» Tag
performance
240 postsMemoryPack: zero-encoding, extreme-performance C# binary serializer
MemoryPack is a zero-encoding .NET 7/C# 11 binary serializer that benchmarks 2-200x faster than existing C# serialization libraries.
Real-Time Audio Synthesis in C#: Inside the Sigilgraph DAW
How a C# engineer built Sigilgraph, an allocation-free, SIMD-powered real-time audio synthesis DAW, disproving myths about managed languages.
LLM Latency Budgets: Speed Up AI Workflows Without Guesswork
A practical guide to building stage-level LLM latency budgets, workflow classes, and runtime enforcement for reliable AI products.
CommitBrief — AI code reviews, right in your terminal
A provider-agnostic, local-first CLI that reviews your staged changes, a historic range, or a whole GitHub pull request. Zero telemetry, no server. Free and open source.
commitbrief.comPHP 8's JIT Compiler: How It Removes the Interpreter Tax
PHP 8's JIT compiles opcodes into native machine code, removing interpreter overhead. This piece covers OPcache, Tracing vs Function JIT, and CRTO tuning.
ILNumerics Lets .NET Array Code Parallelize Itself
ILNumerics Accelerator replaces global static dependency analysis with local runtime checks to auto-parallelize .NET array code, claiming 3x-100x speedups.
Kino: A Rust-powered Ractor web server for Ruby 4.0
Kino brings Ractor-based parallelism to Ruby 4.0 with a Rust core, beating Puma clusters on speed and using up to 7x less memory.
Why TanStack.com Removed React Server Components
TanStack removed React Server Components from tanstack.com, shrinking its markdown/highlight stack and returning to simpler, faster regular SSR.
Headroom: Open-Source Tool Compresses AI Agent Context by Up to 95%
Headroom is an open-source library that compresses AI agent tool outputs by 55-95%, losslessly, via proxy or library mode. 59K GitHub stars.
LLM Agent Performance Is a Distributed Systems Problem
FixBugs's team shows how token math, parallel chunking, and queues cut LLM agent latency: file analysis dropped from 10 minutes to 40 seconds.
Why Distributed Systems Are Often Slower Than One Laptop
The 2015 COST paper showed distributed systems often lose to a single thread. A worked case study reveals the real cost of premature scaling.