200 311×	N=500 811× N=1000 842× N=2000 851×	rep=3 499× rep=5 557× rep=8 652×	N=4 373× N=6 431× N=8 447×	N=4 2,577× N=5 4,868× N=6 6,524×	N=50 630× N=75 790× N=100 889×	N=32,seed=1 897× N=64,seed=1 1,116× N=128,seed=1 1,247×
vs C	N=16 71,137× N=32 69,492× N=64 67,146×	depth=4 1,722× depth=5 2,841× depth=6 3,813×	n=8 188,924× n=9 229,771× n=10 267,682×	N=50 31,409× N=100 37,207× N=200 47,563×	N=500 71,231× N=1000 85,491× N=2000 96,043×	rep=3 96,812× rep=5 118,097× rep=8 149,359×	N=4 30,928× N=6 65,756× N=8 86,385×	N=4 134,878× N=5 211,755× N=6 224,955×	N=50 1,309× N=75 2,412× N=100 3,187×	N=32,seed=1 15,401× N=64,seed=1 17,901× N=128,seed=1 17,369×
v0.2.5
time	N=16 975 ms N=32 2.32 s N=64 5.43 s	depth=4 671 ms depth=5 1.36 s depth=6 2.74 s	n=8 2.11 s n=9 4.03 s n=10 7.64 s	N=50 625 ms N=100 1.24 s N=200 2.48 s	N=500 1.74 s N=1000 3.44 s N=2000 6.90 s	rep=3 1.50 s rep=5 2.50 s rep=8 4.00 s	N=4 596 ms N=6 1.85 s N=8 4.22 s	N=4 690 ms N=5 2.89 s N=6 10.2 s	N=50 2.71 s N=75 5.49 s N=100 8.17 s	N=32,seed=1 1.41 s N=64,seed=1 3.29 s N=128,seed=1 7.26 s
vs Python	N=16 69,062× N=32 95,046× N=64 119,699×	depth=4 46,943× depth=5 60,059× depth=6 67,221×	n=8 206,620× n=9 330,596× n=10 501,995×	N=50 18,799× N=100 20,406× N=200 21,936×	N=500 74,017× N=1000 76,641× N=2000 76,407×	rep=3 38,228× rep=5 39,332× rep=8 39,652×	N=4 29,005× N=6 33,545× N=8 38,009×	N=4 50,137× N=5 123,719× N=6 203,974×	N=50 66,069× N=75 89,703× N=100 101,013×	N=32,seed=1 75,908× N=64,seed=1 97,552× N=128,seed=1 110,700×
vs C	N=16 2,123,604× N=32 2,138,706× N=64 2,481,079×	depth=4 114,994× depth=5 231,924× depth=6 305,928×	n=8 6,400,197× n=9 9,860,498× n=10 13,326,081×	N=50 2,299,259× N=100 2,699,277× N=200 3,026,072×	N=500 5,863,882× N=1000 7,917,590× N=2000 9,025,180×	rep=3 6,609,949× rep=5 8,051,147× rep=8 8,904,458×	N=4 2,671,172× N=6 5,500,327× N=8 7,270,982×	N=4 2,481,606× N=5 5,782,815× N=6 6,828,429×	N=50 145,962× N=75 280,407× N=100 406,075×	N=32,seed=1 1,354,116× N=64,seed=1 1,560,778× N=128,seed=1 1,617,070×

All runs on the same CPU model unless noted. The v0.2.17 snapshot on a different CPU is excluded. Ratios are kvlang / baseline; a lower ratio is faster. C is the reference baseline (ratio to C = 1 is the primary metric). benchmark/README.md