Updating toolboxes with a two-stage build process to reduce size

toolboxes/Dockerfile.rocm-7beta

@@ -1,7 +1,8 @@
-FROM fedora:rawhide
+# build stage
+FROM registry.fedoraproject.org/fedora:rawhide AS builder
 
-# getting Rocm7 repos
-RUN <<EOF
+# rocm 7.0 repos
+RUN <<'EOF'
 tee /etc/yum.repos.d/rocm.repo <<REPO
 [ROCm-7.0.0]
 name=ROCm7.0.0
@@ -21,49 +22,92 @@ priority=50
 gpgcheck=1
 gpgkey=https://repo.radeon.com/rocm/rocm.gpg.key
 REPO
-
 EOF
 
-# Install build dependencies and tools
-RUN dnf install -y \
+# deps
+RUN dnf -y --nodocs --setopt=install_weak_deps=False install \
        make gcc cmake lld clang clang-devel compiler-rt libcurl-devel \
        rocm rocminfo radeontop \
        git vim \
-    && dnf clean all
+    && dnf clean all && rm -rf /var/cache/dnf/*
 
-# Set up working directory
+# llama.cpp
 WORKDIR /opt/llama.cpp
-
-# Clone llama.cpp repository (with submodules)
 RUN git clone --recursive https://github.com/ggerganov/llama.cpp.git .
 
-# Build llama.cpp with HIP support
+# build
 RUN git clean -xdf \
-    && git pull \
-    && git submodule update --recursive \
-    && \
-    # Configure and compile with HIP toolchain
-    HIPCXX="$(hipconfig -l)/clang" HIP_PATH="$(hipconfig -R)" \
-      cmake -S . -B build \
-            -DGGML_HIP=ON \
-            -DAMDGPU_TARGETS=gfx1151 \
-            -DCMAKE_BUILD_TYPE=Release \
-            -DLLAMA_HIP_UMA=ON \
-    && cmake --build build --config Release -- -j$(nproc) \
-    && cmake --install build --config Release
+ && git pull \
+ && git submodule update --recursive \
+ && HIPCXX="$(hipconfig -l)/clang" HIP_PATH="$(hipconfig -R)" \
+    cmake -S . -B build \
+      -DGGML_HIP=ON \
+      -DAMDGPU_TARGETS=gfx1151 \
+      -DCMAKE_BUILD_TYPE=Release \
+      -DLLAMA_HIP_UMA=ON \
+ && cmake --build build --config Release -- -j$(nproc) \
+ && cmake --install build --config Release
 
+# libs
 RUN find /opt/llama.cpp/build -type f -name 'lib*.so*' -exec cp {} /usr/lib64/ \; \
  && ldconfig
 
+# helper
 COPY gguf-vram-estimator.py /usr/local/bin/gguf-vram-estimator.py
 RUN chmod +x /usr/local/bin/gguf-vram-estimator.py
 
-# Ensure we are using hipblaslt
+
+# runtime stage
+FROM registry.fedoraproject.org/fedora-minimal:rawhide
+
+# rocm 7.0 repos
+RUN <<'EOF'
+tee /etc/yum.repos.d/rocm.repo <<REPO
+[ROCm-7.0.0]
+name=ROCm7.0.0
+baseurl=https://repo.radeon.com/rocm/el9/7.0_beta/main
+enabled=1
+priority=50
+gpgcheck=1
+gpgkey=https://repo.radeon.com/rocm/rocm.gpg.key
+REPO
+
+tee /etc/yum.repos.d/rocm-graphics.repo <<REPO
+[ROCm-7.0.0-Graphics]
+name=ROCm7.0.0-Graphics
+baseurl=https://repo.radeon.com/graphics/7.0_beta/rhel/9/main/x86_64/
+enabled=1
+priority=50
+gpgcheck=1
+gpgkey=https://repo.radeon.com/rocm/rocm.gpg.key
+REPO
+EOF
+
+# runtime deps
+RUN microdnf -y --nodocs --setopt=install_weak_deps=0 install \
+      bash ca-certificates libatomic libstdc++ libgcc \
+      rocm rocminfo radeontop \
+  && microdnf clean all && rm -rf /var/cache/dnf/*
+
+# copy
+COPY --from=builder /usr/local/ /usr/local/
+
+# ld
+RUN echo "/usr/local/lib"  > /etc/ld.so.conf.d/local.conf \
+ && echo "/usr/local/lib64" >> /etc/ld.so.conf.d/local.conf \
+ && ldconfig \
+ && cp -n /usr/local/lib/libllama*.so* /usr/lib64/ 2>/dev/null || true \
+ && ldconfig
+
+# helper
+COPY gguf-vram-estimator.py /usr/local/bin/gguf-vram-estimator.py
+RUN chmod +x /usr/local/bin/gguf-vram-estimator.py
+
+# profile
 RUN printf '%s\n' \
   'export ROCBLAS_USE_HIPBLASLT=1' \
-  > /etc/profile.d/rocm.sh \
-  && chmod +x /etc/profile.d/rocm.sh \
+  > /etc/profile.d/rocm.sh && chmod +x /etc/profile.d/rocm.sh \
   && echo 'source /etc/profile.d/rocm.sh' >> /etc/bashrc
 
-# Default to interactive shell
+# shell
 CMD ["/bin/bash"]