benchmark_funds.rake_v0

benchmark_funds.rake

desc 'Benchmark funds_raised methods'

ITERATIONS = 100_000

task benchmark_sql: :environment do
  GC.disable
  campaign = Campaign.last

  Benchmark.bm do |bm| 
    bm.report('sql') do
      ITERATIONS.times do
        campaign.funds_raised_sql
        Campaign.connection.clear_query_cache
      end  
    end  
  end  
  GC.enable
end


task benchmark_ruby: :environment do
  GC.disable
  campaign = Campaign.last

  Benchmark.bm do |bm| 
    bm.report('ruby') do
      ITERATIONS.times do
        campaign.funds_raised_ruby
        Campaign.connection.clear_query_cache
      end  
    end  
  end  
  GC.enable
end

I don't think you want to execute the campaign.funds_raised_sql and campaign.funds_raised_ruby methods 100,000 times. I think you want to create 100,000+ records in the database (with something logical like 20-100 loans per campaign) and run those two methods once each, benchmarking their one execution time.

@croaky, sorry for the delay. I didn't see your comment. The data that is behind the benchmark is as you mentioned (with 1000 loans per campaign, actually)

campaigns = FactoryGirl.create_list(:campaign, 100)

campaigns.each do |campaign|
  FactoryGirl.create_list(:loan, 1000, campaign: campaign)
end

The methods get called 100_000 times (perhaps too many times) to throw out any local blips or hiccups in memory, GC, other processes. The difference in time turned out to be so massive that, in the end, I think a single run would have sufficed.

There is an updated gist at https://gist.github.com/srt32/10687302.