What a Revenue Headline Leaves Out
Working through Ro’s $3 billion run-rate claim, GLP-1 economics, and what Robin might mean for its membership business.
Writing
7 entries / newest first
Cache Lines, Padding, and the Cost of Trying Again
What padding and x86 PAUSE change in a bounded queue, and what the measured backoff and layout comparisons actually support.
SPSC vs MPSC Queues: Who Owns the Next Slot?
Compare SPSC fan-in and MPSC queues in C++20 with runnable examples, acquire/release ordering, bounded capacity, and measured throughput tradeoffs.
Threads, Mutexes, and Atomics: Start With What Can Go Wrong
How threads share memory, what a mutex actually does on Linux, and why replacing locks with atomics is not automatically an optimization.
Optimizing the Wrong Thing, Faster
AI makes it easier than ever to build impressive-looking systems that are architecturally defeated before they start.
AI and the Illusion of Optimization
AI can accelerate engineering, but it can also create false confidence in bad solutions.
Why I'm Building a Server Framework in Zig
F1 telemetry, low-latency trading, and choosing Zig over C++ and Rust