{
"$type": "site.standard.document",
"bskyPostRef": {
"cid": "bafyreibnadzsw23eu77h6eixu3itjyidh22ntkgfqr7salpcyyjsxmlu3i",
"uri": "at://did:plc:7fzsvgbjbyp2vxleodbcwyif/app.bsky.feed.post/3mpf3myn3cbl2"
},
"path": "/2026/06/28/cpu-optimized-container-images.html",
"publishedAt": "2026-06-28T00:00:00.000Z",
"site": "https://jens.dev",
"tags": [
"earlier post"
],
"textContent": "In an earlier post about gguf-runner I made a point that kept nagging at me afterwards.\n\nThe point was this: prebuilt binaries are compiled for a conservative baseline so they run everywhere, and if you build locally with `target-cpu=native` instead, the compiler can light up the SIMD instructions your specific CPU supports. On an AMD Ryzen 7 PRO 8700GE that was worth about **+20.8% tokens/sec** for LLM inference. Real, measurable, free.",
"title": "Shipping CPU-optimized Rust binaries in container images"
}