optimize Array.of and Array.from for cpu pipelining and branch prediction

[evanwashere]

passes 262 webkit tests (Tools/Scripts/test262-runner --release, build first)
todo: check cpu cycles and instructions count (needs bun build for @mitata/counters)

patch: https://gist.github.com/evanwashere/17778413aba8b1eec1adfe5c6ef04a69

before

clk: ~3.33 GHz cpu: null runtime: jsc (null) benchmark avg (min … max) p75 p99 (min … top 1%) ------------------------------------------------------- ------------------------------- Array.of(512) 1.63 µs/iter 1.67 µs 1.68 µs ▃▆▇▁▁▂▁▂▄█▂ Array.of(4096) 10.47 µs/iter 10.49 µs 10.50 µs ▃▃▁▆▁▆▆▃█▆▆ Array.of(32768) 88.37 µs/iter 93.33 µs 110.25 µs ▂▅█▃▂▁▁▁▄▁▁ Array.of(65536) 155.96 µs/iter 157.58 µs 174.00 µs ▂██▇▄▃▂▁▁▁▁ ------------------------------------------------------- ------------------------------- Array.from(512) 1.02 µs/iter 1.02 µs 1.32 µs ▃█▂▁▁▁▁▁▁▁▁ Array.from(4096) 6.75 µs/iter 6.77 µs 6.81 µs ▂▅█▇▄▇▄▂▅▄▂ Array.from(32768) 53.83 µs/iter 53.92 µs 54.25 µs ▃▁▃█▆▃▃▁▁▃▃ Array.from(65536) 110.81 µs/iter 112.46 µs 124.75 µs ▂▆█▇▇▃▂▁▁▁▁ ------------------------------------------------------- ------------------------------- Array.from(arr 512) 200.16 ns/iter 202.21 ns 221.45 ns ▁▄█▆▄▂▂▁▁▁▁ Array.from(arr 4096) 764.40 ns/iter 763.08 ns 989.76 ns ▂█▃▁▁▁▁▁▁▁▁ Array.from(arr 32768) 6.12 µs/iter 6.07 µs 6.66 µs █▆▂▁▁▁▁▁▁▁▂ Array.from(arr 65536) 11.80 µs/iter 11.87 µs 11.87 µs ▂▁▁▁▁▁▁▁▁▄█ ------------------------------------------------------- ------------------------------- Array.from(512, () => 1) 2.56 µs/iter 2.56 µs 2.58 µs ▂▃▄▅▄█▃▃▃▂▂ Array.from(4096, () => 1) 14.27 µs/iter 14.30 µs 14.34 µs ▃▁▁▃▆▁▁█▆▃▃ Array.from(32768, () => 1) 118.04 µs/iter 119.67 µs 132.92 µs ▁█▅▆▄▂▂▁▁▁▁ Array.from(65536, () => 1) 229.38 µs/iter 231.33 µs 246.58 µs ▁▂▆█▆▄▂▂▁▁▁ ------------------------------------------------------- ------------------------------- Array.from(iterator 512) 2.21 µs/iter 2.22 µs 2.27 µs ▃▆█▆▂▁▁▁▁▁▁ Array.from(iterator 4096) 16.04 µs/iter 16.04 µs 16.06 µs █▅▁▅▅▁█▅▁▁▅ Array.from(iterator 32768) 140.09 µs/iter 141.79 µs 161.83 µs ▁▁▂▂█▃▂▁▁▁▁ Array.from(iterator 65536) 262.21 µs/iter 263.79 µs 310.17 µs ▁▁█▇▃▂▁▁▁▁▁ ------------------------------------------------------- ------------------------------- Array.from(iterator 512, () => 1) 2.58 µs/iter 2.58 µs 2.81 µs ▂█▂▁▁▁▁▁▁▁▁ Array.from(iterator 4096, () => 1) 18.93 µs/iter 18.90 µs 18.92 µs ▃▃▆▆▁▁▃▁█▁▃ Array.from(iterator 32768, () => 1) 164.04 µs/iter 165.50 µs 178.83 µs ▁▁▁▄█▅▃▂▁▁▁ Array.from(iterator 65536, () => 1) 305.42 µs/iter 307.25 µs 326.67 µs ▁▃██▅▂▂▂▁▁▁ ------------------------------------------------------- ------------------------------- Array.fromAsync(512) 26.16 µs/iter 26.08 µs 26.34 µs ▃▆█▆▃▁▁▁▃▁▃ Array.fromAsync(4096) 214.39 µs/iter 216.50 µs 231.75 µs ▂▂▆█▂▄▂▁▁▁▁ Array.fromAsync(32768) 1.64 ms/iter 1.65 ms 1.68 ms ▁▅▇▅▃▆█▅▃▂▁ Array.fromAsync(65536) 3.26 ms/iter 3.28 ms 3.30 ms ▂▅▆▄▃▂▂▄█▄▂ ------------------------------------------------------- ------------------------------- Array.fromAsync(arr 512) 81.31 µs/iter 82.42 µs 100.88 µs ▁█▅▄▂▂▁▁▁▁▁ Array.fromAsync(arr 4096) 559.67 µs/iter 564.25 µs 591.58 µs ▁▅█▆▅▃▃▂▁▁▁ Array.fromAsync(arr 32768) 4.20 ms/iter 4.21 ms 4.30 ms ▁▄██▅▂▂▁▁▁▁ Array.fromAsync(arr 65536) 8.35 ms/iter 8.37 ms 8.49 ms ▄▅██▆▃▁▂▂▁▂ ------------------------------------------------------- ------------------------------- Array.fromAsync(512, () => 1) 56.75 µs/iter 57.21 µs 69.75 µs ▂█▂▂▂▁▁▁▁▁▁ Array.fromAsync(4096, () => 1) 417.08 µs/iter 420.08 µs 438.79 µs ▁▃██▆▄▃▂▂▁▁ Array.fromAsync(32768, () => 1) 3.24 ms/iter 3.24 ms 3.32 ms ▃▆▇█▄▂▁▁▁▁▁ Array.fromAsync(65536, () => 1) 6.45 ms/iter 6.47 ms 6.49 ms ▂▄█▄▄▄██▆▄▃ ------------------------------------------------------- ------------------------------- Array.fromAsync(iterator 512) 83.20 µs/iter 84.29 µs 97.50 µs ▁▁█▆▃▂▂▁▁▁▁ Array.fromAsync(iterator 4096) 571.32 µs/iter 575.21 µs 599.50 µs ▁▁▄█▆▄▃▂▂▁▁ Array.fromAsync(iterator 32768) 4.28 ms/iter 4.29 ms 4.34 ms ▂▄▆█▇▇▃▂▂▁▁ Array.fromAsync(iterator 65536) 8.53 ms/iter 8.55 ms 8.75 ms ▃▄█▅▂▂▂▁▁▁▁ ------------------------------------------------------- ------------------------------- Array.fromAsync(iterator 512, () => 1) 105.42 µs/iter 106.00 µs 121.00 µs ▁▁█▄▂▂▁▁▁▁▁ Array.fromAsync(iterator 4096, () => 1) 747.20 µs/iter 752.13 µs 778.75 µs ▁▁▁▃█▄▃▂▂▁▁ Array.fromAsync(iterator 32768, () => 1) 5.70 ms/iter 5.72 ms 5.79 ms ▁▃▅▆▇█▄▃▃▂▂ Array.fromAsync(iterator 65536, () => 1) 11.38 ms/iter 11.40 ms 11.65 ms ▃▃█▆▅▂▂▁▂▁▂

after

clk: ~3.41 GHz cpu: null runtime: jsc (null) benchmark avg (min … max) p75 p99 (min … top 1%) ------------------------------------------------------- ------------------------------- Array.of(512) 1.45 µs/iter 1.48 µs 1.51 µs ▂▅█▃▂▁▂▇▇▂▁ Array.of(4096) 9.15 µs/iter 9.17 µs 9.18 µs ▅▁▃▃▆▃▅▅▅█▃ Array.of(32768) 71.90 µs/iter 71.94 µs 72.43 µs ▃▆▃▁▆█▁▁▃▁▃ Array.of(65536) 138.11 µs/iter 139.79 µs 154.50 µs ▁▁▂▇█▄▃▂▁▁▁ ------------------------------------------------------- ------------------------------- Array.from(512) 393.27 ns/iter 395.49 ns 479.08 ns ▃█▃▂▁▁▁▁▁▁▁ Array.from(4096) 1.54 µs/iter 1.55 µs 1.59 µs ▁▃▆█▇▅▃▁▁▁▁ Array.from(32768) 12.18 µs/iter 12.20 µs 12.39 µs ▂█▅▂▂▁▁▁▂▁▄ Array.from(65536) 24.60 µs/iter 24.61 µs 24.66 µs ▃▁▁▃▆▃▁█▆▁▃ ------------------------------------------------------- ------------------------------- Array.from(arr 512) 192.91 ns/iter 194.31 ns 224.31 ns ▁▆█▄▂▁▁▁▁▁▁ Array.from(arr 4096) 765.31 ns/iter 755.46 ns 1.02 µs ▂█▂▁▁▁▁▁▁▁▁ Array.from(arr 32768) 5.59 µs/iter 5.82 µs 5.89 µs ▄▁▁▁▁▁▁▁▁█▂ Array.from(arr 65536) 11.59 µs/iter 11.73 µs 11.74 µs ▂▁▃▁▁▁▁▁▂▄█ ------------------------------------------------------- ------------------------------- Array.from(512, () => 1) 411.13 ns/iter 412.54 ns 461.70 ns ▂█▆▃▁▁▁▁▁▁▁ Array.from(4096, () => 1) 2.74 µs/iter 2.74 µs 2.85 µs ▄█▆▂▁▁▁▂▁▁▁ Array.from(32768, () => 1) 22.48 µs/iter 22.54 µs 22.64 µs ▃█▃▁▃▁▁▅▁▃▃ Array.from(65536, () => 1) 45.53 µs/iter 45.58 µs 45.78 µs ▅█▅▅█▅█▁▁▁▅ ------------------------------------------------------- ------------------------------- Array.from(iterator 512) 2.13 µs/iter 2.13 µs 2.34 µs ▄█▂▁▁▁▁▁▁▁▁ Array.from(iterator 4096) 15.42 µs/iter 15.43 µs 15.45 µs ▅▅▅█▁▅▅▅▁▁█ Array.from(iterator 32768) 140.76 µs/iter 142.79 µs 158.58 µs ▁▁▂▁▂█▅▂▂▁▁ Array.from(iterator 65536) 257.83 µs/iter 260.17 µs 279.75 µs ▁▁▁▃█▅▃▂▂▁▁ ------------------------------------------------------- ------------------------------- Array.from(iterator 512, () => 1) 2.06 µs/iter 2.06 µs 2.12 µs ▂▄█▄▃▁▁▁▁▁▁ Array.from(iterator 4096, () => 1) 14.86 µs/iter 14.88 µs 14.89 µs ▆▃▁▃█▁▁▁▁▃▆ Array.from(iterator 32768, () => 1) 133.87 µs/iter 136.42 µs 149.58 µs ▁▁▁▂▃▆█▄▂▁▁ Array.from(iterator 65536, () => 1) 246.52 µs/iter 248.83 µs 266.54 µs ▁▁▁▂▆█▄▂▁▁▁ ------------------------------------------------------- ------------------------------- Array.fromAsync(512) 579.35 ns/iter 581.24 ns 670.80 ns ▃█▄▂▁▁▁▁▁▁▁ Array.fromAsync(4096) 1.88 µs/iter 1.88 µs 1.94 µs ▁▁▆█▇▃▁▂▁▁▁ Array.fromAsync(32768) 13.24 µs/iter 13.25 µs 13.36 µs ▅▆▃█▁▃▁▁▁▁▃ Array.fromAsync(65536) 26.89 µs/iter 26.84 µs 26.93 µs ▃▁▃█▃▃▆▃▁▁▃ ------------------------------------------------------- ------------------------------- Array.fromAsync(arr 512) 77.46 µs/iter 78.71 µs 93.96 µs ▁▁█▄▃▂▁▁▁▁▁ Array.fromAsync(arr 4096) 525.49 µs/iter 530.42 µs 554.63 µs ▁▁▇█▆▄▄▂▂▁▁ Array.fromAsync(arr 32768) 3.89 ms/iter 3.90 ms 3.99 ms ▂▆█▅▃▂▂▂▂▁▁ Array.fromAsync(arr 65536) 7.81 ms/iter 7.87 ms 7.98 ms ▃█▄▅▆▆▅▅▄▂▂ ------------------------------------------------------- ------------------------------- Array.fromAsync(512, () => 1) 26.76 µs/iter 26.83 µs 27.03 µs ▃▁█▁▃▆▃▆▁▁▃ Array.fromAsync(4096, () => 1) 212.37 µs/iter 215.58 µs 229.96 µs ▁▂█▂▃▂▁▁▁▁▁ Array.fromAsync(32768, () => 1) 1.62 ms/iter 1.63 ms 1.66 ms ▂▃▆██▆▄▃▁▁▁ Array.fromAsync(65536, () => 1) 3.24 ms/iter 3.25 ms 3.41 ms ▃█▅▃▂▂▁▁▁▁▁ ------------------------------------------------------- ------------------------------- Array.fromAsync(iterator 512) 79.03 µs/iter 80.29 µs 92.04 µs ▁▁▆█▄▂▂▂▁▁▁ Array.fromAsync(iterator 4096) 538.00 µs/iter 542.71 µs 571.71 µs ▁▂▄█▇▄▃▂▂▁▁ Array.fromAsync(iterator 32768) 4.01 ms/iter 4.02 ms 4.10 ms ▂▆█▇▄▂▂▂▁▂▁ Array.fromAsync(iterator 65536) 8.05 ms/iter 8.11 ms 8.35 ms ▆█▆▅▇▇▃▃▃▂▁ ------------------------------------------------------- ------------------------------- Array.fromAsync(iterator 512, () => 1) 106.25 µs/iter 107.38 µs 123.33 µs ▁▁▆█▃▂▂▁▁▁▁ Array.fromAsync(iterator 4096, () => 1) 745.55 µs/iter 751.38 µs 794.75 µs ▁▁▃█▅▄▂▂▁▁▁ Array.fromAsync(iterator 32768, () => 1) 5.71 ms/iter 5.73 ms 5.83 ms ▂▄█▇▆▃▃▂▂▂▂ Array.fromAsync(iterator 65536, () => 1) 11.51 ms/iter 11.57 ms 11.75 ms ▂▅▄▅▄█▆▄▃▁▂

[evanwashere]

@Constellation what do you think of having this upstreamed