Use intel mkl instead of openblas

Signed-off-by: baalajimaestro <me@baalajimaestro.me>

Dockerfile

@@ -1,14 +1,26 @@
-FROM python:3-alpine
+FROM intelpython/intelpython3_full
 
-ARG CMAKE_ARGS="-DLLAMA_BLAS=ON -DLLAMA_BLAS_VENDOR=OpenBLAS"
-ARG FORCE_CMAKE=1
+ARG DEBIAN_FRONTEND=noninteractive
 
-RUN apk update && apk add --no-cache cmake make gcc g++ openblas-dev openblas git && \
-	pip install git+https://git.baalajimaestro.me/baalajimaestro/llama.cpp.git fastapi uvicorn sse-starlette pydantic-settings && \
-	apk del make cmake gcc g++ openblas-dev git && \
-	apk add --no-cache libstdc++
+RUN apt update && \
+    apt install -y software-properties-common gnupg apt-utils && \
+    wget -O- https://apt.repos.intel.com/intel-gpg-keys/GPG-PUB-KEY-INTEL-SW-PRODUCTS.PUB | gpg --dearmor | tee /usr/share/keyrings/oneapi-archive-keyring.gpg > /dev/null && \
+    echo "deb [signed-by=/usr/share/keyrings/oneapi-archive-keyring.gpg] https://apt.repos.intel.com/oneapi all main" | tee /etc/apt/sources.list.d/oneAPI.list && \
+    apt update && \
+    apt install intel-oneapi-mkl -y
+
+RUN apt install intel-oneapi-compiler-dpcpp-cpp intel-oneapi-mkl-devel gcc g++ pkg-config cmake -y && \
+    . /opt/intel/oneapi/setvars.sh && \
+    export CMAKE_ARGS="-DLLAMA_BLAS=ON -DLLAMA_BLAS_VENDOR=Intel10_64lp -DCMAKE_C_COMPILER=icx -DCMAKE_CXX_COMPILER=icpx" && \
+    export FORCE_CMAKE=1 && \
+    pip install git+https://git.baalajimaestro.me/baalajimaestro/llama.cpp.git fastapi uvicorn sse-starlette pydantic-settings && \
+    apt remove intel-oneapi-compiler-dpcpp-cpp intel-oneapi-mkl-devel gcc g++ pkg-config cmake -y && \
+    apt autoremove -y
+
+ADD entrypoint.sh /
 
 EXPOSE 8000
 ENV HOST 0.0.0.0
 
-ENTRYPOINT ["python3", "-m", "llama_cpp.server"]
+ENTRYPOINT ["/entrypoint.sh"]
+

entrypoint.sh

@@ -0,0 +1,5 @@
+#! /usr/bin/env bash
+
+source /opt/intel/oneapi/setvars.sh
+python -m llama_cpp.server $@
+