DEV · hot matching “gemma”

A field report on serving Gemma 4 E2B under vLLM on AWS G5g — the only aarch64 + SM 7.5 hardware there is. No published build covers that combination, AWS quietly solves half of it, and the thing that actually blocks you is 64 KiB of shared memory.

I'd been waiting for more than 30 minutes. The terminal just sat there, blinking, without returning a...