zenbench 1774487377-5ec33

git: c8cb79faad7dbb87456f233405c3c1942a8bba08total: 39.8s177 noisy roundstimer: 10nsoverhead: 0.37ns/iterAMD Ryzen 9 7950X 16-Core Processor (x86_64/linux)cal: int=0.22ns mem_bw=117.2GiB/s lat=9.9ns

sort_1000

30 rounds · 3436 calls/sample

Each round, all 4 benchmarks run in shuffled order. Each benchmark runs 3436 calls (iterations) per round to produce one sample. After 30 rounds, paired statistics compare the round-by-round differences.

benchmark

min

mean

±mad

95% CI

throughput

reversed

std_sort

421.5ns

443.3ns

±15.7ns

[435.1ns–452.6ns]

2.26 Gitems/s

Summary

n	30
min	421.5ns
mean	443.3ns
median	434.8ns
max	534.8ns
MAD	15.7ns
std dev	24.9ns
CV	5.6% good
cold start	982.0ns
mean CI	[435.1ns – 452.6ns]

sort_unstable

241.3ns

258.3ns

±6.6ns

[-44.4%–-40.8%]

3.87 Gitems/s

Summary

n	30
min	241.3ns
mean	258.3ns
median	248.3ns
max	356.2ns
MAD	6.6ns
std dev	26.4ns
CV	10.2% good
cold start	310.0ns
mean CI	[250.0ns – 268.5ns]

vs Baseline

change	-42.61%
significant	yes — real difference
effect size	d=-11.22 large
Wilcoxon p	0.0000
CI	[-195.1ns – -179.3ns]
drift	r=0.10 stable — no systematic trend
samples	30
outliers removed	3

already sorted

std_sort_sorted

200.1ns

211.1ns

±5.2ns

[-53.5%–-50.9%]

4.74 Gitems/s

Summary

n	30
min	200.1ns
mean	211.1ns
median	204.6ns
max	254.5ns
MAD	5.2ns
std dev	13.9ns
CV	6.6% good
cold start	301.0ns
mean CI	[206.6ns – 216.3ns]

vs Baseline

change	-52.13%
significant	yes — real difference
effect size	d=-15.83 large
Wilcoxon p	0.0000
CI	[-234.8ns – -223.4ns]
drift	r=0.24 stable — no systematic trend
samples	30
outliers removed	3

unstable_sorted

194.6ns

205.6ns

±4.8ns

[-54.5%–-51.8%]

4.86 Gitems/s

Summary

n	30
min	194.6ns
mean	205.6ns
median	199.8ns
max	269.2ns
MAD	4.8ns
std dev	15.3ns
CV	7.4% good
cold start	371.0ns
mean CI	[200.8ns – 211.5ns]

vs Baseline

change	-53.10%
significant	yes — real difference
effect size	d=-14.30 large
Wilcoxon p	0.0000
CI	[-239.0ns – -227.1ns]
drift	r=0.15 stable — no systematic trend
samples	30
outliers removed	2

sub_ns_ops

200 rounds · 1000000 calls/sample

Each round, all 3 benchmarks run in shuffled order. Each benchmark runs 1000000 calls (iterations) per round to produce one sample. After 200 rounds, paired statistics compare the round-by-round differences.

benchmark

min

mean

±mad

95% CI

black_box_unit

0.000ns

0.001ns

±0.000ns

[0.000ns–0.003ns]

Summary

n	200
min	0.000ns
mean	0.001ns
median	0.000ns
max	0.1ns
MAD	0.000ns
std dev	0.0ns
CV	831.9% very noisy — try a quieter system
cold start	100.0ns
mean CI	[0.000ns – 0.003ns]

black_box_add

0.000ns

0.0ns

±0.006ns

[+1407.5%–+2707.6%]

Summary

n	200
min	0.000ns
mean	0.0ns
median	0.004ns
max	0.3ns
MAD	0.006ns
std dev	0.1ns
CV	223.6% very noisy — try a quieter system
cold start	10.0ns
mean CI	[0.0ns – 0.0ns]

vs Baseline

change	+2023.33%
significant	yes — real difference
effect size	d=0.59 medium
Wilcoxon p	0.0000
CI	[0.0ns – 0.0ns]
drift	r=0.36 later rounds slower — possible thermal throttling
samples	200
outliers removed	0

black_box_bool

0.000ns

0.0ns

±0.006ns

[+1241.8%–+2499.4%]

Summary

n	200
min	0.000ns
mean	0.0ns
median	0.004ns
max	0.3ns
MAD	0.006ns
std dev	0.1ns
CV	229.6% very noisy — try a quieter system
cold start	10.0ns
mean CI	[0.0ns – 0.0ns]

vs Baseline

change	+1839.68%
significant	yes — real difference
effect size	d=0.57 medium
Wilcoxon p	0.0000
CI	[0.0ns – 0.0ns]
drift	r=0.27 stable — no systematic trend
samples	200
outliers removed	0

contention

30 rounds · 9523 calls/sample

Each round, all 3 benchmarks run in shuffled order. Each benchmark runs 9523 calls (iterations) per round to produce one sample. After 30 rounds, paired statistics compare the round-by-round differences.

benchmark

min

mean

±mad

95% CI

mutex_map_1t

8.0ns

8.6ns

±0.3ns

[8.4ns–8.8ns]

Summary

n	30
min	8.0ns
mean	8.6ns
median	8.5ns
max	10.9ns
MAD	0.3ns
std dev	0.5ns
CV	5.9% good
cold start	12.45µs
mean CI	[8.4ns – 8.8ns]

Tags

threads

1

mutex_map_2t

51.1ns

96.5ns

±11.4ns

[+989.5%–+1088.4%]

Summary

n	30
min	51.1ns
mean	96.5ns
median	99.5ns
max	114.6ns
MAD	11.4ns
std dev	14.6ns
CV	15.1% noisy
cold start	58.74µs
mean CI	[91.0ns – 101.3ns]

vs Baseline

change	+1040.94%
significant	yes — real difference
effect size	d=10.48 large
Wilcoxon p	0.0000
CI	[85.0ns – 93.5ns]
drift	r=0.30 stable — no systematic trend
samples	30
outliers removed	1

Tags

threads

2

mutex_map_4t

148.7ns

186.6ns

±16.7ns

[+1967.4%–+2117.9%]

Summary

n	30
min	148.7ns
mean	186.6ns
median	183.1ns
max	257.4ns
MAD	16.7ns
std dev	22.0ns
CV	11.8% good
cold start	147.16µs
mean CI	[179.0ns – 194.8ns]

vs Baseline

change	+2042.91%
significant	yes — real difference
effect size	d=13.93 large
Wilcoxon p	0.0000
CI	[169.1ns – 182.0ns]
drift	r=-0.05 stable — no systematic trend
samples	30
outliers removed	1

Tags

threads

4

process_elements

30 rounds · 122 calls/sample

Each round, all 3 benchmarks run in shuffled order. Each benchmark runs 122 calls (iterations) per round to produce one sample. After 30 rounds, paired statistics compare the round-by-round differences.

benchmark

min

mean

±mad

95% CI

throughput

single-threaded

sum_sequential

947.0ns

985.9ns

±11.0ns

[962.6ns–1.02µs]

10.14 Gitems/s

Summary

n	30
min	947.0ns
mean	985.9ns
median	958.0ns
max	1.29µs
MAD	11.0ns
std dev	77.4ns
CV	7.8% good
cold start	581.0ns
mean CI	[962.6ns – 1.02µs]

map_sqrt

7.91µs

8.19µs

±135.1ns

[+709.1%–+717.8%]

1.22 Gitems/s

Summary

n	30
min	7.91µs
mean	8.19µs
median	8.05µs
max	10.16µs
MAD	135.1ns
std dev	441.3ns
CV	5.4% good
cold start	9.14µs
mean CI	[8.06µs – 8.36µs]

vs Baseline

change	+713.46%
significant	yes — real difference
effect size	d=80.29 large
Wilcoxon p	0.0000
CI	[7.02µs – 7.10µs]
drift	r=0.30 stable — no systematic trend
samples	30
outliers removed	4

contended (4 threads)

shared_vec_push

281.76µs

340.15µs

±39.99µs

[+33099.6%–+35740.9%]

29.40 Mitems/s

Summary

n	30
min	281.76µs
mean	340.15µs
median	334.76µs
max	424.82µs
MAD	39.99µs
std dev	36.90µs
CV	10.8% good
cold start	312.58µs
mean CI	[327.34µs – 353.38µs]

vs Baseline

change	+34400.03%
significant	yes — real difference
effect size	d=13.00 large
Wilcoxon p	0.0000
CI	[326.34µs – 352.38µs]
drift	r=0.19 stable — no systematic trend
samples	30
outliers removed	0

Tags

threads

4

scaling

200 rounds · 69 calls/sample

Each round, all 17 benchmarks run in shuffled order. Each benchmark runs 69 calls (iterations) per round to produce one sample. After 200 rounds, paired statistics compare the round-by-round differences.

benchmark

min

mean

±mad

95% CI

throughput

sqrt_1t

2.97µs

4.30µs

±114.9ns

[4.25µs–4.35µs]

2.33 Gitems/s

Summary

n	200
min	2.97µs
mean	4.30µs
median	4.20µs
max	6.94µs
MAD	114.9ns
std dev	383.4ns
CV	8.9% good
cold start	12.53µs
mean CI	[4.25µs – 4.35µs]

Tags

threads

1

sqrt_2t

4.17µs

4.81µs

±337.4ns

[+9.4%–+11.8%]

2.08 Gitems/s

Summary

n	200
min	4.17µs
mean	4.81µs
median	4.70µs
max	8.42µs
MAD	337.4ns
std dev	554.5ns
CV	11.5% good
cold start	204.33µs
mean CI	[4.74µs – 4.89µs]

vs Baseline

change	+10.62%
significant	yes — real difference
effect size	d=1.79 large
Wilcoxon p	0.0000
CI	[402.3ns – 502.5ns]
drift	r=0.05 stable — no systematic trend
samples	200
outliers removed	16

Tags

threads

2

sqrt_3t

4.67µs

5.72µs

±338.9ns

[+30.4%–+33.0%]

1.75 Gitems/s

Summary

n	200
min	4.67µs
mean	5.72µs
median	5.58µs
max	9.01µs
MAD	338.9ns
std dev	626.8ns
CV	11.0% good
cold start	110.51µs
mean CI	[5.64µs – 5.81µs]

vs Baseline

change	+31.69%
significant	yes — real difference
effect size	d=4.70 large
Wilcoxon p	0.0000
CI	[1.29µs – 1.40µs]
drift	r=0.06 stable — no systematic trend
samples	200
outliers removed	16

Tags

threads

3

sqrt_4t

4.93µs

6.20µs

±424.6ns

[+39.6%–+42.7%]

1.61 Gitems/s

Summary

n	200
min	4.93µs
mean	6.20µs
median	6.02µs
max	15.56µs
MAD	424.6ns
std dev	967.8ns
CV	15.6% noisy
cold start	139.40µs
mean CI	[6.08µs – 6.34µs]

vs Baseline

change	+41.17%
significant	yes — real difference
effect size	d=5.35 large
Wilcoxon p	0.0000
CI	[1.69µs – 1.82µs]
drift	r=-0.00 stable — no systematic trend
samples	200
outliers removed	19

Tags

threads

4

sqrt_5t

5.27µs

6.87µs

±525.7ns

[+54.6%–+58.2%]

1.46 Gitems/s

Summary

n	200
min	5.27µs
mean	6.87µs
median	6.64µs
max	10.79µs
MAD	525.7ns
std dev	908.0ns
CV	13.2% good
cold start	189.74µs
mean CI	[6.75µs – 7.00µs]

vs Baseline

change	+56.39%
significant	yes — real difference
effect size	d=5.92 large
Wilcoxon p	0.0000
CI	[2.33µs – 2.48µs]
drift	r=0.09 stable — no systematic trend
samples	200
outliers removed	20

Tags

threads

5

sqrt_6t

5.14µs

7.49µs

±615.5ns

[+69.3%–+73.9%]

1.33 Gitems/s

Summary

n	200
min	5.14µs
mean	7.49µs
median	7.30µs
max	11.71µs
MAD	615.5ns
std dev	1.08µs
CV	14.4% good
cold start	226.56µs
mean CI	[7.35µs – 7.64µs]

vs Baseline

change	+71.59%
significant	yes — real difference
effect size	d=6.06 large
Wilcoxon p	0.0000
CI	[2.96µs – 3.15µs]
drift	r=0.17 stable — no systematic trend
samples	200
outliers removed	26

Tags

threads

6

sqrt_7t

5.52µs

8.27µs

±889.6ns

[+85.5%–+91.9%]

1.21 Gitems/s

Summary

n	200
min	5.52µs
mean	8.27µs
median	8.04µs
max	12.42µs
MAD	889.6ns
std dev	1.28µs
CV	15.5% noisy
cold start	248.01µs
mean CI	[8.10µs – 8.45µs]

vs Baseline

change	+88.72%
significant	yes — real difference
effect size	d=5.32 large
Wilcoxon p	0.0000
CI	[3.66µs – 3.94µs]
drift	r=0.07 stable — no systematic trend
samples	200
outliers removed	17

Tags

threads

7

sqrt_8t

4.57µs

9.13µs

±1.12µs

[+107.6%–+115.7%]

1.10 Gitems/s

Summary

n	200
min	4.57µs
mean	9.13µs
median	8.93µs
max	14.08µs
MAD	1.12µs
std dev	1.44µs
CV	15.8% noisy
cold start	292.49µs
mean CI	[8.93µs – 9.33µs]

vs Baseline

change	+111.65%
significant	yes — real difference
effect size	d=5.44 large
Wilcoxon p	0.0000
CI	[4.62µs – 4.97µs]
drift	r=0.02 stable — no systematic trend
samples	200
outliers removed	9

Tags

threads

8

sqrt_9t

5.82µs

10.08µs

±1.47µs

[+125.4%–+136.0%]

992.33 Mitems/s

Summary

n	200
min	5.82µs
mean	10.08µs
median	9.91µs
max	16.45µs
MAD	1.47µs
std dev	1.83µs
CV	18.2% noisy
cold start	311.60µs
mean CI	[9.83µs – 10.33µs]

vs Baseline

change	+130.68%
significant	yes — real difference
effect size	d=4.81 large
Wilcoxon p	0.0000
CI	[5.39µs – 5.85µs]
drift	r=0.02 stable — no systematic trend
samples	200
outliers removed	6

Tags

threads

9

sqrt_10t

6.03µs

11.05µs

±1.80µs

[+148.5%–+160.8%]

905.35 Mitems/s

Summary

n	200
min	6.03µs
mean	11.05µs
median	10.86µs
max	18.28µs
MAD	1.80µs
std dev	2.02µs
CV	18.3% noisy
cold start	372.77µs
mean CI	[10.77µs – 11.33µs]

vs Baseline

change	+154.66%
significant	yes — real difference
effect size	d=4.91 large
Wilcoxon p	0.0000
CI	[6.39µs – 6.92µs]
drift	r=0.01 stable — no systematic trend
samples	200
outliers removed	3

Tags

threads

10

sqrt_11t

6.47µs

11.98µs

±2.29µs

[+171.8%–+184.2%]

834.74 Mitems/s

Summary

n	200
min	6.47µs
mean	11.98µs
median	11.81µs
max	18.77µs
MAD	2.29µs
std dev	1.95µs
CV	16.3% noisy
cold start	371.32µs
mean CI	[11.71µs – 12.25µs]

vs Baseline

change	+178.03%
significant	yes — real difference
effect size	d=5.58 large
Wilcoxon p	0.0000
CI	[7.38µs – 7.91µs]
drift	r=0.05 stable — no systematic trend
samples	200
outliers removed	1

Tags

threads

11

sqrt_12t

6.60µs

12.64µs

±2.15µs

[+186.3%–+199.5%]

791.26 Mitems/s

Summary

n	200
min	6.60µs
mean	12.64µs
median	12.23µs
max	21.14µs
MAD	2.15µs
std dev	2.20µs
CV	17.4% noisy
cold start	395.67µs
mean CI	[12.33µs – 12.94µs]

vs Baseline

change	+192.98%
significant	yes — real difference
effect size	d=5.99 large
Wilcoxon p	0.0000
CI	[8.00µs – 8.57µs]
drift	r=-0.04 stable — no systematic trend
samples	200
outliers removed	5

Tags

threads

12

sqrt_13t

6.82µs

13.19µs

±2.11µs

[+199.9%–+212.4%]

758.01 Mitems/s

Summary

n	200
min	6.82µs
mean	13.19µs
median	12.94µs
max	20.54µs
MAD	2.11µs
std dev	2.05µs
CV	15.6% noisy
cold start	466.13µs
mean CI	[12.91µs – 13.48µs]

vs Baseline

change	+206.16%
significant	yes — real difference
effect size	d=6.48 large
Wilcoxon p	0.0000
CI	[8.59µs – 9.13µs]
drift	r=-0.08 stable — no systematic trend
samples	200
outliers removed	3

Tags

threads

13

sqrt_14t

7.21µs

13.61µs

±2.06µs

[+210.4%–+223.2%]

734.72 Mitems/s

Summary

n	200
min	7.21µs
mean	13.61µs
median	13.55µs
max	19.75µs
MAD	2.06µs
std dev	2.02µs
CV	14.8% good
cold start	517.29µs
mean CI	[13.33µs – 13.89µs]

vs Baseline

change	+216.86%
significant	yes — real difference
effect size	d=6.70 large
Wilcoxon p	0.0000
CI	[9.04µs – 9.59µs]
drift	r=0.01 stable — no systematic trend
samples	200
outliers removed	2

Tags

threads

14

sqrt_15t

8.16µs

14.65µs

±1.93µs

[+234.3%–+246.2%]

682.73 Mitems/s

Summary

n	200
min	8.16µs
mean	14.65µs
median	14.54µs
max	22.65µs
MAD	1.93µs
std dev	2.03µs
CV	13.9% good
cold start	578.32µs
mean CI	[14.37µs – 14.93µs]

vs Baseline

change	+240.15%
significant	yes — real difference
effect size	d=8.16 large
Wilcoxon p	0.0000
CI	[10.03µs – 10.54µs]
drift	r=-0.01 stable — no systematic trend
samples	200
outliers removed	6

Tags

threads

15

sqrt_16t

8.83µs

14.85µs

±1.96µs

[+240.0%–+253.5%]

673.53 Mitems/s

Summary

n	200
min	8.83µs
mean	14.85µs
median	14.72µs
max	20.49µs
MAD	1.96µs
std dev	2.20µs
CV	14.8% good
cold start	667.51µs
mean CI	[14.55µs – 15.16µs]

vs Baseline

change	+246.61%
significant	yes — real difference
effect size	d=7.35 large
Wilcoxon p	0.0000
CI	[10.33µs – 10.91µs]
drift	r=-0.04 stable — no systematic trend
samples	200
outliers removed	6

Tags

threads

16

sqrt_32t

14.63µs

24.81µs

±2.67µs

[+469.3%–+487.8%]

403.06 Mitems/s

Summary

n	200
min	14.63µs
mean	24.81µs
median	24.70µs
max	32.36µs
MAD	2.67µs
std dev	2.98µs
CV	12.0% good
cold start	1.07ms
mean CI	[24.39µs – 25.22µs]

vs Baseline

change	+478.71%
significant	yes — real difference
effect size	d=10.29 large
Wilcoxon p	0.0000
CI	[20.17µs – 20.96µs]
drift	r=-0.06 stable — no systematic trend
samples	200
outliers removed	3

Tags

threads

32

sort_sizes

30 rounds · 4016 calls/sample

Each round, all 4 benchmarks run in shuffled order. Each benchmark runs 4016 calls (iterations) per round to produce one sample. After 30 rounds, paired statistics compare the round-by-round differences.

benchmark

min

mean

±mad

95% CI

unstable_10

21.6ns

22.8ns

±0.3ns

[22.0ns–23.8ns]

Summary

n	30
min	21.6ns
mean	22.8ns
median	21.9ns
max	35.0ns
MAD	0.3ns
std dev	2.6ns
CV	11.6% good
cold start	90.0ns
mean CI	[22.0ns – 23.8ns]

unstable_100

31.0ns

32.5ns

±0.4ns

[+41.7%–+43.9%]

Summary

n	30
min	31.0ns
mean	32.5ns
median	31.5ns
max	55.3ns
MAD	0.4ns
std dev	4.4ns
CV	13.4% good
cold start	91.0ns
mean CI	[31.5ns – 34.2ns]

vs Baseline

change	+42.80%
significant	yes — real difference
effect size	d=18.25 large
Wilcoxon p	0.0000
CI	[9.2ns – 9.7ns]
drift	r=0.10 stable — no systematic trend
samples	30
outliers removed	6

unstable_1000

243.7ns

259.7ns

±5.4ns

[+1010.0%–+1031.3%]

Summary

n	30
min	243.7ns
mean	259.7ns
median	249.5ns
max	466.6ns
MAD	5.4ns
std dev	40.1ns
CV	15.5% noisy
cold start	261.0ns
mean CI	[250.2ns – 276.0ns]

vs Baseline

change	+1019.81%
significant	yes — real difference
effect size	d=51.59 large
Wilcoxon p	0.0000
CI	[225.6ns – 230.4ns]
drift	r=-0.12 stable — no systematic trend
samples	30
outliers removed	4

unstable_10000

2.36µs

2.50µs

±85.4ns

[+10723.0%–+10922.8%]

Summary

n	30
min	2.36µs
mean	2.50µs
median	2.44µs
max	4.21µs
MAD	85.4ns
std dev	330.0ns
CV	13.2% good
cold start	4.66µs
mean CI	[2.42µs – 2.63µs]

vs Baseline

change	+10822.87%
significant	yes — real difference
effect size	d=54.36 large
Wilcoxon p	0.0000
CI	[2.39µs – 2.44µs]
drift	r=-0.16 stable — no systematic trend
samples	30
outliers removed	1