Show HN: Open-source engine running Gemma 4 26B in 2 GB RAM on any M-series Mac

Re:Show HN: Open-source engine running Gemma 4 26B in 2 GB R... by a-dub on Wednesday July 29, 2026 @12:47PM (#49099886)

for a given expert, do you have a sense for what the spatiotemporal access pattern looks like?

[ Parent | on HN ]

Re:Show HN: Open-source engine running Gemma 4 26B in 2 GB R... by gitpusher42 on Wednesday July 29, 2026 @05:59PM (#49103630)

Yeah, I checked it. One expert is about a 3.36mb block. If a cache miss happens I read whole block with one pread.

And there is some reuse. ~41% selected again for the next token, ~57% within two. Each layer has its own experts, so no reuse between these layers.

[ Parent | on HN ]