Skip to main content

memory_allocation_tracking

🧠 Track Memory Allocation and GC Impact​

Microbenchmarks often ignore memory churn, but large heaps or frequent GCs can kill performance. By pairing Benchmark.realtime with GC.stat and ObjectSpace, you can correlate time cost with allocation patterns and tune both code and GC settings.

require 'benchmark'

# Warm up and clear heap for consistent stats
GC.start(full_mark: true, immediate_sweep: true)

before_stats = GC.stat
before_count = ObjectSpace.each_object(String).count

time = Benchmark.realtime do
  100_000.times { "x" * 100 }
end

after_stats = GC.stat
after_count = ObjectSpace.each_object(String).count

puts "Time: #{(time*1000).round(2)}ms"
puts "Heap live slots Δ: #{after_stats[:heap_live_slots] - before_stats[:heap_live_slots]}"
puts "String objects Δ: #{after_count - before_count}"