Map implementations in various ways.
You can add various_map as dependencies in your mix.exs. At the moment you will have to use a Git dependency while we work on our first release:
def deps do
[
{:various_map, "~> 0.1", github: "zacky1972/various_map"}
]
endDocumentation can be generated with ExDoc by mix docs.
Run mix run -r bench/map_bench.exs, then you'll get results similar to the following:
% mix run -r bench/map_bench.exs
Operating System: macOS
CPU Information: Apple M2
Number of Available Cores: 8
Available memory: 24 GB
Elixir 1.14.0
Erlang 25.0.4
Benchmark suite executing with the following configuration:
warmup: 2 s
time: 5 s
memory time: 0 ns
reduction time: 0 ns
parallel: 1
inputs: size 10, size 100, size 1000, size 10000, size 100000
Estimated total run time: 1.17 min
Benchmarking Map get with input size 10 ...
Benchmarking Map get with input size 100 ...
Benchmarking Map get with input size 1000 ...
Benchmarking Map get with input size 10000 ...
Benchmarking Map get with input size 100000 ...
Benchmarking Map put with input size 10 ...
Benchmarking Map put with input size 100 ...
Benchmarking Map put with input size 1000 ...
Benchmarking Map put with input size 10000 ...
Benchmarking Map put with input size 100000 ...
##### With input size 10 #####
Name ips average deviation median 99th %
Map get 13.88 M 72.05 ns ±21396.29% 42 ns 84 ns
Map put 6.98 M 143.27 ns ±2200.77% 120.90 ns 162.50 ns
Comparison:
Map get 13.88 M
Map put 6.98 M - 1.99x slower +71.22 ns
##### With input size 100 #####
Name ips average deviation median 99th %
Map get 11.14 M 89.78 ns ±15571.08% 83 ns 125 ns
Map put 8.60 M 116.29 ns ±20030.50% 83 ns 125 ns
Comparison:
Map get 11.14 M
Map put 8.60 M - 1.30x slower +26.51 ns
##### With input size 1000 #####
Name ips average deviation median 99th %
Map get 10.92 M 91.54 ns ±14534.79% 83 ns 125 ns
Map put 7.95 M 125.78 ns ±25238.19% 83 ns 125 ns
Comparison:
Map get 10.92 M
Map put 7.95 M - 1.37x slower +34.24 ns
##### With input size 10000 #####
Name ips average deviation median 99th %
Map get 9.11 M 109.80 ns ±13974.75% 83 ns 125 ns
Map put 8.09 M 123.65 ns ±24992.66% 83 ns 125 ns
Comparison:
Map get 9.11 M
Map put 8.09 M - 1.13x slower +13.85 ns
##### With input size 100000 #####
Name ips average deviation median 99th %
Map get 8.16 M 122.55 ns ±15375.64% 84 ns 209 ns
Map put 6.86 M 145.67 ns ±2532.76% 120.90 ns 162.50 ns
Comparison:
Map get 8.16 M
Map put 6.86 M - 1.19x slower +23.12 ns
The operations of Map.get/3 and Map.put/3 work in logarithmic time, which means that the time it takes to find keys grows as the map grows, but it's not directly proportional to the map size.
The results show the growth of the execution time of the operations seems to be very gentle, even considering that they work in logarithmic time.
Map is 12.1-18.2x and 77.4-119x faster than VariousMap.EtsMap and VariousMap.MnesiaMap, respectively.
VariousMap.EtsMap is a module compatible to Map but using ETS.
Run mix run -r bench/ets_map_bench.exs, then you'll get results similar to the following:
% mix run -r bench/ets_map_bench.exs
Operating System: macOS
CPU Information: Apple M2
Number of Available Cores: 8
Available memory: 24 GB
Elixir 1.14.0
Erlang 25.0.4
Benchmark suite executing with the following configuration:
warmup: 2 s
time: 5 s
memory time: 0 ns
reduction time: 0 ns
parallel: 1
inputs: size 10, size 100, size 1000, size 10000, size 100000
Estimated total run time: 1.17 min
Benchmarking ETS Map get with input size 10 ...
Benchmarking ETS Map get with input size 100 ...
Benchmarking ETS Map get with input size 1000 ...
Benchmarking ETS Map get with input size 10000 ...
Benchmarking ETS Map get with input size 100000 ...
Benchmarking ETS Map put with input size 10 ...
Benchmarking ETS Map put with input size 100 ...
Benchmarking ETS Map put with input size 1000 ...
Benchmarking ETS Map put with input size 10000 ...
Benchmarking ETS Map put with input size 100000 ...
##### With input size 10 #####
Name ips average deviation median 99th %
ETS Map get 367.06 K 2.72 μs ±11.40% 2.67 μs 3.49 μs
ETS Map put 358.20 K 2.79 μs ±10.34% 2.75 μs 3.63 μs
Comparison:
ETS Map get 367.06 K
ETS Map put 358.20 K - 1.02x slower +0.0674 μs
##### With input size 100 #####
Name ips average deviation median 99th %
ETS Map get 10.11 M 0.0989 μs ±37.50% 0.0840 μs 0.125 μs
ETS Map put 0.0415 M 24.11 μs ±3.89% 23.80 μs 26.77 μs
Comparison:
ETS Map get 10.11 M
ETS Map put 0.0415 M - 243.84x slower +24.01 μs
##### With input size 1000 #####
Name ips average deviation median 99th %
ETS Map get 2.92 K 342.08 μs ±1.93% 341.70 μs 357.35 μs
ETS Map put 2.91 K 343.56 μs ±2.44% 341.52 μs 367.19 μs
Comparison:
ETS Map get 2.92 K
ETS Map put 2.91 K - 1.00x slower +1.48 μs
##### With input size 10000 #####
Name ips average deviation median 99th %
ETS Map get 260.82 3.83 ms ±4.39% 3.82 ms 5.15 ms
ETS Map put 258.13 3.87 ms ±1.33% 3.88 ms 4.02 ms
Comparison:
ETS Map get 260.82
ETS Map put 258.13 - 1.01x slower +0.0399 ms
##### With input size 100000 #####
Name ips average deviation median 99th %
ETS Map put 566.85 K 1.76 μs ±82.06% 0.71 μs 5.14 μs
ETS Map get 447.76 K 2.23 μs ±63.62% 2.50 μs 5.19 μs
Comparison:
ETS Map put 566.85 K
ETS Map get 447.76 K - 1.27x slower +0.47 μs
VariousMap.EtsMap is 12.1-18.2x slower than Map.
VariousMap.EtsMap is 6.40-6.51x faster than VariousMap.MnesiaMap.
VariousMap.MnesiaMap is a module compatible to Map but using Mnesia.
Run mix run -r bench/mnesia_map_bench.exs, then you'll get results similar to the following:
% mix run -r bench/mnesia_map_bench.exs
Operating System: macOS
CPU Information: Apple M2
Number of Available Cores: 8
Available memory: 24 GB
Elixir 1.14.0
Erlang 25.0.4
Benchmark suite executing with the following configuration:
warmup: 2 s
time: 5 s
memory time: 0 ns
reduction time: 0 ns
parallel: 1
inputs: size 10, size 100, size 1000, size 10000, size 100000
Estimated total run time: 1.17 min
Benchmarking Mnesia Map get with input size 10 ...
Benchmarking Mnesia Map get with input size 100 ...
Benchmarking Mnesia Map get with input size 1000 ...
Benchmarking Mnesia Map get with input size 10000 ...
Benchmarking Mnesia Map get with input size 100000 ...
Benchmarking Mnesia Map put with input size 10 ...
Benchmarking Mnesia Map put with input size 100 ...
Benchmarking Mnesia Map put with input size 1000 ...
Benchmarking Mnesia Map put with input size 10000 ...
Benchmarking Mnesia Map put with input size 100000 ...
##### With input size 10 #####
Name ips average deviation median 99th %
Mnesia Map put 91.70 K 10.90 μs ±13.83% 10.67 μs 18.70 μs
Mnesia Map get 90.34 K 11.07 μs ±13.04% 10.83 μs 18.11 μs
Comparison:
Mnesia Map put 91.70 K
Mnesia Map get 90.34 K - 1.02x slower +0.165 μs
##### With input size 100 #####
Name ips average deviation median 99th %
Mnesia Map put 88.46 K 11.30 μs ±26.82% 9.88 μs 21.53 μs
Mnesia Map get 86.74 K 11.53 μs ±27.53% 10.50 μs 22.88 μs
Comparison:
Mnesia Map put 88.46 K
Mnesia Map get 86.74 K - 1.02x slower +0.22 μs
##### With input size 1000 #####
Name ips average deviation median 99th %
Mnesia Map put 96.27 K 10.39 μs ±19.80% 9.75 μs 19.92 μs
Mnesia Map get 91.63 K 10.91 μs ±15.61% 10.17 μs 17.70 μs
Comparison:
Mnesia Map put 96.27 K
Mnesia Map get 91.63 K - 1.05x slower +0.53 μs
##### With input size 10000 #####
Name ips average deviation median 99th %
Mnesia Map put 104.36 K 9.58 μs ±37.39% 8.46 μs 29.58 μs
Mnesia Map get 95.75 K 10.44 μs ±26.64% 9.52 μs 25.75 μs
Comparison:
Mnesia Map put 104.36 K
Mnesia Map get 95.75 K - 1.09x slower +0.86 μs
##### With input size 100000 #####
Name ips average deviation median 99th %
Mnesia Map put 88.56 K 11.29 μs ±23.40% 10.29 μs 16 μs
Mnesia Map get 68.77 K 14.54 μs ±10.44% 14.96 μs 15.96 μs
Comparison:
Mnesia Map put 88.56 K
Mnesia Map get 68.77 K - 1.29x slower +3.25 μs
VariousMap.MnesiaMap is 77.4-119x and 6.40-6.51x slower than Map and VariousMap.EtsMap, respectively.
VariousMap.MapGraph is a module for a graph that can use for both a directed and a undirected graph using Map.
Run mix run -r bench/map_graph_bench.exs, then you'll get results similar to the following:
% mix run -r bench/map_graph_bench.exs
time of create random_edges: 0.562msec.
time of create random_edges: 2.969msec.
time of create random_edges: 602.218msec.
time of create random_edges: 24840.19msec.
time of create MapGraph: 0.007msec
time of create MapGraph: 0.001msec
time of create MapGraph: 0.012msec
time of create MapGraph: 0.017msec
time of create MapGraph: 0.028msec
time of create MapGraph: 2.025msec
time of create MapGraph: 0.157msec
time of create MapGraph: 0.311msec
time of create MapGraph: 817.008msec
time of create MapGraph: 1.718msec
time of create MapGraph: 2.562msec
time of create MapGraph: 22880.225msec
Operating System: macOS
CPU Information: Apple M2
Number of Available Cores: 8
Available memory: 24 GB
Elixir 1.14.0
Erlang 25.0.4
Benchmark suite executing with the following configuration:
warmup: 2 s
time: 5 s
memory time: 0 ns
reduction time: 0 ns
parallel: 1
inputs: Graph {num_vertex, num_edge} = {10, 10}, Graph {num_vertex, num_edge} = {10, 20}, Graph {num_vertex, num_edge} = {10, 90}, Graph {num_vertex, num_edge} = {100, 100}, Graph {num_vertex, num_edge} = {100, 200}, Graph {num_vertex, num_edge} = {100, 9900}, Graph {num_vertex, num_edge} = {1000, 1000}, Graph {num_vertex, num_edge} = {1000, 2000}, Graph {num_vertex, num_edge} = {1000, 999000}, Graph {num_vertex, num_edge} = {5000, 10000}, Graph {num_vertex, num_edge} = {5000, 24995000}, Graph {num_vertex, num_edge} = {5000, 5000}
Estimated total run time: 2.80 min
Benchmarking MapGraph.get with input Graph {num_vertex, num_edge} = {10, 10} ...
Benchmarking MapGraph.get with input Graph {num_vertex, num_edge} = {10, 20} ...
Benchmarking MapGraph.get with input Graph {num_vertex, num_edge} = {10, 90} ...
Benchmarking MapGraph.get with input Graph {num_vertex, num_edge} = {100, 100} ...
Benchmarking MapGraph.get with input Graph {num_vertex, num_edge} = {100, 200} ...
Benchmarking MapGraph.get with input Graph {num_vertex, num_edge} = {100, 9900} ...
Benchmarking MapGraph.get with input Graph {num_vertex, num_edge} = {1000, 1000} ...
Benchmarking MapGraph.get with input Graph {num_vertex, num_edge} = {1000, 2000} ...
Benchmarking MapGraph.get with input Graph {num_vertex, num_edge} = {1000, 999000} ...
Benchmarking MapGraph.get with input Graph {num_vertex, num_edge} = {5000, 10000} ...
Benchmarking MapGraph.get with input Graph {num_vertex, num_edge} = {5000, 24995000} ...
Benchmarking MapGraph.get with input Graph {num_vertex, num_edge} = {5000, 5000} ...
Benchmarking MapGraph.put with input Graph {num_vertex, num_edge} = {10, 10} ...
Benchmarking MapGraph.put with input Graph {num_vertex, num_edge} = {10, 20} ...
Benchmarking MapGraph.put with input Graph {num_vertex, num_edge} = {10, 90} ...
Benchmarking MapGraph.put with input Graph {num_vertex, num_edge} = {100, 100} ...
Benchmarking MapGraph.put with input Graph {num_vertex, num_edge} = {100, 200} ...
Benchmarking MapGraph.put with input Graph {num_vertex, num_edge} = {100, 9900} ...
Benchmarking MapGraph.put with input Graph {num_vertex, num_edge} = {1000, 1000} ...
Benchmarking MapGraph.put with input Graph {num_vertex, num_edge} = {1000, 2000} ...
Benchmarking MapGraph.put with input Graph {num_vertex, num_edge} = {1000, 999000} ...
Benchmarking MapGraph.put with input Graph {num_vertex, num_edge} = {5000, 10000} ...
Benchmarking MapGraph.put with input Graph {num_vertex, num_edge} = {5000, 24995000} ...
Benchmarking MapGraph.put with input Graph {num_vertex, num_edge} = {5000, 5000} ...
##### With input Graph {num_vertex, num_edge} = {10, 10} #####
Name ips average deviation median 99th %
MapGraph.get 10.63 M 94.05 ns ±25752.82% 83 ns 166 ns
MapGraph.put 4.27 M 234.29 ns ±24652.22% 166 ns 333 ns
Comparison:
MapGraph.get 10.63 M
MapGraph.put 4.27 M - 2.49x slower +140.24 ns
##### With input Graph {num_vertex, num_edge} = {10, 20} #####
Name ips average deviation median 99th %
MapGraph.get 4.44 M 225.10 ns ±61686.85% 83 ns 125 ns
MapGraph.put 2.87 M 348.55 ns ±29631.32% 167 ns 334 ns
Comparison:
MapGraph.get 4.44 M
MapGraph.put 2.87 M - 1.55x slower +123.45 ns
##### With input Graph {num_vertex, num_edge} = {10, 90} #####
Name ips average deviation median 99th %
MapGraph.get 4.01 M 249.12 ns ±57871.76% 83 ns 125 ns
MapGraph.put 3.22 M 310.66 ns ±20983.83% 167 ns 334 ns
Comparison:
MapGraph.get 4.01 M
MapGraph.put 3.22 M - 1.25x slower +61.54 ns
##### With input Graph {num_vertex, num_edge} = {100, 100} #####
Name ips average deviation median 99th %
MapGraph.get 3.81 M 262.67 ns ±55318.78% 84 ns 167 ns
MapGraph.put 3.11 M 321.91 ns ±21244.53% 208 ns 1083 ns
Comparison:
MapGraph.get 3.81 M
MapGraph.put 3.11 M - 1.23x slower +59.25 ns
##### With input Graph {num_vertex, num_edge} = {100, 200} #####
Name ips average deviation median 99th %
MapGraph.put 3.07 M 325.28 ns ±18568.61% 209 ns 375 ns
MapGraph.get 3.03 M 330.42 ns ±42673.45% 125 ns 209 ns
Comparison:
MapGraph.put 3.07 M
MapGraph.get 3.03 M - 1.02x slower +5.14 ns
##### With input Graph {num_vertex, num_edge} = {100, 9900} #####
Name ips average deviation median 99th %
MapGraph.get 6.65 M 150.45 ns ±44246.77% 125 ns 208 ns
MapGraph.put 2.66 M 375.39 ns ±22716.75% 250 ns 500 ns
Comparison:
MapGraph.get 6.65 M
MapGraph.put 2.66 M - 2.50x slower +224.94 ns
##### With input Graph {num_vertex, num_edge} = {1000, 1000} #####
Name ips average deviation median 99th %
MapGraph.get 8.83 M 113.20 ns ±30.55% 125 ns 167 ns
MapGraph.put 2.84 M 351.72 ns ±15344.77% 250 ns 1125 ns
Comparison:
MapGraph.get 8.83 M
MapGraph.put 2.84 M - 3.11x slower +238.52 ns
##### With input Graph {num_vertex, num_edge} = {1000, 2000} #####
Name ips average deviation median 99th %
MapGraph.get 6.37 M 157.11 ns ±46602.02% 125 ns 167 ns
MapGraph.put 2.56 M 391.06 ns ±23783.38% 250 ns 1250 ns
Comparison:
MapGraph.get 6.37 M
MapGraph.put 2.56 M - 2.49x slower +233.95 ns
##### With input Graph {num_vertex, num_edge} = {1000, 999000} #####
Name ips average deviation median 99th %
MapGraph.get 2.06 M 485.63 ns ±29438.06% 250 ns 583 ns
MapGraph.put 1.79 M 557.52 ns ±10558.53% 458 ns 1458 ns
Comparison:
MapGraph.get 2.06 M
MapGraph.put 1.79 M - 1.15x slower +71.89 ns
##### With input Graph {num_vertex, num_edge} = {5000, 10000} #####
Name ips average deviation median 99th %
MapGraph.get 6.28 M 159.29 ns ±44283.03% 125 ns 208 ns
MapGraph.put 2.47 M 405.52 ns ±21239.02% 292 ns 1292 ns
Comparison:
MapGraph.get 6.28 M
MapGraph.put 2.47 M - 2.55x slower +246.23 ns
##### With input Graph {num_vertex, num_edge} = {5000, 24995000} #####
Name ips average deviation median 99th %
MapGraph.put 1.34 M 744.50 ns ±3462.61% 709 ns 1167 ns
MapGraph.get 1.32 M 760.03 ns ±33377.01% 542 ns 875 ns
Comparison:
MapGraph.put 1.34 M
MapGraph.get 1.32 M - 1.02x slower +15.52 ns
##### With input Graph {num_vertex, num_edge} = {5000, 5000} #####
Name ips average deviation median 99th %
MapGraph.get 6.48 M 154.35 ns ±46204.45% 125 ns 208 ns
MapGraph.put 2.64 M 378.96 ns ±26497.26% 250 ns 1250 ns
Comparison:
MapGraph.get 6.48 M
MapGraph.put 2.64 M - 2.46x slower +224.62 ns
- The maximum number of edges when the number of vertexes is 5,000 is 24,995,000.
- The average execution time of
MapGraph.putin this case is 760.03 ns. - Then, total execution time of
MapGraph.putwhen generating the graph in this case is 18,996.9 msec. - But, total execution time of generating the graph in this case is 22,880.225 msec.
- About 83% is the ratio of generating the graph to
MapGraph.put, and about 17% is some overheads. What does the overheads arise from?
VariousMap.EtsGraph is a module for a global graph that can use for a directed graph using ETS.
Run mix run -r bench/ets_graph_bench.exs, then you'll get results similar to the following:
% mix run -r bench/ets_graph_bench.exs
time of create random_edges: 0.431msec.
time of create random_edges: 2.596msec.
time of create random_edges: 566.238msec.
time of create random_edges: 23104.238msec.
time of create EtsGraph: 0.016msec
time of create EtsGraph: 0.009msec
time of create EtsGraph: 0.024msec
time of create EtsGraph: 0.023msec
time of create EtsGraph: 0.044msec
time of create EtsGraph: 2.557msec
time of create EtsGraph: 0.248msec
time of create EtsGraph: 0.516msec
time of create EtsGraph: 419.51msec
time of create EtsGraph: 2.238msec
time of create EtsGraph: 3.499msec
time of create EtsGraph: 21322.16msec
Operating System: macOS
CPU Information: Apple M2
Number of Available Cores: 8
Available memory: 24 GB
Elixir 1.14.0
Erlang 25.0.4
Benchmark suite executing with the following configuration:
warmup: 2 s
time: 5 s
memory time: 0 ns
reduction time: 0 ns
parallel: 1
inputs: Graph {num_vertex, num_edge} = {10, 10}, Graph {num_vertex, num_edge} = {10, 20}, Graph {num_vertex, num_edge} = {10, 90}, Graph {num_vertex, num_edge} = {100, 100}, Graph {num_vertex, num_edge} = {100, 200}, Graph {num_vertex, num_edge} = {100, 9900}, Graph {num_vertex, num_edge} = {1000, 1000}, Graph {num_vertex, num_edge} = {1000, 2000}, Graph {num_vertex, num_edge} = {1000, 999000}, Graph {num_vertex, num_edge} = {5000, 10000}, Graph {num_vertex, num_edge} = {5000, 24995000}, Graph {num_vertex, num_edge} = {5000, 5000}
Estimated total run time: 2.80 min
Benchmarking EtsGraph.get with input Graph {num_vertex, num_edge} = {10, 10} ...
Benchmarking EtsGraph.get with input Graph {num_vertex, num_edge} = {10, 20} ...
Benchmarking EtsGraph.get with input Graph {num_vertex, num_edge} = {10, 90} ...
Benchmarking EtsGraph.get with input Graph {num_vertex, num_edge} = {100, 100} ...
Benchmarking EtsGraph.get with input Graph {num_vertex, num_edge} = {100, 200} ...
Benchmarking EtsGraph.get with input Graph {num_vertex, num_edge} = {100, 9900} ...
Benchmarking EtsGraph.get with input Graph {num_vertex, num_edge} = {1000, 1000} ...
Benchmarking EtsGraph.get with input Graph {num_vertex, num_edge} = {1000, 2000} ...
Benchmarking EtsGraph.get with input Graph {num_vertex, num_edge} = {1000, 999000} ...
Benchmarking EtsGraph.get with input Graph {num_vertex, num_edge} = {5000, 10000} ...
Benchmarking EtsGraph.get with input Graph {num_vertex, num_edge} = {5000, 24995000} ...
Benchmarking EtsGraph.get with input Graph {num_vertex, num_edge} = {5000, 5000} ...
Benchmarking EtsGraph.put with input Graph {num_vertex, num_edge} = {10, 10} ...
Benchmarking EtsGraph.put with input Graph {num_vertex, num_edge} = {10, 20} ...
Benchmarking EtsGraph.put with input Graph {num_vertex, num_edge} = {10, 90} ...
Benchmarking EtsGraph.put with input Graph {num_vertex, num_edge} = {100, 100} ...
Benchmarking EtsGraph.put with input Graph {num_vertex, num_edge} = {100, 200} ...
Benchmarking EtsGraph.put with input Graph {num_vertex, num_edge} = {100, 9900} ...
Benchmarking EtsGraph.put with input Graph {num_vertex, num_edge} = {1000, 1000} ...
Benchmarking EtsGraph.put with input Graph {num_vertex, num_edge} = {1000, 2000} ...
Benchmarking EtsGraph.put with input Graph {num_vertex, num_edge} = {1000, 999000} ...
Benchmarking EtsGraph.put with input Graph {num_vertex, num_edge} = {5000, 10000} ...
Benchmarking EtsGraph.put with input Graph {num_vertex, num_edge} = {5000, 24995000} ...
Benchmarking EtsGraph.put with input Graph {num_vertex, num_edge} = {5000, 5000} ...
##### With input Graph {num_vertex, num_edge} = {10, 10} #####
Name ips average deviation median 99th %
EtsGraph.get 3.80 M 263.22 ns ±6875.80% 208 ns 333 ns
EtsGraph.put 3.23 M 309.50 ns ±7289.78% 291 ns 417 ns
Comparison:
EtsGraph.get 3.80 M
EtsGraph.put 3.23 M - 1.18x slower +46.28 ns
##### With input Graph {num_vertex, num_edge} = {10, 20} #####
Name ips average deviation median 99th %
EtsGraph.get 3.80 M 263.27 ns ±9314.22% 209 ns 334 ns
EtsGraph.put 3.17 M 315.60 ns ±7245.80% 291 ns 417 ns
Comparison:
EtsGraph.get 3.80 M
EtsGraph.put 3.17 M - 1.20x slower +52.33 ns
##### With input Graph {num_vertex, num_edge} = {10, 90} #####
Name ips average deviation median 99th %
EtsGraph.get 3.69 M 270.96 ns ±8590.21% 209 ns 333 ns
EtsGraph.put 3.08 M 324.79 ns ±7095.67% 292 ns 417 ns
Comparison:
EtsGraph.get 3.69 M
EtsGraph.put 3.08 M - 1.20x slower +53.83 ns
##### With input Graph {num_vertex, num_edge} = {100, 100} #####
Name ips average deviation median 99th %
EtsGraph.get 3.64 M 274.76 ns ±8197.40% 209 ns 333 ns
EtsGraph.put 2.76 M 362.55 ns ±4963.25% 333 ns 500 ns
Comparison:
EtsGraph.get 3.64 M
EtsGraph.put 2.76 M - 1.32x slower +87.79 ns
##### With input Graph {num_vertex, num_edge} = {100, 200} #####
Name ips average deviation median 99th %
EtsGraph.get 3.64 M 274.54 ns ±7411.32% 250 ns 334 ns
EtsGraph.put 2.61 M 382.66 ns ±4916.99% 334 ns 541 ns
Comparison:
EtsGraph.get 3.64 M
EtsGraph.put 2.61 M - 1.39x slower +108.12 ns
##### With input Graph {num_vertex, num_edge} = {100, 9900} #####
Name ips average deviation median 99th %
EtsGraph.get 3.44 M 290.87 ns ±7528.34% 250 ns 375 ns
EtsGraph.put 2.47 M 405.19 ns ±5147.82% 375 ns 542 ns
Comparison:
EtsGraph.get 3.44 M
EtsGraph.put 2.47 M - 1.39x slower +114.32 ns
##### With input Graph {num_vertex, num_edge} = {1000, 1000} #####
Name ips average deviation median 99th %
EtsGraph.get 3.67 M 272.81 ns ±6156.71% 250 ns 334 ns
EtsGraph.put 1.61 M 621.70 ns ±1615.55% 583 ns 1083 ns
Comparison:
EtsGraph.get 3.67 M
EtsGraph.put 1.61 M - 2.28x slower +348.90 ns
##### With input Graph {num_vertex, num_edge} = {1000, 2000} #####
Name ips average deviation median 99th %
EtsGraph.get 3.60 M 277.85 ns ±4413.52% 250 ns 375 ns
EtsGraph.put 1.56 M 640.60 ns ±1763.92% 583 ns 1125 ns
Comparison:
EtsGraph.get 3.60 M
EtsGraph.put 1.56 M - 2.31x slower +362.75 ns
##### With input Graph {num_vertex, num_edge} = {1000, 999000} #####
Name ips average deviation median 99th %
EtsGraph.get 1.97 M 507.73 ns ±2888.02% 458 ns 833 ns
EtsGraph.put 1.47 M 680.72 ns ±2795.66% 625 ns 1042 ns
Comparison:
EtsGraph.get 1.97 M
EtsGraph.put 1.47 M - 1.34x slower +173.00 ns
##### With input Graph {num_vertex, num_edge} = {5000, 10000} #####
Name ips average deviation median 99th %
EtsGraph.get 3.52 M 0.28 μs ±5377.75% 0.25 μs 0.38 μs
EtsGraph.put 0.96 M 1.04 μs ±943.51% 1 μs 1.71 μs
Comparison:
EtsGraph.get 3.52 M
EtsGraph.put 0.96 M - 3.68x slower +0.76 μs
##### With input Graph {num_vertex, num_edge} = {5000, 24995000} #####
Name ips average deviation median 99th %
EtsGraph.get 1.65 M 604.38 ns ±2433.01% 500 ns 958 ns
EtsGraph.put 1.26 M 796.30 ns ±2518.92% 708 ns 1208 ns
Comparison:
EtsGraph.get 1.65 M
EtsGraph.put 1.26 M - 1.32x slower +191.92 ns
##### With input Graph {num_vertex, num_edge} = {5000, 5000} #####
Name ips average deviation median 99th %
EtsGraph.get 3.48 M 0.29 μs ±6289.53% 0.25 μs 0.38 μs
EtsGraph.put 0.95 M 1.05 μs ±938.52% 1 μs 1.71 μs
Comparison:
EtsGraph.get 3.48 M
EtsGraph.put 0.95 M - 3.67x slower +0.77 μs
- The maximum number of edges when the number of vertexes is 5,000 is 24,995,000.
- The average execution time of
EtsGraph.putin this case is 796.30 ns. - Then, total execution time of
EtsGraph.putwhen generating the graph in this case is 19,904 msec. - Total execution time of generating the graph in this case is 21,322.16 msec. It is 1.07 times faster than that of
MapGraph. - About 93% is the ratio of generating the graph to
EtsGraph.put, and about 7% is some overheads. This is 2.4 times less than that ofMapGraph.
VariousMap.MnesiaGraph is a module for a global graph that can use for a directed graph using Mnesia.
Run mix run -r bench/mnesia_graph_bench.exs, then you'll get results similar to the following:
% mix run -r bench/mnesia_graph_bench.exs
time of create random_edges: 0.65msec.
time of create random_edges: 2.954msec.
time of create random_edges: 612.574msec.
time of create random_edges: 24972.64msec.
time of create MnesiaGraph: 4.301msec
time of create MnesiaGraph: 1.969msec
time of create MnesiaGraph: 2.784msec
time of create MnesiaGraph: 2.441msec
time of create MnesiaGraph: 3.853msec
time of create MnesiaGraph: 139.017msec
time of create MnesiaGraph: 15.735msec
time of create MnesiaGraph: 29.552msec
time of create MnesiaGraph: 18311.401msec
time of create MnesiaGraph: 91.298msec
time of create MnesiaGraph: 180.053msec
time of create MnesiaGraph: 532464.164msec
Operating System: macOS
CPU Information: Apple M2
Number of Available Cores: 8
Available memory: 24 GB
Elixir 1.14.0
Erlang 25.0.4
Benchmark suite executing with the following configuration:
warmup: 2 s
time: 5 s
memory time: 0 ns
reduction time: 0 ns
parallel: 1
inputs: Graph {num_vertex, num_edge} = {10, 10}, Graph {num_vertex, num_edge} = {10, 20}, Graph {num_vertex, num_edge} = {10, 90}, Graph {num_vertex, num_edge} = {100, 100}, Graph {num_vertex, num_edge} = {100, 200}, Graph {num_vertex, num_edge} = {100, 9900}, Graph {num_vertex, num_edge} = {1000, 1000}, Graph {num_vertex, num_edge} = {1000, 2000}, Graph {num_vertex, num_edge} = {1000, 999000}, Graph {num_vertex, num_edge} = {5000, 10000}, Graph {num_vertex, num_edge} = {5000, 24995000}, Graph {num_vertex, num_edge} = {5000, 5000}
Estimated total run time: 2.80 min
Benchmarking MnesiaGraph.get with input Graph {num_vertex, num_edge} = {10, 10} ...
Benchmarking MnesiaGraph.get with input Graph {num_vertex, num_edge} = {10, 20} ...
Benchmarking MnesiaGraph.get with input Graph {num_vertex, num_edge} = {10, 90} ...
Benchmarking MnesiaGraph.get with input Graph {num_vertex, num_edge} = {100, 100} ...
Benchmarking MnesiaGraph.get with input Graph {num_vertex, num_edge} = {100, 200} ...
Benchmarking MnesiaGraph.get with input Graph {num_vertex, num_edge} = {100, 9900} ...
Benchmarking MnesiaGraph.get with input Graph {num_vertex, num_edge} = {1000, 1000} ...
Benchmarking MnesiaGraph.get with input Graph {num_vertex, num_edge} = {1000, 2000} ...
Benchmarking MnesiaGraph.get with input Graph {num_vertex, num_edge} = {1000, 999000} ...
Benchmarking MnesiaGraph.get with input Graph {num_vertex, num_edge} = {5000, 10000} ...
Benchmarking MnesiaGraph.get with input Graph {num_vertex, num_edge} = {5000, 24995000} ...
Benchmarking MnesiaGraph.get with input Graph {num_vertex, num_edge} = {5000, 5000} ...
Benchmarking MnesiaGraph.put with input Graph {num_vertex, num_edge} = {10, 10} ...
Benchmarking MnesiaGraph.put with input Graph {num_vertex, num_edge} = {10, 20} ...
Benchmarking MnesiaGraph.put with input Graph {num_vertex, num_edge} = {10, 90} ...
Benchmarking MnesiaGraph.put with input Graph {num_vertex, num_edge} = {100, 100} ...
Benchmarking MnesiaGraph.put with input Graph {num_vertex, num_edge} = {100, 200} ...
Benchmarking MnesiaGraph.put with input Graph {num_vertex, num_edge} = {100, 9900} ...
Benchmarking MnesiaGraph.put with input Graph {num_vertex, num_edge} = {1000, 1000} ...
Benchmarking MnesiaGraph.put with input Graph {num_vertex, num_edge} = {1000, 2000} ...
Benchmarking MnesiaGraph.put with input Graph {num_vertex, num_edge} = {1000, 999000} ...
Benchmarking MnesiaGraph.put with input Graph {num_vertex, num_edge} = {5000, 10000} ...
Benchmarking MnesiaGraph.put with input Graph {num_vertex, num_edge} = {5000, 24995000} ...
Benchmarking MnesiaGraph.put with input Graph {num_vertex, num_edge} = {5000, 5000} ...
##### With input Graph {num_vertex, num_edge} = {10, 10} #####
Name ips average deviation median 99th %
MnesiaGraph.put 54.56 K 18.33 μs ±16.47% 17.88 μs 26.33 μs
MnesiaGraph.get 53.58 K 18.66 μs ±25.68% 18.21 μs 27.42 μs
Comparison:
MnesiaGraph.put 54.56 K
MnesiaGraph.get 53.58 K - 1.02x slower +0.34 μs
##### With input Graph {num_vertex, num_edge} = {10, 20} #####
Name ips average deviation median 99th %
MnesiaGraph.get 54.80 K 18.25 μs ±26.60% 17.83 μs 27.63 μs
MnesiaGraph.put 54.76 K 18.26 μs ±17.20% 17.83 μs 26.25 μs
Comparison:
MnesiaGraph.get 54.80 K
MnesiaGraph.put 54.76 K - 1.00x slower +0.0149 μs
##### With input Graph {num_vertex, num_edge} = {10, 90} #####
Name ips average deviation median 99th %
MnesiaGraph.get 54.47 K 18.36 μs ±26.11% 17.96 μs 27 μs
MnesiaGraph.put 54.43 K 18.37 μs ±16.01% 17.92 μs 26.25 μs
Comparison:
MnesiaGraph.get 54.47 K
MnesiaGraph.put 54.43 K - 1.00x slower +0.0117 μs
##### With input Graph {num_vertex, num_edge} = {100, 100} #####
Name ips average deviation median 99th %
MnesiaGraph.put 53.80 K 18.59 μs ±24.04% 18.13 μs 26.83 μs
MnesiaGraph.get 46.86 K 21.34 μs ±54.82% 17.29 μs 64.17 μs
Comparison:
MnesiaGraph.put 53.80 K
MnesiaGraph.get 46.86 K - 1.15x slower +2.75 μs
##### With input Graph {num_vertex, num_edge} = {100, 200} #####
Name ips average deviation median 99th %
MnesiaGraph.get 57.85 K 17.29 μs ±79.87% 16.63 μs 26.13 μs
MnesiaGraph.put 54.30 K 18.42 μs ±16.74% 18 μs 26.46 μs
Comparison:
MnesiaGraph.get 57.85 K
MnesiaGraph.put 54.30 K - 1.07x slower +1.13 μs
##### With input Graph {num_vertex, num_edge} = {100, 9900} #####
Name ips average deviation median 99th %
MnesiaGraph.get 58.99 K 16.95 μs ±30.83% 16.42 μs 24.54 μs
MnesiaGraph.put 54.71 K 18.28 μs ±16.94% 17.79 μs 26.50 μs
Comparison:
MnesiaGraph.get 58.99 K
MnesiaGraph.put 54.71 K - 1.08x slower +1.33 μs
##### With input Graph {num_vertex, num_edge} = {1000, 1000} #####
Name ips average deviation median 99th %
MnesiaGraph.put 53.75 K 18.60 μs ±25.63% 18.08 μs 27.17 μs
MnesiaGraph.get 53.53 K 18.68 μs ±1544.73% 16.38 μs 31.13 μs
Comparison:
MnesiaGraph.put 53.75 K
MnesiaGraph.get 53.53 K - 1.00x slower +0.0763 μs
##### With input Graph {num_vertex, num_edge} = {1000, 2000} #####
Name ips average deviation median 99th %
MnesiaGraph.get 60.40 K 16.56 μs ±30.60% 16.04 μs 24.08 μs
MnesiaGraph.put 53.82 K 18.58 μs ±16.45% 18.08 μs 26.92 μs
Comparison:
MnesiaGraph.get 60.40 K
MnesiaGraph.put 53.82 K - 1.12x slower +2.02 μs
##### With input Graph {num_vertex, num_edge} = {1000, 999000} #####
Name ips average deviation median 99th %
MnesiaGraph.get 59.03 K 16.94 μs ±30.41% 16.42 μs 24.58 μs
MnesiaGraph.put 45.05 K 22.20 μs ±274.81% 18.54 μs 73.96 μs
Comparison:
MnesiaGraph.get 59.03 K
MnesiaGraph.put 45.05 K - 1.31x slower +5.25 μs
##### With input Graph {num_vertex, num_edge} = {5000, 10000} #####
Name ips average deviation median 99th %
MnesiaGraph.get 60.20 K 16.61 μs ±19.91% 16.21 μs 23.79 μs
MnesiaGraph.put 54.23 K 18.44 μs ±26.21% 17.96 μs 26.50 μs
Comparison:
MnesiaGraph.get 60.20 K
MnesiaGraph.put 54.23 K - 1.11x slower +1.83 μs
##### With input Graph {num_vertex, num_edge} = {5000, 24995000} #####
Name ips average deviation median 99th %
MnesiaGraph.get 56.05 K 17.84 μs ±35.85% 17.13 μs 26.79 μs
MnesiaGraph.put 53.50 K 18.69 μs ±48.00% 18.04 μs 27.63 μs
Comparison:
MnesiaGraph.get 56.05 K
MnesiaGraph.put 53.50 K - 1.05x slower +0.85 μs
##### With input Graph {num_vertex, num_edge} = {5000, 5000} #####
Name ips average deviation median 99th %
MnesiaGraph.get 58.86 K 16.99 μs ±25.95% 16.46 μs 24.75 μs
MnesiaGraph.put 53.76 K 18.60 μs ±24.24% 18.13 μs 26.79 μs
Comparison:
MnesiaGraph.get 58.86 K
MnesiaGraph.put 53.76 K - 1.09x slower +1.61 μs
- The maximum number of edges when the number of vertexes is 5,000 is 24,995,000.
- The average execution time of
MnesiaGraph.putin this case is 18.69 μs. - Then, total execution time of
MnesiaGraph.putwhen generating the graph in this case is 467,156.55 msec. - Total execution time of generating the graph in this case is 532,464.164 msec. It is and 23.3 and 25.0 times slower than that of
MapGraphandEtsGraph, respectively. - About 88% is the ratio of generating the graph to
MnesiaGraph.put, and about 12% is some overheads. This is 1.4 times less than that ofMapGraph, and is 1.7 times more than that ofEtsGraph.
Copyright (c) 2022 University of Kitakyushu
Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. You may obtain a copy of the License at http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License.