benchmark.md

Benchmarks

These benchmarks have been done using criterium. You may find the code in vector_ops.benchmarks.bench.

As with any benchmarks, these don't project a perfect view of reality and I don't claim they do as such. They solely deem as an indicator of the degree of improvement these optimisations might have over the Clojure alternatives.

Benchmarks have been run on:

Operating System: Ubuntu 16.04 (64 bit)

CPU: Intel i7-6700HQ CPU (6MB cache, 2.60 - 3.50 GHz x 4 cores)

Memory: 16 GB DDR4

Concatenation

Note: -

Clojure: (vec (concat v1 v2))

Optimised: (vector-ops.core/concatv v1 v2)

Size	Clojure	Optimised
0	361.810252 ns	7.196989 ns
10	558.245696 ns	398.796734 ns
100	2.240790 µs	846.869074 ns
1000	19.031484 µs	7.450031 µs
10000	189.564562 µs	71.827575 µs
100000	1.927494 ms	900.493711 µs
1000000	23.703857 ms	12.814612 ms

Batch concatenation

Note: Concatenated 10 vectors of given size at once.

Clojure: (vec (apply concat vs))

Optimised: (apply vector-ops.core/concatv vs)

Size	Clojure	Optimised
0	756.808988 ns	393.664749 ns
10	5.480622 µs	2.163517 µs
100	34.627662 µs	6.678997 µs
1000	318.344501 µs	80.677171 µs
10000	3.214963 ms	1.030669 ms
100000	45.329293 ms	10.913499 ms
1000000	893.050373 ms	162.112949 ms

Dropping

Note: -

Clojure: (vec (drop n v))

Optimised: (vector-ops.core/dropv n v)

Size	Clojure	Optimised
0	154.673231 ns	18.564517 ns
10	461.134753 ns	444.081258 ns
100	3.125726 µs	517.383259 ns
1000	29.222027 µs	1.110373 µs
10000	347.917215 µs	8.011953 µs
100000	3.791349 ms	95.015811 µs
1000000	32.418843 ms	695.911450 µs

Dropping last

Note: -

Clojure: (vec (drop-last n v))

Optimised: (vector-ops.core/dropv-last n v)

Size	Clojure	Optimised
0	180.366026 ns	53.649340 ns
10	903.780801 ns	445.236752 ns
100	7.046274 µs	481.406993 ns
1000	68.173964 µs	722.727751 ns
10000	686.190882 µs	3.796002 µs
100000	7.932386 ms	46.238683 µs
1000000	75.833936 ms	517.512106 µs

Dropping last while

Note: -

Clojure: (vec (->> (reverse v) (drop-while p) (reverse)))

Optimised: (vector-ops.core/dropv-last-while n v)

Size	Clojure	Optimised
0	68.505042 ns	8.913599 ns
10	1.167079 µs	544.383748 ns
100	9.456833 µs	1.522609 µs
1000	90.773551 µs	11.015628 µs
10000	900.016235 µs	105.808669 µs
100000	9.058089 ms	1.067737 ms
1000000	93.121276 ms	11.336951 ms

Dropping while

Note: -

Clojure: (vec (drop-while n v))

Optimised: (vector-ops.core/dropv-while n v)

Size	Clojure	Optimised
0	131.099668 ns	9.535428 ns
10	500.303606 ns	613.369386 ns
100	3.462573 µs	1.614100 µs
1000	33.210243 µs	11.893156 µs
10000	334.205152 µs	116.636202 µs
100000	3.334074 ms	1.225122 ms
1000000	36.378544 ms	16.444744 ms

Mapping

Note: -

Clojure: (mapv f vs)

Optimised: (vector-ops.core/mapv f vs)

Size	Clojure	Optimised
0	181.796290 ns	7.399992 ns
10	454.735210 ns	675.667162 ns
100	3.040831 µs	2.270018 µs
1000	29.574956 µs	18.165056 µs
10000	292.305066 µs	196.545751 µs
100000	2.959955 ms	1.815747 ms
1000000	33.691650 ms	21.845780 ms

Slicing

Note: Like clojure.core/subvec, but returns a proper vector, not a subvector

Clojure: (vec (->> v (take m) (drop n)))

Optimised: (vector-ops.core/slicev v n m)

Size	Clojure	Optimised
0	176.707260 ns	42.609155 ns
10	923.916195 ns	536.723375 ns
100	6.906388 µs	621.573771 ns
1000	67.022146 µs	1.253480 µs
10000	667.791529 µs	8.570844 µs
100000	6.780595 ms	46.800204 µs
1000000	71.247426 ms	914.541700 µs

Splitting at

Note: -

Clojure: (mapv vec (split-at n v))

Optimised: (vector-ops.core/splitv-at n v)

Size	Clojure	Optimised
0	611.454342 ns	47.140462 ns
10	1.410785 µs	807.516807 ns
100	8.159660 µs	931.020481 ns
1000	75.594338 µs	1.731148 µs
10000	752.465039 µs	11.221048 µs
100000	7.625630 ms	132.211750 µs
1000000	84.335322 ms	1.310956 ms

Taking

Note: -

Clojure: (vec (take n v))

Optimised: (vector-ops.core/takev n v)

Size	Clojure	Optimised
0	121.605930 ns	19.252959 ns
10	575.795633 ns	387.547367 ns
100	4.717715 µs	428.134181 ns
1000	45.774762 µs	660.637803 ns
10000	457.472861 µs	3.668902 µs
100000	4.612023 ms	49.418577 µs
1000000	47.498904 ms	510.973851 µs

Taking last

Note: -

Clojure: (vec (take-last n v))

Optimised: (vector-ops.core/takev-last n v)

Size	Clojure	Optimised
0	81.464990 ns	53.871355 ns
10	550.774035 ns	458.354913 ns
100	3.929223 µs	535.625752 ns
1000	37.912209 µs	1.116335 µs
10000	384.062047 µs	7.406457 µs
100000	3.869972 ms	85.214376 µs
1000000	41.998881 ms	510.553739 µs

Taking last while

Note: -

Clojure: (vec (->> (reverse v) (take-while p) (reverse)))

Optimised: (vector-ops.core/takev-last-while n v)

Size	Clojure	Optimised
0	62.448233 ns	8.927027 ns
10	1.085941 µs	575.353576 ns
100	11.152364 µs	1.588850 µs
1000	111.231348 µs	11.537438 µs
10000	1.115974 ms	111.004553 µs
100000	11.322195 ms	1.125774 ms
1000000	115.946913 ms	12.238548 ms

Taking while

Note: -

Clojure: (vec (take-while n v))

Optimised: (vector-ops.core/takev-while n v)

Size	Clojure	Optimised
0	123.973344 ns	9.025158 ns
10	674.887858 ns	547.634108 ns
100	5.387527 µs	1.537828 µs
1000	52.125426 µs	11.015713 µs
10000	523.327036 µs	105.406068 µs
100000	5.242220 ms	1.080573 ms
1000000	53.885682 ms	13.177093 ms