Benchmark.ips: show allocated bytes per operation (#5522)

crystal-lang · Jan 4, 2018 · ca555d2 · ca555d2
1 parent aaf7d96
commit ca555d2
Showing 1 changed file with 14 additions and 1 deletion.
diff --git a/src/benchmark/ips.cr b/src/benchmark/ips.cr
@@ -43,13 +43,15 @@ module Benchmark
       def report
         max_label = ran_items.max_of &.label.size
         max_compare = ran_items.max_of &.human_compare.size
+        max_bytes_per_op = ran_items.max_of &.bytes_per_op.to_s.size
 
         ran_items.each do |item|
-          printf "%s %s (%s) (±%5.2f%%) %s\n",
+          printf "%s %s (%s) (±%5.2f%%)  %s B/op  %s\n",
             item.label.rjust(max_label),
             item.human_mean,
             item.human_iteration_time,
             item.relative_stddev,
+            item.bytes_per_op.to_s.rjust(max_bytes_per_op),
             item.human_compare.rjust(max_compare)
         end
       end
@@ -79,17 +81,25 @@ module Benchmark
           GC.collect
 
           measurements = [] of Time::Span
+          bytes = 0_i64
+          cycles = 0_i64
+
           target = Time.monotonic + @calculation_time
 
           loop do
+            bytes_before_measure = GC.stats.total_bytes
             elapsed = Time.measure { item.call_for_100ms }
+            bytes += (GC.stats.total_bytes - bytes_before_measure).to_i64
+            cycles += item.cycles
             measurements << elapsed
             break if Time.monotonic >= target
           end
 
           ips = measurements.map { |m| item.cycles.to_f / m.total_seconds }
           item.calculate_stats(ips)
 
+          item.bytes_per_op = (bytes.to_f / cycles.to_f).round.to_i
+
           if @interactive
             run_comparison
             report
@@ -139,6 +149,9 @@ module Benchmark
       # Multiple slower than the fastest entry
       property! slower : Float64
 
+      # Number of bytes allocated per operation
+      property! bytes_per_op : Int32
+
       @ran : Bool
       @ran = false