No usize except uniform

[dhardy]

• Added a CHANGELOG.md entry

Summary

This is #1471 without Rng::gen_index. Since we can't have UniformUsize without directly supporting usize in Rng::gen_range this is probably the best way to go.

[dhardy]

At this point, we can remove the internal rand::seq::gen_index function without affecting tests. Benchmarking this changeset shows some changes, but mostly noise.

$ cargo bench "seq|choose|shuffle|indices"
   Compiling rand v0.9.0-alpha.1 (/home/dhardy/projects/rand/rand)
   Compiling rand_distr v0.5.0-alpha.1 (/home/dhardy/projects/rand/rand/rand_distr)
   Compiling benches v0.1.0 (/home/dhardy/projects/rand/rand/benches)
    Finished `bench` profile [optimized] target(s) in 5.22s
     Running benches/base_distributions.rs (target/release/deps/base_distributions-4d7c8319c6c903ac)
running 0 tests
test result: ok. 0 passed; 0 failed; 0 ignored; 0 measured; 130 filtered out; finished in 0.00s
 Running src/distr.rs (target/release/deps/distr-0435b5737d5e6436)
 Running benches/generators.rs (target/release/deps/generators-16c771dd48c7abd1)

running 0 tests
test result: ok. 0 passed; 0 failed; 0 ignored; 0 measured; 47 filtered out; finished in 0.00s
 Running benches/misc.rs (target/release/deps/misc-5858d7fc1cad7292)

running 0 tests
test result: ok. 0 passed; 0 failed; 0 ignored; 0 measured; 14 filtered out; finished in 0.00s
 Running benches/seq.rs (target/release/deps/seq-ad06a4eecb35fde5)

running 0 tests
test result: ok. 0 passed; 0 failed; 0 ignored; 0 measured; 24 filtered out; finished in 0.00s
 Running src/seq_choose.rs (target/release/deps/seq_choose-07855b064554e80a)

choose_size-hinted_from_1_ChaCha20

time:   [420.20 ps 421.09 ps 422.49 ps]

change: [-80.420% -80.349% -80.265%] (p = 0.00 < 0.05)

Performance has improved.

Found 1 outliers among 100 measurements (1.00%)

1 (1.00%) high severe
choose_stable_from_1_ChaCha20

time:   [5.3811 ns 5.4025 ns 5.4255 ns]

change: [-2.9571% -2.0250% -1.0752%] (p = 0.00 < 0.05)

Performance has improved.

Found 6 outliers among 100 measurements (6.00%)

5 (5.00%) high mild

1 (1.00%) high severe
choose_unhinted_from_1_ChaCha20

time:   [4.1099 ns 4.1119 ns 4.1142 ns]

change: [+0.1323% +0.2572% +0.3843%] (p = 0.00 < 0.05)

Change within noise threshold.

Found 9 outliers among 100 measurements (9.00%)

3 (3.00%) low severe

1 (1.00%) low mild

4 (4.00%) high mild

1 (1.00%) high severe
choose_windowed_from_1_ChaCha20

time:   [5.2392 ns 5.2492 ns 5.2601 ns]

change: [-3.6257% -2.8153% -1.8643%] (p = 0.00 < 0.05)

Performance has improved.

Found 17 outliers among 100 measurements (17.00%)

2 (2.00%) high mild

15 (15.00%) high severe
choose_size-hinted_from_2_ChaCha20

time:   [4.2299 ns 4.2552 ns 4.2882 ns]

change: [-5.9766% -5.5698% -5.1654%] (p = 0.00 < 0.05)

Performance has improved.

Found 20 outliers among 100 measurements (20.00%)

4 (4.00%) high mild

16 (16.00%) high severe
choose_stable_from_2_ChaCha20

time:   [13.357 ns 13.380 ns 13.409 ns]

change: [+1.5822% +1.8502% +2.0972%] (p = 0.00 < 0.05)

Performance has regressed.

Found 1 outliers among 100 measurements (1.00%)

1 (1.00%) high mild
choose_unhinted_from_2_ChaCha20

time:   [11.450 ns 11.511 ns 11.579 ns]

change: [+0.3157% +1.0968% +1.7696%] (p = 0.00 < 0.05)

Change within noise threshold.

Found 9 outliers among 100 measurements (9.00%)

7 (7.00%) high mild

2 (2.00%) high severe
choose_windowed_from_2_ChaCha20

time:   [10.896 ns 10.902 ns 10.908 ns]

change: [+15.622% +15.756% +15.875%] (p = 0.00 < 0.05)

Performance has regressed.

Found 4 outliers among 100 measurements (4.00%)

2 (2.00%) low mild

1 (1.00%) high mild

1 (1.00%) high severe
choose_size-hinted_from_3_ChaCha20

time:   [4.1573 ns 4.1711 ns 4.1853 ns]

change: [-6.9772% -6.8399% -6.6869%] (p = 0.00 < 0.05)

Performance has improved.

Found 8 outliers among 100 measurements (8.00%)

1 (1.00%) high mild

7 (7.00%) high severe
choose_stable_from_3_ChaCha20

time:   [24.523 ns 24.573 ns 24.633 ns]

change: [+5.0714% +5.5322% +6.0273%] (p = 0.00 < 0.05)

Performance has regressed.

Found 4 outliers among 100 measurements (4.00%)

2 (2.00%) high mild

2 (2.00%) high severe
choose_unhinted_from_3_ChaCha20

time:   [21.728 ns 21.762 ns 21.798 ns]

change: [+0.9964% +1.1999% +1.3999%] (p = 0.00 < 0.05)

Change within noise threshold.

Found 12 outliers among 100 measurements (12.00%)

4 (4.00%) high mild

8 (8.00%) high severe
choose_windowed_from_3_ChaCha20

time:   [12.184 ns 12.219 ns 12.267 ns]

change: [+12.798% +13.298% +13.657%] (p = 0.00 < 0.05)

Performance has regressed.

Found 11 outliers among 100 measurements (11.00%)

1 (1.00%) low mild

3 (3.00%) high mild

7 (7.00%) high severe
choose_size-hinted_from_10_ChaCha20

time:   [4.2229 ns 4.2244 ns 4.2260 ns]

change: [-5.7623% -5.5768% -5.4285%] (p = 0.00 < 0.05)

Performance has improved.

Found 14 outliers among 100 measurements (14.00%)

2 (2.00%) low mild

9 (9.00%) high mild

3 (3.00%) high severe
choose_stable_from_10_ChaCha20

time:   [61.407 ns 61.687 ns 61.949 ns]

change: [+1.7071% +1.9575% +2.2228%] (p = 0.00 < 0.05)

Performance has regressed.

Found 18 outliers among 100 measurements (18.00%)

1 (1.00%) high mild

17 (17.00%) high severe
choose_unhinted_from_10_ChaCha20

time:   [55.321 ns 55.364 ns 55.436 ns]

change: [+0.1581% +0.2399% +0.3271%] (p = 0.00 < 0.05)

Change within noise threshold.

Found 2 outliers among 100 measurements (2.00%)

1 (1.00%) high mild

1 (1.00%) high severe
choose_windowed_from_10_ChaCha20

time:   [21.895 ns 21.927 ns 21.964 ns]

change: [+9.9881% +10.193% +10.405%] (p = 0.00 < 0.05)

Performance has regressed.

Found 10 outliers among 100 measurements (10.00%)

5 (5.00%) high mild

5 (5.00%) high severe
choose_size-hinted_from_100_ChaCha20

time:   [4.2295 ns 4.2304 ns 4.2312 ns]

change: [-5.6818% -5.4132% -5.1829%] (p = 0.00 < 0.05)

Performance has improved.

Found 2 outliers among 100 measurements (2.00%)

2 (2.00%) high severe
choose_stable_from_100_ChaCha20

time:   [377.31 ns 378.45 ns 379.61 ns]

change: [+2.1705% +2.4440% +2.6774%] (p = 0.00 < 0.05)

Performance has regressed.

Found 5 outliers among 100 measurements (5.00%)

5 (5.00%) high mild
choose_unhinted_from_100_ChaCha20

time:   [334.08 ns 334.41 ns 334.86 ns]

change: [-0.2562% +0.8827% +1.8004%] (p = 0.10 > 0.05)

No change in performance detected.

Found 4 outliers among 100 measurements (4.00%)

1 (1.00%) high mild

3 (3.00%) high severe
choose_windowed_from_100_ChaCha20

time:   [125.49 ns 125.66 ns 125.85 ns]

change: [+23.492% +23.829% +24.238%] (p = 0.00 < 0.05)

Performance has regressed.

Found 7 outliers among 100 measurements (7.00%)

2 (2.00%) low mild

2 (2.00%) high mild

3 (3.00%) high severe
Benchmarking choose_size-hinted_from_1000_ChaCha20: Collecting 100 samples in estimated 5.0000 s (1.2B iterationschoose_size-hinted_from_1000_ChaCha20

time:   [4.2411 ns 4.2439 ns 4.2468 ns]

change: [-3.6234% -3.3519% -2.9879%] (p = 0.00 < 0.05)

Performance has improved.

Found 9 outliers among 100 measurements (9.00%)

3 (3.00%) low severe

4 (4.00%) high mild

2 (2.00%) high severe
choose_stable_from_1000_ChaCha20

time:   [2.9220 µs 2.9234 µs 2.9251 µs]

change: [+2.9325% +3.1211% +3.3032%] (p = 0.00 < 0.05)

Performance has regressed.

Found 14 outliers among 100 measurements (14.00%)

2 (2.00%) low severe

6 (6.00%) high mild

6 (6.00%) high severe
choose_unhinted_from_1000_ChaCha20

time:   [2.5877 µs 2.5937 µs 2.5994 µs]

change: [+4.2999% +4.6570% +5.0714%] (p = 0.00 < 0.05)

Performance has regressed.

Found 4 outliers among 100 measurements (4.00%)

3 (3.00%) high mild

1 (1.00%) high severe
choose_windowed_from_1000_ChaCha20

time:   [915.85 ns 921.40 ns 928.01 ns]

change: [+36.010% +36.576% +37.193%] (p = 0.00 < 0.05)

Performance has regressed.

Found 14 outliers among 100 measurements (14.00%)

2 (2.00%) high mild

12 (12.00%) high severe
choose_size-hinted_from_1_Pcg32

time:   [429.96 ps 430.20 ps 430.48 ps]

change: [+2.2400% +2.6419% +2.9365%] (p = 0.00 < 0.05)

Performance has regressed.

Found 10 outliers among 100 measurements (10.00%)

1 (1.00%) low mild

7 (7.00%) high mild

2 (2.00%) high severe
choose_stable_from_1_Pcg32

time:   [5.6490 ns 5.6615 ns 5.6759 ns]

change: [-15.651% -13.773% -11.886%] (p = 0.00 < 0.05)

Performance has improved.

Found 7 outliers among 100 measurements (7.00%)

6 (6.00%) high mild

1 (1.00%) high severe
choose_unhinted_from_1_Pcg32

time:   [4.1208 ns 4.1285 ns 4.1349 ns]

change: [+6.6747% +6.9684% +7.2107%] (p = 0.00 < 0.05)

Performance has regressed.
choose_windowed_from_1_Pcg32

time:   [5.3411 ns 5.3677 ns 5.3969 ns]

change: [-3.7357% -2.8700% -1.9170%] (p = 0.00 < 0.05)

Performance has improved.

Found 11 outliers among 100 measurements (11.00%)

7 (7.00%) high mild

4 (4.00%) high severe
choose_size-hinted_from_2_Pcg32

time:   [2.1998 ns 2.2006 ns 2.2017 ns]

change: [+86.310% +86.618% +86.940%] (p = 0.00 < 0.05)

Performance has regressed.

Found 5 outliers among 100 measurements (5.00%)

1 (1.00%) high mild

4 (4.00%) high severe
choose_stable_from_2_Pcg32

time:   [11.848 ns 11.850 ns 11.854 ns]

change: [+4.0385% +4.1321% +4.2191%] (p = 0.00 < 0.05)

Performance has regressed.

Found 6 outliers among 100 measurements (6.00%)

2 (2.00%) high mild

4 (4.00%) high severe
choose_unhinted_from_2_Pcg32

time:   [10.363 ns 10.384 ns 10.404 ns]

change: [+11.625% +12.011% +12.460%] (p = 0.00 < 0.05)

Performance has regressed.

Found 5 outliers among 100 measurements (5.00%)

3 (3.00%) high mild

2 (2.00%) high severe
choose_windowed_from_2_Pcg32

time:   [8.8771 ns 8.8879 ns 8.9022 ns]

change: [+12.376% +12.599% +12.812%] (p = 0.00 < 0.05)

Performance has regressed.

Found 11 outliers among 100 measurements (11.00%)

7 (7.00%) high mild

4 (4.00%) high severe
choose_size-hinted_from_3_Pcg32

time:   [2.1523 ns 2.1543 ns 2.1563 ns]

change: [+83.730% +84.097% +84.436%] (p = 0.00 < 0.05)

Performance has regressed.

Found 3 outliers among 100 measurements (3.00%)

3 (3.00%) high mild
choose_stable_from_3_Pcg32

time:   [22.232 ns 22.353 ns 22.492 ns]

change: [+1.3781% +1.5846% +1.8387%] (p = 0.00 < 0.05)

Performance has regressed.

Found 4 outliers among 100 measurements (4.00%)

4 (4.00%) high severe
choose_unhinted_from_3_Pcg32

time:   [20.813 ns 20.818 ns 20.822 ns]

change: [+8.5935% +8.6926% +8.7793%] (p = 0.00 < 0.05)

Performance has regressed.

Found 1 outliers among 100 measurements (1.00%)

1 (1.00%) high mild
choose_windowed_from_3_Pcg32

time:   [10.484 ns 10.494 ns 10.505 ns]

change: [+16.239% +16.449% +16.652%] (p = 0.00 < 0.05)

Performance has regressed.

Found 1 outliers among 100 measurements (1.00%)

1 (1.00%) high severe
choose_size-hinted_from_10_Pcg32

time:   [2.1910 ns 2.1924 ns 2.1937 ns]

change: [+85.907% +86.242% +86.567%] (p = 0.00 < 0.05)

Performance has regressed.

Found 7 outliers among 100 measurements (7.00%)

3 (3.00%) low severe

3 (3.00%) high mild

1 (1.00%) high severe
choose_stable_from_10_Pcg32

time:   [59.869 ns 59.938 ns 60.027 ns]

change: [+3.3939% +3.5207% +3.6508%] (p = 0.00 < 0.05)

Performance has regressed.

Found 4 outliers among 100 measurements (4.00%)

1 (1.00%) low mild

2 (2.00%) high mild

1 (1.00%) high severe
choose_unhinted_from_10_Pcg32

time:   [56.683 ns 56.705 ns 56.729 ns]

change: [+6.4272% +6.5180% +6.6045%] (p = 0.00 < 0.05)

Performance has regressed.

Found 3 outliers among 100 measurements (3.00%)

1 (1.00%) low mild

1 (1.00%) high mild

1 (1.00%) high severe
choose_windowed_from_10_Pcg32

time:   [18.747 ns 18.828 ns 18.918 ns]

change: [+20.566% +21.009% +21.493%] (p = 0.00 < 0.05)

Performance has regressed.

Found 14 outliers among 100 measurements (14.00%)

2 (2.00%) low severe

7 (7.00%) high mild

5 (5.00%) high severe
choose_size-hinted_from_100_Pcg32

time:   [2.1903 ns 2.1972 ns 2.2037 ns]

change: [+85.257% +85.891% +86.510%] (p = 0.00 < 0.05)

Performance has regressed.

Found 6 outliers among 100 measurements (6.00%)

6 (6.00%) high mild
choose_stable_from_100_Pcg32

time:   [364.40 ns 364.74 ns 365.37 ns]

change: [+2.3807% +2.5940% +2.7882%] (p = 0.00 < 0.05)

Performance has regressed.

Found 20 outliers among 100 measurements (20.00%)

6 (6.00%) low mild

4 (4.00%) high mild

10 (10.00%) high severe
choose_unhinted_from_100_Pcg32

time:   [349.07 ns 350.40 ns 351.89 ns]

change: [+5.5706% +5.8794% +6.2248%] (p = 0.00 < 0.05)

Performance has regressed.

Found 3 outliers among 100 measurements (3.00%)

3 (3.00%) high mild
choose_windowed_from_100_Pcg32

time:   [106.57 ns 107.04 ns 107.64 ns]

change: [+36.386% +37.128% +37.985%] (p = 0.00 < 0.05)

Performance has regressed.

Found 16 outliers among 100 measurements (16.00%)

2 (2.00%) high mild

14 (14.00%) high severe
choose_size-hinted_from_1000_Pcg32

time:   [2.2684 ns 2.2957 ns 2.3274 ns]

change: [+92.035% +92.996% +94.047%] (p = 0.00 < 0.05)

Performance has regressed.

Found 18 outliers among 100 measurements (18.00%)

7 (7.00%) low severe

1 (1.00%) low mild

3 (3.00%) high mild

7 (7.00%) high severe
choose_stable_from_1000_Pcg32

time:   [2.8277 µs 2.8294 µs 2.8315 µs]

change: [+2.0175% +2.5036% +2.8025%] (p = 0.00 < 0.05)

Performance has regressed.

Found 11 outliers among 100 measurements (11.00%)

1 (1.00%) low severe

2 (2.00%) low mild

4 (4.00%) high mild

4 (4.00%) high severe
choose_unhinted_from_1000_Pcg32

time:   [2.6369 µs 2.6384 µs 2.6401 µs]

change: [+2.6229% +2.7625% +2.9061%] (p = 0.00 < 0.05)

Performance has regressed.

Found 9 outliers among 100 measurements (9.00%)

1 (1.00%) low severe

1 (1.00%) low mild

5 (5.00%) high mild

2 (2.00%) high severe
choose_windowed_from_1000_Pcg32

time:   [775.98 ns 780.42 ns 785.17 ns]

change: [+57.292% +58.290% +59.351%] (p = 0.00 < 0.05)

Performance has regressed.

Found 1 outliers among 100 measurements (1.00%)

1 (1.00%) high severe
choose_size-hinted_from_1_Pcg64

time:   [422.75 ps 422.91 ps 423.06 ps]

change: [-78.014% -77.995% -77.977%] (p = 0.00 < 0.05)

Performance has improved.

Found 6 outliers among 100 measurements (6.00%)

1 (1.00%) low mild

2 (2.00%) high mild

3 (3.00%) high severe
choose_stable_from_1_Pcg64

time:   [6.7144 ns 6.7639 ns 6.8084 ns]

change: [-19.701% -18.096% -16.326%] (p = 0.00 < 0.05)

Performance has improved.
choose_unhinted_from_1_Pcg64

time:   [4.3655 ns 4.3715 ns 4.3796 ns]

change: [-1.4495% -1.1652% -0.7308%] (p = 0.00 < 0.05)

Change within noise threshold.

Found 11 outliers among 100 measurements (11.00%)

6 (6.00%) high mild

5 (5.00%) high severe
choose_windowed_from_1_Pcg64

time:   [5.5656 ns 5.5840 ns 5.6019 ns]

change: [-15.194% -14.445% -13.646%] (p = 0.00 < 0.05)

Performance has improved.

Found 3 outliers among 100 measurements (3.00%)

3 (3.00%) high mild
choose_size-hinted_from_2_Pcg64

time:   [3.0539 ns 3.0556 ns 3.0579 ns]

change: [-5.0176% -4.7436% -4.3441%] (p = 0.00 < 0.05)

Performance has improved.

Found 4 outliers among 100 measurements (4.00%)

1 (1.00%) high mild

3 (3.00%) high severe
choose_stable_from_2_Pcg64

time:   [12.695 ns 12.698 ns 12.701 ns]

change: [+1.5912% +1.7476% +1.9125%] (p = 0.00 < 0.05)

Performance has regressed.

Found 2 outliers among 100 measurements (2.00%)

2 (2.00%) low mild
choose_unhinted_from_2_Pcg64

time:   [10.981 ns 11.025 ns 11.081 ns]

change: [+2.4321% +2.8392% +3.2573%] (p = 0.00 < 0.05)

Performance has regressed.

Found 9 outliers among 100 measurements (9.00%)

9 (9.00%) high mild
choose_windowed_from_2_Pcg64

time:   [9.9539 ns 9.9566 ns 9.9592 ns]

change: [+12.049% +12.176% +12.296%] (p = 0.00 < 0.05)

Performance has regressed.

Found 9 outliers among 100 measurements (9.00%)

2 (2.00%) low severe

3 (3.00%) low mild

3 (3.00%) high mild

1 (1.00%) high severe
choose_size-hinted_from_3_Pcg64

time:   [3.0602 ns 3.0604 ns 3.0606 ns]

change: [-4.4339% -4.3124% -4.1954%] (p = 0.00 < 0.05)

Performance has improved.

Found 16 outliers among 100 measurements (16.00%)

5 (5.00%) low severe

2 (2.00%) low mild

8 (8.00%) high mild

1 (1.00%) high severe
choose_stable_from_3_Pcg64

time:   [23.798 ns 23.806 ns 23.816 ns]

change: [+1.5472% +1.6436% +1.7350%] (p = 0.00 < 0.05)

Performance has regressed.

Found 7 outliers among 100 measurements (7.00%)

1 (1.00%) low severe

2 (2.00%) high mild

4 (4.00%) high severe
choose_unhinted_from_3_Pcg64

time:   [21.441 ns 21.460 ns 21.483 ns]

change: [+1.6660% +1.8470% +2.1155%] (p = 0.00 < 0.05)

Performance has regressed.

Found 12 outliers among 100 measurements (12.00%)

1 (1.00%) low mild

4 (4.00%) high mild

7 (7.00%) high severe
choose_windowed_from_3_Pcg64

time:   [11.323 ns 11.328 ns 11.332 ns]

change: [+11.506% +11.604% +11.695%] (p = 0.00 < 0.05)

Performance has regressed.

Found 6 outliers among 100 measurements (6.00%)

1 (1.00%) low mild

4 (4.00%) high mild

1 (1.00%) high severe
choose_size-hinted_from_10_Pcg64

time:   [3.0400 ns 3.0422 ns 3.0447 ns]

change: [-5.5479% -5.4691% -5.3933%] (p = 0.00 < 0.05)

Performance has improved.

Found 5 outliers among 100 measurements (5.00%)

4 (4.00%) high mild

1 (1.00%) high severe
choose_stable_from_10_Pcg64

time:   [62.691 ns 62.853 ns 63.022 ns]

change: [+3.6084% +4.1060% +4.6822%] (p = 0.00 < 0.05)

Performance has regressed.

Found 7 outliers among 100 measurements (7.00%)

5 (5.00%) high mild

2 (2.00%) high severe
choose_unhinted_from_10_Pcg64

time:   [55.935 ns 56.072 ns 56.213 ns]

change: [+3.0192% +3.3180% +3.6652%] (p = 0.00 < 0.05)

Performance has regressed.

Found 4 outliers among 100 measurements (4.00%)

3 (3.00%) high mild

1 (1.00%) high severe
choose_windowed_from_10_Pcg64

time:   [20.435 ns 20.512 ns 20.598 ns]

change: [+12.132% +12.356% +12.629%] (p = 0.00 < 0.05)

Performance has regressed.

Found 14 outliers among 100 measurements (14.00%)

3 (3.00%) low mild

2 (2.00%) high mild

9 (9.00%) high severe
choose_size-hinted_from_100_Pcg64

time:   [3.1388 ns 3.1474 ns 3.1567 ns]

change: [-2.8788% -2.4525% -2.0175%] (p = 0.00 < 0.05)

Performance has improved.
choose_stable_from_100_Pcg64

time:   [369.21 ns 369.45 ns 369.74 ns]

change: [+3.2396% +3.3271% +3.4248%] (p = 0.00 < 0.05)

Performance has regressed.
choose_unhinted_from_100_Pcg64

time:   [321.43 ns 321.64 ns 321.86 ns]

change: [+1.3127% +1.4030% +1.4982%] (p = 0.00 < 0.05)

Performance has regressed.

Found 4 outliers among 100 measurements (4.00%)

2 (2.00%) high mild

2 (2.00%) high severe
choose_windowed_from_100_Pcg64

time:   [108.48 ns 108.55 ns 108.65 ns]

change: [+9.1049% +9.2832% +9.4614%] (p = 0.00 < 0.05)

Performance has regressed.

Found 7 outliers among 100 measurements (7.00%)

2 (2.00%) low mild

2 (2.00%) high mild

3 (3.00%) high severe
choose_size-hinted_from_1000_Pcg64

time:   [3.0872 ns 3.0998 ns 3.1163 ns]

change: [-2.1530% -1.6326% -1.0692%] (p = 0.00 < 0.05)

Performance has improved.

Found 11 outliers among 100 measurements (11.00%)

4 (4.00%) high mild

7 (7.00%) high severe
choose_stable_from_1000_Pcg64

time:   [2.8468 µs 2.8499 µs 2.8547 µs]

change: [+2.6786% +2.8290% +3.0302%] (p = 0.00 < 0.05)

Performance has regressed.

Found 9 outliers among 100 measurements (9.00%)

6 (6.00%) high mild

3 (3.00%) high severe
choose_unhinted_from_1000_Pcg64

time:   [2.4865 µs 2.4922 µs 2.4978 µs]

change: [+3.1856% +3.3348% +3.4862%] (p = 0.00 < 0.05)

Performance has regressed.

Found 16 outliers among 100 measurements (16.00%)

3 (3.00%) low mild

5 (5.00%) high mild

8 (8.00%) high severe
choose_windowed_from_1000_Pcg64

time:   [746.74 ns 746.90 ns 747.08 ns]

change: [+11.577% +11.689% +11.798%] (p = 0.00 < 0.05)

Performance has regressed.

Found 10 outliers among 100 measurements (10.00%)

5 (5.00%) high mild

5 (5.00%) high severe
 Running src/shuffle.rs (target/release/deps/shuffle-68b64b946bda8444)

shuffle_1_ChaCha12      time:   [212.49 ps 213.08 ps 213.65 ps]

change: [+0.8434% +1.2614% +1.5863%] (p = 0.00 < 0.05)

Change within noise threshold.
shuffle_2_ChaCha12      time:   [7.0658 ns 7.1450 ns 7.2180 ns]

change: [+4.7196% +5.5800% +6.4672%] (p = 0.00 < 0.05)

Performance has regressed.

Found 2 outliers among 100 measurements (2.00%)

2 (2.00%) high mild
shuffle_3_ChaCha12      time:   [7.8170 ns 7.8459 ns 7.8808 ns]

change: [+4.6635% +4.9737% +5.2525%] (p = 0.00 < 0.05)

Performance has regressed.
shuffle_10_ChaCha12     time:   [18.745 ns 18.784 ns 18.829 ns]

change: [+1.7949% +1.9896% +2.1897%] (p = 0.00 < 0.05)

Performance has regressed.

Found 1 outliers among 100 measurements (1.00%)

1 (1.00%) high mild
partial_shuffle_10_ChaCha12

time:   [14.428 ns 14.462 ns 14.501 ns]

change: [+3.3524% +3.6645% +3.9495%] (p = 0.00 < 0.05)

Performance has regressed.

Found 8 outliers among 100 measurements (8.00%)

7 (7.00%) high mild

1 (1.00%) high severe
shuffle_100_ChaCha12    time:   [232.08 ns 232.49 ns 232.89 ns]

change: [-0.7536% -0.5515% -0.3411%] (p = 0.00 < 0.05)

Change within noise threshold.
partial_shuffle_100_ChaCha12

time:   [116.43 ns 116.75 ns 117.05 ns]

change: [+3.7060% +4.4136% +5.0969%] (p = 0.00 < 0.05)

Performance has regressed.

Found 3 outliers among 100 measurements (3.00%)

1 (1.00%) low mild

1 (1.00%) high mild

1 (1.00%) high severe
shuffle_1000_ChaCha12   time:   [2.0994 µs 2.1037 µs 2.1082 µs]

change: [-4.6414% -3.8207% -3.1850%] (p = 0.00 < 0.05)

Performance has improved.

Found 8 outliers among 100 measurements (8.00%)

8 (8.00%) high mild
partial_shuffle_1000_ChaCha12

time:   [993.52 ns 995.41 ns 997.41 ns]

change: [+4.9496% +5.4320% +5.9154%] (p = 0.00 < 0.05)

Performance has regressed.

Found 6 outliers among 100 measurements (6.00%)

4 (4.00%) low mild

2 (2.00%) high mild
shuffle_10000_ChaCha12  time:   [21.619 µs 21.663 µs 21.721 µs]

change: [+2.6866% +3.6380% +4.9686%] (p = 0.00 < 0.05)

Performance has regressed.

Found 20 outliers among 100 measurements (20.00%)

10 (10.00%) high mild

10 (10.00%) high severe
partial_shuffle_10000_ChaCha12

time:   [10.625 µs 10.703 µs 10.786 µs]

change: [+6.8013% +7.2338% +7.6390%] (p = 0.00 < 0.05)

Performance has regressed.

Found 15 outliers among 100 measurements (15.00%)

3 (3.00%) low mild

5 (5.00%) high mild

7 (7.00%) high severe
shuffle_1_Pcg32         time:   [215.31 ps 215.74 ps 216.29 ps]

change: [+2.7588% +3.0068% +3.2805%] (p = 0.00 < 0.05)

Performance has regressed.

Found 3 outliers among 100 measurements (3.00%)

2 (2.00%) high mild

1 (1.00%) high severe
shuffle_2_Pcg32         time:   [9.0827 ns 9.5501 ns 9.9967 ns]

change: [+4.3966% +10.521% +17.494%] (p = 0.00 < 0.05)

Performance has regressed.
shuffle_3_Pcg32         time:   [7.4608 ns 7.5080 ns 7.5516 ns]

change: [-2.3084% -1.3876% -0.4683%] (p = 0.00 < 0.05)

Change within noise threshold.

Found 1 outliers among 100 measurements (1.00%)

1 (1.00%) high mild
shuffle_10_Pcg32        time:   [18.158 ns 18.317 ns 18.478 ns]

change: [+5.7873% +6.8524% +8.0101%] (p = 0.00 < 0.05)

Performance has regressed.

Found 2 outliers among 100 measurements (2.00%)

1 (1.00%) high mild

1 (1.00%) high severe
partial_shuffle_10_Pcg32

time:   [14.409 ns 14.468 ns 14.520 ns]

change: [-0.3578% +0.1410% +0.6260%] (p = 0.52 > 0.05)

No change in performance detected.
shuffle_100_Pcg32       time:   [172.94 ns 173.21 ns 173.52 ns]

change: [+1.2509% +1.4167% +1.5827%] (p = 0.00 < 0.05)

Performance has regressed.

Found 6 outliers among 100 measurements (6.00%)

6 (6.00%) high mild
partial_shuffle_100_Pcg32

time:   [93.122 ns 93.297 ns 93.494 ns]

change: [+1.1609% +1.7122% +2.1805%] (p = 0.00 < 0.05)

Performance has regressed.

Found 2 outliers among 100 measurements (2.00%)

1 (1.00%) high mild

1 (1.00%) high severe
shuffle_1000_Pcg32      time:   [1.7067 µs 1.7083 µs 1.7098 µs]

change: [-4.4770% -2.9869% -1.7272%] (p = 0.00 < 0.05)

Performance has improved.

Found 9 outliers among 100 measurements (9.00%)

2 (2.00%) low severe

1 (1.00%) low mild

3 (3.00%) high mild

3 (3.00%) high severe
partial_shuffle_1000_Pcg32

time:   [857.50 ns 859.27 ns 861.87 ns]

change: [+0.5617% +1.0207% +1.4152%] (p = 0.00 < 0.05)

Change within noise threshold.

Found 6 outliers among 100 measurements (6.00%)

3 (3.00%) high mild

3 (3.00%) high severe
shuffle_10000_Pcg32     time:   [16.735 µs 16.755 µs 16.777 µs]

change: [-0.6981% -0.1695% +0.4371%] (p = 0.62 > 0.05)

No change in performance detected.

Found 4 outliers among 100 measurements (4.00%)

1 (1.00%) low mild

2 (2.00%) high mild

1 (1.00%) high severe
partial_shuffle_10000_Pcg32

time:   [8.2061 µs 8.2146 µs 8.2227 µs]

change: [+0.3206% +0.6257% +0.9335%] (p = 0.00 < 0.05)

Change within noise threshold.

Found 3 outliers among 100 measurements (3.00%)

1 (1.00%) low severe

1 (1.00%) low mild

1 (1.00%) high mild
shuffle_1_Pcg64         time:   [213.99 ps 214.05 ps 214.14 ps]

change: [-0.0089% +0.0851% +0.2149%] (p = 0.12 > 0.05)

No change in performance detected.

Found 10 outliers among 100 measurements (10.00%)

5 (5.00%) high mild

5 (5.00%) high severe
shuffle_2_Pcg64         time:   [7.0886 ns 7.1192 ns 7.1507 ns]

change: [-31.118% -27.988% -24.609%] (p = 0.00 < 0.05)

Performance has improved.

Found 2 outliers among 100 measurements (2.00%)

2 (2.00%) high mild
shuffle_3_Pcg64         time:   [8.5383 ns 8.6103 ns 8.6956 ns]

change: [-29.647% -28.206% -26.673%] (p = 0.00 < 0.05)

Performance has improved.

Found 3 outliers among 100 measurements (3.00%)

3 (3.00%) high mild
shuffle_10_Pcg64        time:   [18.648 ns 18.664 ns 18.680 ns]

change: [-8.3002% -7.7604% -7.2105%] (p = 0.00 < 0.05)

Performance has improved.

Found 7 outliers among 100 measurements (7.00%)

7 (7.00%) high mild
partial_shuffle_10_Pcg64

time:   [14.545 ns 14.565 ns 14.583 ns]

change: [-7.1614% -6.8663% -6.5580%] (p = 0.00 < 0.05)

Performance has improved.

Found 7 outliers among 100 measurements (7.00%)

3 (3.00%) low severe

4 (4.00%) high mild
shuffle_100_Pcg64       time:   [187.73 ns 188.80 ns 189.83 ns]

change: [-4.0263% -3.7459% -3.4286%] (p = 0.00 < 0.05)

Performance has improved.

Found 18 outliers among 100 measurements (18.00%)

2 (2.00%) low mild

6 (6.00%) high mild

10 (10.00%) high severe
partial_shuffle_100_Pcg64

time:   [96.249 ns 96.532 ns 96.885 ns]

change: [-6.4110% -5.9369% -5.4841%] (p = 0.00 < 0.05)

Performance has improved.

Found 1 outliers among 100 measurements (1.00%)

1 (1.00%) high mild
shuffle_1000_Pcg64      time:   [1.8515 µs 1.8551 µs 1.8588 µs]

change: [-4.9568% -4.4750% -3.7605%] (p = 0.00 < 0.05)

Performance has improved.

Found 7 outliers among 100 measurements (7.00%)

7 (7.00%) high severe
partial_shuffle_1000_Pcg64

time:   [942.37 ns 948.80 ns 957.16 ns]

change: [-5.0863% -4.1937% -2.8587%] (p = 0.00 < 0.05)

Performance has improved.

Found 6 outliers among 100 measurements (6.00%)

3 (3.00%) high mild

3 (3.00%) high severe
shuffle_10000_Pcg64     time:   [19.041 µs 19.062 µs 19.083 µs]

change: [-6.9676% -6.8235% -6.6512%] (p = 0.00 < 0.05)

Performance has improved.

Found 4 outliers among 100 measurements (4.00%)

2 (2.00%) high mild

2 (2.00%) high severe
partial_shuffle_10000_Pcg64

time:   [9.4146 µs 9.4263 µs 9.4428 µs]

change: [-6.7491% -6.4804% -6.0440%] (p = 0.00 < 0.05)

Performance has improved.

Found 7 outliers among 100 measurements (7.00%)

2 (2.00%) low severe

2 (2.00%) low mild

1 (1.00%) high mild

2 (2.00%) high severe
 Running src/uniform.rs (target/release/deps/uniform-a5c2754802b04fc5)
 Running src/uniform_float.rs (target/release/deps/uniform_float-c70085eeebdc4543)
 Running benches/weighted.rs (target/release/deps/weighted-fbd407d40100d55b)

running 0 tests
test result: ok. 0 passed; 0 failed; 0 ignored; 0 measured; 2 filtered out; finished in 0.00s

[vks]

Some regressions are quite substantial, do we understand why? E.g.:

choose_windowed_from_1000_Pcg32
time:   [775.98 ns 780.42 ns 785.17 ns]
change: [+57.292% +58.290% +59.351%] (p = 0.00 < 0.05)

[dhardy]

Some regressions are quite substantial, do we understand why?

choose is impacted since it uses gen_range in a loop... but that previously used gen_index which should be equivalent. So no.

More half-baked thoughts:

• Code like IteratorRandom::choose should be able to select a 32-bit or 64-bit sampler once, then always use that. There are probably other locations that we could take this approach.
• The idea not to support gen_range for usize was partly to force people to think more closely about such cases — but this is quite disruptive.
• Maybe we should only support 32-bit sampling for usize? There can't be that many uses requiring 64-bit sampling.

---

I just pushed a slightly different impl of UniformUsize but this doesn't have a big impact on benches.

[dhardy]

Adding support for RangeTo* improves performance a little in some cases (presumably due to constant folding).

[dhardy]

choose_windowed_from_1000_Pcg32

Given that the choose_windowed (and related) benchmarks were deliberately designed to be hard to optimise for (and are likely not common patterns), I don't think we should be too concerned about these.

I'd like to get some more general benchmarking done, but decided to push #1490 first.

[dhardy]

Rebased. Running benches now...

[dhardy]

Comparing 7bdd833 to master (9e030aa): see the attachment.
output.zip
output.html.gz

In summary, the choose benchmarks have some significant changes (+105% to -80%, though the latter is on a <500ps test) and the shuffle benchmarks have some moderately significant changes. Wins and losses vaguely balance each other out with one exception (the worst relative result):

seq_slice_choose_1_of_1000
                        time:   [2.3746 ns 2.3785 ns 2.3823 ns]
                        change: [+105.12% +105.42% +105.73%] (p = 0.00 < 0.05)
                        Performance has regressed.

Overall, benchmark results are probably a smidgen worse, with most of the differences likely due to different inlining.

I find this acceptable (given our goal of portable ranged usize variates).