First project

• Project: Profiling ethereumjs-vm
• Goal: Identify performance optimisation opportunities
• Implemented using ganache-core and synthetic benchmarks
• Repository: https://github.com/EthWorks/ethereumjs-vm-benchmarks
• Results: https://github.com/EthWorks/ethereumjs-vm-benchmarks/blob/master/results/results.md
• Takeaways:
  • Some low-hanging fruit optimisations were identified
  • Identified Merkle Patricia Tree (MPT) as the major performance sink
    • Lots of complexity to expose features without a clear need for them
    • An alternative implementation was proposed
  • Exposed the difficulty to profile the VM because of MPT's complexity

Second project

• Project: Alternative MPT implementation performance analysis
• Goals: Understand how big the performance improvement of implementing the alternative MPT would be
• Side goals:
  • Simplify MPT's code, making ethereumjs-vm easier to profile
  • Create an initial set of benchmarks that can be used to track performance changes in ethereumjs-vm
• Non-goals:
  • Create a production-ready alternative implementation of MPT
  • Integrate the benchmarks into EthereumJS' development workflow

This project had three different phases, with the results of each of the leading to the next one

Phase 1: Alternative MPT implementation and MPT benchmarking

• Description: The proposed alternative MPT was implemented and was benchmarked against the pervious versions.