Charles Fonseca
@barddoo.bsky.social
Software Engineer. Follower of Jesus.
Married to the prettiest woman.
Making databases in my spare time and occasionally writing about it.
Substack: https://charlesfonseca.substack.com/
Married to the prettiest woman.
Making databases in my spare time and occasionally writing about it.
Substack: https://charlesfonseca.substack.com/
It's astonishing how one can scale a system to thousands of machines, but the cost is often complexity.
Each dependency one doesn't fully comprehend is a potential risk to the system architecture. In most cases, all you need is a VPS with Docker and software you're already familiar with.
Each dependency one doesn't fully comprehend is a potential risk to the system architecture. In most cases, all you need is a VPS with Docker and software you're already familiar with.
October 21, 2025 at 10:16 PM
It's astonishing how one can scale a system to thousands of machines, but the cost is often complexity.
Each dependency one doesn't fully comprehend is a potential risk to the system architecture. In most cases, all you need is a VPS with Docker and software you're already familiar with.
Each dependency one doesn't fully comprehend is a potential risk to the system architecture. In most cases, all you need is a VPS with Docker and software you're already familiar with.
Hashing function: initially, I believed that FNV-1a would perform better than all other algorithms, but CityHash64 was the best.
Results: gist.github.com/barddoo/0a24...
Results: gist.github.com/barddoo/0a24...
October 15, 2025 at 8:41 PM
Hashing function: initially, I believed that FNV-1a would perform better than all other algorithms, but CityHash64 was the best.
Results: gist.github.com/barddoo/0a24...
Results: gist.github.com/barddoo/0a24...
I benchmarked to find the best hashing and equality function for the most common cache workload.
TL;DR:
Equality checker: You won't beat the standard library. Starting at 40 bytes, std.mem.eql outperforms all SIMD implementations.
Results: gist.github.com/barddoo/e37e...
TL;DR:
Equality checker: You won't beat the standard library. Starting at 40 bytes, std.mem.eql outperforms all SIMD implementations.
Results: gist.github.com/barddoo/e37e...
October 15, 2025 at 8:41 PM
I benchmarked to find the best hashing and equality function for the most common cache workload.
TL;DR:
Equality checker: You won't beat the standard library. Starting at 40 bytes, std.mem.eql outperforms all SIMD implementations.
Results: gist.github.com/barddoo/e37e...
TL;DR:
Equality checker: You won't beat the standard library. Starting at 40 bytes, std.mem.eql outperforms all SIMD implementations.
Results: gist.github.com/barddoo/e37e...