Oossa

llama.cpp b12530 adds WebAssembly SIMD support

The release updates llama.cpp’s WebAssembly kernels and adds support for q3_K and q6_K formats. It also provides downloadable builds for several platforms.

NoteBy Published by Oossa: 1 min read

ggml-org released llama.cpp b12530 with a WebAssembly update: relaxed SIMD kernels were fixed, and support was added for the q3_K and q6_K model formats. The release page also lists downloads for macOS, iOS and several Linux systems.

The release includes CPU builds for Ubuntu x64, arm64 and s390x, among others. The release notes do not say how much the WebAssembly changes affect speed.

Why it matters

Developers building llama.cpp for WebAssembly can use the added q3_K and q6_K support, though the release notes do not quantify any performance change.

Was this article useful?
Share

Read next

Oossa · Newsletter

The week in AI, explained

Every Monday: the stories worth knowing, in plain language. Free, no spam.