bench: add bench_string_key_in_map_vs_string_value_in_array.v

vlib/v/tests/bench/bench_string_key_in_map_vs_string_value_in_array.v

@@ -0,0 +1,70 @@
+import os
+import rand
+import benchmark
+
+// Purpose: this script measures the performance of common lookup operations on the builtin map, vs the builtin array type in V.
+// Usage:
+//   MAX_ITEMS=1000 MAX_ITERATIONS=1_000_000 v -prod -cc gcc-11 crun vlib/v/tests/bench/bench_string_key_in_map_vs_string_value_in_array.v
+//
+// Typical result on Intel i3, Ubuntu 20.04:
+//
+// > max_iterations: 1000000 over max_items: 1000
+// > m.len: 1001 | a_as_first.len: 1001 | a_as_last.len: 1001 | a_as_middle.len: 1001
+// SPENT    17.344 ms in sum: 1000000 | m[key]
+// SPENT    13.402 ms in sum: 1000000 | key in m
+// SPENT   643.429 ms in sum: 1000000 | key in a_as_last
+// SPENT   319.897 ms in sum: 1000000 | key in a_as_middle
+// SPENT     1.842 ms in sum: 1000000 | key in a_as_first
+
+max_iterations := os.getenv_opt('MAX_ITERATIONS') or { '1000' }.int()
+max_items := os.getenv_opt('MAX_ITEMS') or { '100' }.int()
+eprintln('> max_iterations: ${max_iterations} over max_items: ${max_items}')
+key := 'abc'
+mut sum := 0
+mut m := map[string]bool{}
+m[key] = true
+for _ in 0 .. max_items {
+	m[rand.string(20)] = true
+}
+a_as_first := m.keys()
+a_as_last := m.keys().reverse()
+mut a_as_middle := a_as_first[1..a_as_first.len / 2]
+a_as_middle << key
+a_as_middle << a_as_first[a_as_first.len / 2..]
+eprintln('> m.len: ${m.len} | a_as_first.len: ${a_as_first.len} | a_as_last.len: ${a_as_last.len} | a_as_middle.len: ${a_as_middle.len}')
+
+mut b := benchmark.start()
+//
+sum = 0
+for _ in 0 .. max_iterations {
+	sum += int(m[key])
+}
+b.measure('sum: ${sum} | m[key]')
+
+sum = 0
+for _ in 0 .. max_iterations {
+	sum += int(key in m)
+}
+b.measure('sum: ${sum} | key in m')
+
+//
+b = benchmark.start()
+
+//
+sum = 0
+for _ in 0 .. max_iterations {
+	sum += int(key in a_as_last)
+}
+b.measure('sum: ${sum} | key in a_as_last')
+
+sum = 0
+for _ in 0 .. max_iterations {
+	sum += int(key in a_as_middle)
+}
+b.measure('sum: ${sum} | key in a_as_middle')
+
+sum = 0
+for _ in 0 .. max_iterations {
+	sum += int(key in a_as_first)
+}
+b.measure('sum: ${sum} | key in a_as_first')