Benchmarking with Snoop

If you have a piece of code for which you’d like to construct a model predictive of its performance, mavkit-snoop is the tool to help you do that. This tool allows to benchmark any given piece of OCaml code and use these measures to fit cost models predictive of execution time.

It is in particular used to derive the functions in the Michelson gas cost API , computing the gas costs in the Mavryk protocol.

Architecture of mavkit-snoop

• Architecture of mavkit-snoop
  • High-level description
  • Code organization
  • Defining benchmarks: the Generator module
  • Defining a predictive model: the Model module
  • The Measure module
  • Computing parameter fits: the Inference module

Using mavkit-snoop by example

• Snoop usage tutorial
  • Step 1: Defining the benchmark
  • Step 2: Running the benchmark, inferring parameters, and generating code
  • Step 3: checking the generated files
  • Step 4: options
  • Using Snoop in the development cycle

mavkit-snoop: going further for more control

• In-depth usage example: more control over your benchmark
  • Step 1: Checking the timer
  • Step 2: Benchmarking
  • Step 2.5: (optional) Removing outliers
  • Step 3: Fitting the model
  • Step 4: Generating code

Rewriting Micheline terms

• Rewriting Micheline with mavryk-micheline-rewriting
  • Constructing Micheline terms
  • Manipulating paths in Micheline terms
  • Pattern-matching Micheline
  • Performing substitutions
  • An example?

Writing your very own benchmarks and models for the Michelson interpreter

• Snoop and the Michelson Interpreter
  • Interpreter Workload
  • Interpreter Models
  • Interpreter benchmarks
  • Testing