address feedback

.github/workflows/benchmarks.yml

@@ -20,7 +20,7 @@ jobs:
     - name: Set up Ruby
       uses: ruby/setup-ruby@v1
       with:
-        ruby-version: 3.2
+        ruby-version: 3.4
         bundler-cache: true # 'bundle install' and cache
     - name: Run Benchmarks
       run: RUBY_YJIT_ENABLE=1 BENCH_TARGET=all bundle exec bin/benchmark

.github/workflows/profile.yml

@@ -23,6 +23,9 @@ jobs:
         bundler-cache: true # 'bundle install' and cache
     - name: Run Profiles
       run: RUBY_YJIT_ENABLE=1 BENCH_TARGET=all bundle exec bin/profile
+    # NOTE: to pull profile results, visit https://github.com/petergoldstein/dalli/actions/workflows/profile.yml
+    # click to view the run you are interested in (ex https://github.com/petergoldstein/dalli/actions/runs/13296952241)
+    # in the artifacts section, download the profile results
     - name: Upload profile results
       uses: actions/upload-artifact@v4
       with:
@@ -36,3 +39,7 @@ jobs:
           socket_get_multi_profile.json
           client_set_multi_profile.json
           socket_set_multi_profile.json
+          meta_client_get_multi_profile.json
+          meta_client_get_profile.json
+          meta_client_set_multi_profile.json
+          meta_client_set_profile.json

bin/benchmark

@@ -2,28 +2,28 @@
 # frozen_string_literal: true
 
 # This helps benchmark current performance of Dalli
-# as well as compare performance of optimizated and non-optimized calls like multi-set vs set
+# as well as compare performance of optimized and non-optimized calls like multi-set vs set
 #
 # run with:
 # bundle exec bin/benchmark
 # RUBY_YJIT_ENABLE=1 BENCH_TARGET=get bundle exec bin/benchmark
 require 'bundler/inline'
-require 'json'
 
 gemfile do
   source 'https://rubygems.org'
   gem 'benchmark-ips'
   gem 'logger'
 end
 
-require_relative '../lib/dalli'
+require 'json'
 require 'benchmark/ips'
 require 'monitor'
+require_relative '../lib/dalli'
 
 ##
-# StringSerializer is a serializer that avoids the overhead of Marshal or JSON.
+# NoopSerializer is a serializer that avoids the overhead of Marshal or JSON.
 ##
-class StringSerializer
+class NoopSerializer
   def self.dump(value)
     value
   end
@@ -34,7 +34,7 @@ class StringSerializer
 end
 
 dalli_url = ENV['BENCH_CACHE_URL'] || '127.0.0.1:11211'
-bench_target = ENV['BENCH_TARGET'] || 'set'
+bench_target = ENV['BENCH_TARGET'] || 'all'
 bench_time = (ENV['BENCH_TIME'] || 10).to_i
 bench_warmup = (ENV['BENCH_WARMUP'] || 3).to_i
 bench_payload_size = (ENV['BENCH_PAYLOAD_SIZE'] || 700_000).to_i
@@ -54,7 +54,7 @@ sock.setsockopt(Socket::IPPROTO_TCP, Socket::TCP_NODELAY, true)
 sock.setsockopt(Socket::SOL_SOCKET, Socket::SO_KEEPALIVE, true)
 # Benchmarks didn't see any performance gains from increasing the SO_RCVBUF buffer size
 # sock.setsockopt(Socket::SOL_SOCKET, ::Socket::SO_RCVBUF, 1024 * 1024 * 8)
-# Benchamrks did see an improvement in performance when increasing the SO_SNDBUF buffer size
+# Benchmarks did see an improvement in performance when increasing the SO_SNDBUF buffer size
 # sock.setsockopt(Socket::SOL_SOCKET, Socket::SO_SNDBUF, 1024 * 1024 * 8)
 
 # ensure the clients are all connected and working

bin/profile

@@ -7,7 +7,6 @@
 # run with:
 # RUBY_YJIT_ENABLE=1 bundle exec bin/profile
 require 'bundler/inline'
-require 'json'
 
 gemfile do
   source 'https://rubygems.org'
@@ -16,14 +15,15 @@ gemfile do
   gem 'logger'
 end
 
-require_relative '../lib/dalli'
+require 'json'
 require 'benchmark/ips'
 require 'vernier'
+require_relative '../lib/dalli'
 
 ##
-# StringSerializer is a serializer that avoids the overhead of Marshal or JSON.
+# NoopSerializer is a serializer that avoids the overhead of Marshal or JSON.
 ##
-class StringSerializer
+class NoopSerializer
   def self.dump(value)
     value
   end
@@ -34,8 +34,8 @@ class StringSerializer
 end
 
 dalli_url = ENV['BENCH_CACHE_URL'] || '127.0.0.1:11211'
-bench_target = ENV['BENCH_TARGET'] || 'get'
-bench_time = (ENV['BENCH_TIME'] || 10).to_i
+bench_target = ENV['BENCH_TARGET'] || 'all'
+bench_time = (ENV['BENCH_TIME'] || 8).to_i
 bench_payload_size = (ENV['BENCH_PAYLOAD_SIZE'] || 700_000).to_i
 TERMINATOR = "\r\n"
 puts "yjit: #{RubyVM::YJIT.enabled?}"
@@ -50,7 +50,7 @@ sock.setsockopt(Socket::IPPROTO_TCP, Socket::TCP_NODELAY, true)
 sock.setsockopt(Socket::SOL_SOCKET, Socket::SO_KEEPALIVE, true)
 # Benchmarks didn't see any performance gains from increasing the SO_RCVBUF buffer size
 # sock.setsockopt(Socket::SOL_SOCKET, ::Socket::SO_RCVBUF, 1024 * 1024 * 8)
-# Benchamrks did see an improvement in performance when increasing the SO_SNDBUF buffer size
+# Benchmarks did see an improvement in performance when increasing the SO_SNDBUF buffer size
 # sock.setsockopt(Socket::SOL_SOCKET, Socket::SO_SNDBUF, 1024 * 1024 * 8)
 
 payload = 'B' * bench_payload_size