llama.cpp/ggml
Daniel Bevenius 515bd7c9a5
cmake : respect GGML_LIB_INSTALL_DIR and LLAMA_LIB_INSTALL_DIR
This commit addresses an issue where the CMake options
`GGML_LIB_INSTALL_DIR` and `LLAMA_LIB_INSTALL_DIR` currently only affect
the generated cmake config files and not the actual installation of the
libraries.

The motivation for this change is that if these options are set, a
project that uses cmake find_package to find ggml will fail to configure:
```console
$ cmake -S . -B build
CMake Error at /work/ai/llama.cpp/install/lib/cmake/ggml/ggml-config.cmake:11 (message):
  File or directory /work/ai/llama.cpp/install/llama.cpp
  referenced by variable GGML_LIB_DIR does not exist !
Call Stack (most recent call first):
  /work/ai/llama.cpp/install/lib/cmake/ggml/ggml-config.cmake:259 (set_and_check)
  CMakeLists.txt:7 (find_package)
```

Resolves: https://github.com/ggml-org/llama.cpp/issues/19748
2026-02-20 08:53:19 +01:00
..
cmake cmake : respect GGML_LIB_INSTALL_DIR and LLAMA_LIB_INSTALL_DIR 2026-02-20 08:53:19 +01:00
include ggml : make `ggml_is_view` as API (#19539) 2026-02-16 17:43:34 +02:00
src ggml-webgpu: Add unary op (SQR, SQRT, SIN, COS) support. (#19700) 2026-02-19 09:18:30 -07:00
.gitignore vulkan : cmake integration (#8119) 2024-07-13 18:12:39 +02:00
CMakeLists.txt cmake : respect GGML_LIB_INSTALL_DIR and LLAMA_LIB_INSTALL_DIR 2026-02-20 08:53:19 +01:00