My working unpac space for OCaml projects in development
v simple bench
+141
+134
bench/bench_zstd.ml
+134
bench/bench_zstd.ml
···
1
+
(** Benchmarks for pure OCaml zstd implementation.
2
+
3
+
Run with: dune exec bench/bench_zstd.exe -- -quota 5
4
+
5
+
To compare with C zstd, run:
6
+
python3 -c "print('A' * 102400, end='')" > /tmp/rep100k.bin
7
+
zstd -b1 -e9 -i5 /tmp/rep100k.bin
8
+
*)
9
+
10
+
open Core
11
+
open Core_bench
12
+
13
+
(** Generate test data with different characteristics *)
14
+
module Test_data = struct
15
+
(** Highly compressible: repeated pattern *)
16
+
let repetitive size =
17
+
String.init size ~f:(fun i -> Char.of_int_exn ((i mod 26) + 65))
18
+
19
+
(** Medium compressibility: English-like text with spaces *)
20
+
let text_like size =
21
+
let words = [| "the "; "quick "; "brown "; "fox "; "jumps "; "over ";
22
+
"lazy "; "dog "; "and "; "runs "; "now " |] in
23
+
let buf = Buffer.create size in
24
+
let i = ref 0 in
25
+
while Buffer.length buf < size do
26
+
Buffer.add_string buf words.(!i mod Array.length words);
27
+
incr i
28
+
done;
29
+
String.sub (Buffer.contents buf) ~pos:0 ~len:size
30
+
31
+
(** Binary-like: sequential bytes with variation *)
32
+
let binary_like size =
33
+
String.init size ~f:(fun i -> Char.of_int_exn ((i / 4) mod 256))
34
+
end
35
+
36
+
(** Benchmark decompression *)
37
+
let decompression_benchmarks data_name data =
38
+
let size = String.length data in
39
+
(* Pre-compress at level 3 *)
40
+
let compressed = Zstd.compress ~level:3 data in
41
+
let ratio = Float.of_int (String.length compressed) /. Float.of_int size *. 100.0 in
42
+
[
43
+
Bench.Test.create
44
+
~name:(sprintf "decompress/%s/%dKB(%.0f%%)" data_name (size / 1024) ratio)
45
+
(fun () -> ignore (Zstd.decompress_exn compressed : string));
46
+
]
47
+
48
+
(** Benchmark roundtrip *)
49
+
let roundtrip_benchmarks data_name data =
50
+
let size = String.length data in
51
+
[
52
+
Bench.Test.create
53
+
~name:(sprintf "roundtrip/%s/%dKB" data_name (size / 1024))
54
+
(fun () ->
55
+
let compressed = Zstd.compress ~level:3 data in
56
+
ignore (Zstd.decompress_exn compressed : string));
57
+
]
58
+
59
+
(** Print compression ratio summary *)
60
+
let print_compression_ratios () =
61
+
let sizes = [1024; 10240; 102400] in
62
+
let data_types = [
63
+
("repetitive", Test_data.repetitive);
64
+
("text", Test_data.text_like);
65
+
("binary", Test_data.binary_like);
66
+
] in
67
+
68
+
printf "\n=== Compression Ratios (size as %% of original) ===\n";
69
+
printf "%-12s %8s %8s %8s %8s %8s\n"
70
+
"Data Type" "Size" "L1" "L3" "L6" "L9";
71
+
printf "%s\n" (String.make 56 '-');
72
+
73
+
List.iter sizes ~f:(fun size ->
74
+
List.iter data_types ~f:(fun (name, gen) ->
75
+
let data = gen size in
76
+
let ratios = List.map [1; 3; 6; 9] ~f:(fun level ->
77
+
let compressed = Zstd.compress ~level data in
78
+
Float.of_int (String.length compressed) /. Float.of_int size *. 100.0
79
+
) in
80
+
printf "%-12s %7dB " name size;
81
+
List.iter ratios ~f:(fun r -> printf "%7.1f%% " r);
82
+
printf "\n"
83
+
)
84
+
);
85
+
printf "\n"
86
+
87
+
let () =
88
+
(* Print compression ratios first *)
89
+
print_compression_ratios ();
90
+
91
+
(* Generate test data at different sizes *)
92
+
let sizes = [
93
+
(1, "1KB");
94
+
(10, "10KB");
95
+
(100, "100KB");
96
+
] in
97
+
98
+
let all_benchmarks =
99
+
List.concat_map sizes ~f:(fun (kb, _) ->
100
+
let size = kb * 1024 in
101
+
102
+
(* Different data types *)
103
+
let repetitive = Test_data.repetitive size in
104
+
let text_like = Test_data.text_like size in
105
+
let binary_like = Test_data.binary_like size in
106
+
107
+
List.concat [
108
+
(* Compression benchmarks - only level 3 for each type/size *)
109
+
[Bench.Test.create
110
+
~name:(sprintf "compress/rep/%dKB" kb)
111
+
(fun () -> ignore (Zstd.compress ~level:3 repetitive : string))];
112
+
[Bench.Test.create
113
+
~name:(sprintf "compress/text/%dKB" kb)
114
+
(fun () -> ignore (Zstd.compress ~level:3 text_like : string))];
115
+
[Bench.Test.create
116
+
~name:(sprintf "compress/bin/%dKB" kb)
117
+
(fun () -> ignore (Zstd.compress ~level:3 binary_like : string))];
118
+
119
+
(* Decompression benchmarks *)
120
+
decompression_benchmarks "rep" repetitive;
121
+
decompression_benchmarks "text" text_like;
122
+
decompression_benchmarks "bin" binary_like;
123
+
124
+
(* Roundtrip benchmarks *)
125
+
roundtrip_benchmarks "rep" repetitive;
126
+
roundtrip_benchmarks "text" text_like;
127
+
roundtrip_benchmarks "bin" binary_like;
128
+
]
129
+
)
130
+
in
131
+
132
+
printf "Running %d benchmarks...\n" (List.length all_benchmarks);
133
+
printf "Use -quota N to set seconds per benchmark (default 10).\n\n";
134
+
Command_unix.run (Bench.make_command all_benchmarks)