Dockerfile.cuda

# --------------------------------------------------------------
# Copyright (c) AutoDeployAI. All rights reserved.
# Licensed under the Apache License, Version 2.0 (the "License").
# --------------------------------------------------------------

FROM nvcr.io/nvidia/cuda:12.1.1-cudnn8-runtime-ubuntu22.04
MAINTAINER AutoDeployAI "autodeploy.ai@gmail.com"

ADD . /code

ENV DEBIAN_FRONTEND=noninteractive

RUN apt-get update &&\
  apt-get install -y --no-install-recommends \
  wget \
  openjdk-11-jdk \
  ca-certificates-java 

ARG SBT_VERSION=1.6.2
RUN  wget -q -O /tmp/sbt-${SBT_VERSION}.tgz https://github.com/sbt/sbt/releases/download/v${SBT_VERSION}/sbt-${SBT_VERSION}.tgz &&\
  tar -zxf /tmp/sbt-${SBT_VERSION}.tgz --strip=1 -C /usr

# Harvest both PMML4S and ONNXRuntime(GPU) from maven central
RUN cd /code &&\
  sbt -Dgpu=true assembly


FROM nvcr.io/nvidia/cuda:12.1.1-cudnn8-runtime-ubuntu22.04

WORKDIR /ai-serving

ENV \
  LANG=C.UTF-8 \
  JAVA_HOME=/usr/lib/jvm/java-11-openjdk-amd64

RUN apt-get update &&\
  apt-get install -y libgomp1 openjdk-11-jdk ca-certificates-java &&\
  rm -rf /var/lib/apt/lists/*

COPY --from=0 /code/target/scala-2.13/*.jar .

RUN ln -s ai-serving-cuda-assembly-*.jar ai-serving-cuda-assembly.jar

ENTRYPOINT exec java $JAVA_OPTS --add-exports java.base/jdk.internal.math=ALL-UNNAMED -Donnxruntime.backend=cuda -jar ai-serving-cuda-assembly.jar