martinothamar / coreclr

CoreCLR is the runtime for .NET Core. It includes the garbage collector, JIT compiler, primitive data types and low-level classes.

Hi there 👋, my name is Martin

I am a polyglot programmer and open source enthusiast, working as a consultant at Forse

• I've built
  • Mediator - a fast (close to 0 overhead) sourcegenerator-based mediator pattern implementation in .NET. Mostly MediatR-compatible, but faster
  • Fast.PRNGs.NET - fast PRNG implementations (up to 2.5x faster than System.Random)
  • simd-rand - SIMD implementations of common PRNGs in Rust 🦀 (for example Xoshiro256+ ~60 GiB/s u64 gen using AVX512 on my laptop, about 6x faster than non-vectorized)
  • "building x in y" - experiments and learning about systems programming, programming languages, architectures and designs
    • Simulating football seasons using the Monte Carlo method in Rust - lots of SIMD and performance engineering
    • HTTP servers - experimenting with various techniques and architectures for performent HTTP servers: IO Uring, Linux, thread-per-core...
    • Calculation Engine - calculatione engine with a builder API for creating formulas while doing vectorized calculations over columns of data
• Currently exploring Rust, Zig and Go

I strive to make simple, observable systems - employing mechanical sympathy to build efficient and reliable software.

---

---

Some of my favorite technical content 📖:

• Parse, don’t validate - by Alexis King
• Designing with types: Making illegal states unrepresentable - by Scott Wlaschin
• Anemic Domain Model - by Martin Fowler
• Design Smell: Primitive Obsession - by Mark Seemann/ploeh
• Simple Made Easy - by Rich Hickey
• CUPID - for joful coding - reflections on characteristics and properties vs principles - by Dan North
• SOLID? Nope, just Coupling and Cohesion - by CodeOpinion/Derek Comartin
• Modular Monoliths - by Simon Brown
• Vertical Slice Architecture - by Jimmy Bogard
• TigerStyle! (Or How To Design Safer Systems in Less Time) - by Joran Dirk Greef
  • Doc: https://github.com/tigerbeetle/tigerbeetle/blob/main/docs/TIGER_STYLE.md
• What is Going On With Software? - reflections on abstraction and performance - by Matthew Crews
• Data-Oriented Design and C++ - legendary talk by Mike Acton
• Practical Data Oriented Design - by Andrew Kelley
• Designing for Performance - by Martin Thompson
• How NOT to Measure Latency - by Gil Tene
• Performance Excuses Debunked - by Casey Muratori
• "Clean" Code, Horrible Performance - by Casey Muratori
• Adventures in Performance: Efficiency Analysis of Large-scale Compute - by Thomas Dullien/Halvar Flake