|
|
@@ -242,6 +242,23 @@ In order to build llama.cpp you have three different options.
|
|
|
zig build -Doptimize=ReleaseFast
|
|
|
```
|
|
|
|
|
|
+- Using `gmake` (FreeBSD):
|
|
|
+
|
|
|
+ 1. Install and activate [DRM in FreeBSD](https://wiki.freebsd.org/Graphics)
|
|
|
+ 2. Add your user to **video** group
|
|
|
+ 3. Install compilation dependencies.
|
|
|
+
|
|
|
+ ```bash
|
|
|
+ sudo pkg install gmake automake autoconf pkgconf llvm15 clinfo clover \
|
|
|
+ opencl clblast openblas
|
|
|
+
|
|
|
+ gmake CC=/usr/local/bin/clang15 CXX=/usr/local/bin/clang++15 -j4
|
|
|
+ ```
|
|
|
+
|
|
|
+ **Notes:** With this packages you can build llama.cpp with OPENBLAS and
|
|
|
+ CLBLAST support for use OpenCL GPU acceleration in FreeBSD. Please read
|
|
|
+ the instructions for use and activate this options in this document below.
|
|
|
+
|
|
|
### Metal Build
|
|
|
|
|
|
Using Metal allows the computation to be executed on the GPU for Apple devices:
|