change compilation flags

Dockerfile

@@ -5,9 +5,11 @@ RUN apt-get update && apt-get install -y --no-install-recommends \
     git \
     build-essential \
     cmake \
+    ninja-build \
     ca-certificates \
     libcurl4-openssl-dev \
     libssl-dev \
+    libopenblas-dev \
     pkg-config \
     && rm -rf /var/lib/apt/lists/*
 
@@ -16,15 +18,20 @@ WORKDIR /build
 ARG CACHEBUST=6
 RUN git clone https://github.com/ggerganov/llama.cpp.git && \
     cd llama.cpp && \
-    cmake -B build -DCMAKE_BUILD_TYPE=Release \
+    cmake -B build -G Ninja -DCMAKE_BUILD_TYPE=Release \
+        -DCMAKE_C_FLAGS_RELEASE="-O3 -DNDEBUG" \
+        -DCMAKE_CXX_FLAGS_RELEASE="-O3 -DNDEBUG" \
         -DGGML_NATIVE=OFF \
-        -DGGML_AVX2=OFF \
-        -DGGML_AVX=OFF \
-        -DGGML_FMA=OFF \
-        -DGGML_F16C=OFF \
+        -DGGML_AVX2=ON \
+        -DGGML_AVX=ON \
+        -DGGML_FMA=ON \
+        -DGGML_F16C=ON \
+        -DGGML_OPENMP=ON \
+        -DGGML_BLAS=ON \
+        -DGGML_BLAS_VENDOR=OpenBLAS \
         -DLLAMA_CURL=ON \
         -DLLAMA_OPENSSL=ON && \
-    cmake --build build --config Release --target llama-server -j1 && \
+    cmake --build build --config Release --target llama-server -j"$(nproc)" && \
     echo "=== Binary dependencies ===" && \
     ldd build/bin/llama-server || true
 
@@ -36,6 +43,7 @@ RUN apt-get update && apt-get install -y --no-install-recommends \
     libcurl4 \
     ca-certificates \
     libgomp1 \
+    libopenblas0-pthread \
     libstdc++6 \
     openssl \
     && rm -rf /var/lib/apt/lists/*
@@ -89,7 +97,7 @@ EXPOSE 7860
 #
 # Example Qwen2.5 7B Instruct (8k context, safer on 10GB RAM):
 # Verified llama.cpp args from model card.
-CMD ["llama-server", "--hf-repo", "paultimothymooney/Qwen2.5-7B-Instruct-Q4_K_M-GGUF", "--hf-file", "qwen2.5-7b-instruct-q4_k_m.gguf", "--host", "0.0.0.0", "--port", "7860", "-c", "4096", "-t", "4", "-ngl", "0", "--cont-batching", "-b", "128", "--ubatch-size", "64"]
+CMD ["llama-server", "--hf-repo", "paultimothymooney/Qwen2.5-7B-Instruct-Q4_K_M-GGUF", "--hf-file", "qwen2.5-7b-instruct-q4_k_m.gguf", "--host", "0.0.0.0", "--port", "7860", "-c", "8192", "-t", "4", "-ngl", "0", "--cont-batching", "-b", "128", "--ubatch-size", "64"]
 #
 # Example Qwen2.5-Coder 7B Instruct (32k context):
 # CMD ["llama-server", "-hf", "Qwen/Qwen2.5-Coder-7B-Instruct-GGUF:Qwen2.5-Coder-7B-Instruct.Q4_K_M.gguf",