Git commit

413977d

Operating systems

Linux

GGML backends

BLAS

Problem description & steps to reproduce

First, I followed the blas backend doc to cmake the llama.cpp.
I found that there truely have found the blas library.

And also built the blas target in cmake.

However, when I tried to run the llama2-7b model, the system info doesn't have the blas info.
And the backend is used the CPU.

If the model will influence the backend choosen in llama.cpp?

First Bad Commit

No response

Compile command

cmake -B build -DGGML_BLAS=ON -DGGML_BLAS_VENDOR=OpenBLAS
cmake --build build --config Release

Relevant log output

system_info: n_threads = 8 (n_threads_batch = 8) / CPU : SSE3 = 1 | OPENMAP = 1 | REPACK = 1|