2026-11-13 · America/Los_Angeles · 社区动态 · #3
I built speedy Strix Halo kernels for the top-51 LLMs with AMD Loom (and you can too)
I’ve been trying out a new experimental software flow from AMD’s AI Group, and I’ve been having a lot of fun with it, so I thought I’d share and see what y’all think too. hrx-loom is a new end-to-end tool stack (compiler, runtime, serving tools) that lets anyone build, optimize, serve, and share model programs without any modifications to the inference engine or server. The kernels are written in .loom files, a vector programming model reminiscent of MLIR, which can be complex for humans but easy for agents (C++ is also supported). Over the past couple weeks, I developed an agent skill that…
热度 48.4 / 100;排名与评分保留该期记录。