Hi,
On 2026-03-10 15:28, Matt Corallo wrote:
> I did local builds with -DGGML_RPC=ON in both ggml and llama.cpp's rules
> file, renamed rpc-server to llama-rpc-server in llama.cpp-tools-
> extra.install and added usr/lib/${DEB_HOST_MULTIARCH}/ggml/backends0/
> libggml-rpc.so to libggml0.install and it seems to work great.
Some progress:
The ggml RPC backend is now available as libggml0-backend-rpc.
The frontend to this backend is shipped by llama.cpp-tools-extra, under
the path
/usr/libexec/x86_64-linux-gnu/llama/bin/ggml-rpc-server
(Replace x86_64-linux-gnu with your host's multi-arch triplet if you're
not on amd64.)
A few things to note:
* Upstream wanted ggml-rpc-server because there is no tie to
llama.cpp, other than being shipped as part of that project
* Upstream will eventually move the binary to ggml
* Once the way forward has been resolved, we'll ship this in some
new package, with a dependency on libggml-backend-rpc.
Best,
Christian
