feat(iso): split into nvidia and amd variants, fix KVM graphics and PATH

- build.sh: add --variant nvidia|amd; separate work dirs per variant (live-build-work-nvidia / live-build-work-amd); GPU-specific steps (modules, NCCL, cuBLAS, nccl-tests) run only for nvidia; deb package cache synced back to shared location after each lb build so second variant reuses downloaded packages; ISO output named easy-bee-{variant}-v{ver}-amd64.iso - build-in-container.sh: add --variant nvidia|amd|all (default: all); runs build.sh twice in one container for 'all'; --clean-build wipes both variant work dirs - package-lists: remove GPU packages from bee.list.chroot; add bee-nvidia.list.chroot (DCGM) and bee-amd.list.chroot (ROCm) - 9000-bee-setup hook: read /etc/bee-gpu-vendor; enable bee-nvidia.service and DCGM only for nvidia; set up ROCm symlinks only for amd - auto/config: --iso-volume uses BEE_GPU_VENDOR_UPPER env var - grub.cfg: add nomodeset to EASY-BEE and EASY-BEE (load to RAM) entries — fixes X/lightdm on BMC KVM (ASPEED AST chip requires nomodeset for fbdev to work; NVIDIA H100 compute does not need KMS) - bee.sh / smoketest.sh: add /usr/sbin to PATH so dmidecode, smartctl, nvme are found - 9100-memtest hook: add diagnostic listing of chroot/boot/memtest* files Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-03-30 22:24:37 +03:00
parent 905c581ece
commit ace1a9dba6
11 changed files with 305 additions and 173 deletions
--- a/iso/builder/auto/config
+++ b/iso/builder/auto/config
@@ -30,8 +30,8 @@ lb config noauto \
    --linux-flavours "amd64" \
    --linux-packages "${LB_LINUX_PACKAGES}" \
    --memtest none \
-    --iso-volume "EASY-BEE" \
-    --iso-application "EASY-BEE" \
+    --iso-volume "EASY-BEE-${BEE_GPU_VENDOR_UPPER:-NVIDIA}" \
+    --iso-application "EASY-BEE-${BEE_GPU_VENDOR_UPPER:-NVIDIA}" \
    --bootappend-live "boot=live components video=1920x1080 console=tty0 console=ttyS0,115200n8 loglevel=7 username=bee user-fullname=Bee modprobe.blacklist=nouveau" \
    --apt-recommends false \
    --chroot-squashfs-compression-type zstd \
--- a/iso/builder/build-in-container.sh
+++ b/iso/builder/build-in-container.sh
@@ -12,6 +12,7 @@ CACHE_DIR="${BEE_BUILDER_CACHE_DIR:-${REPO_ROOT}/dist/container-cache}"
 AUTH_KEYS=""
 REBUILD_IMAGE=0
 CLEAN_CACHE=0
+VARIANT="all"

 . "${BUILDER_DIR}/VERSIONS"

@@ -34,14 +35,23 @@ while [ $# -gt 0 ]; do
            REBUILD_IMAGE=1
            shift
            ;;
+        --variant)
+            VARIANT="$2"
+            shift 2
+            ;;
        *)
            echo "unknown arg: $1" >&2
-            echo "usage: $0 [--cache-dir /path] [--rebuild-image] [--clean-build] [--authorized-keys /path/to/authorized_keys]" >&2
+            echo "usage: $0 [--cache-dir /path] [--rebuild-image] [--clean-build] [--authorized-keys /path/to/authorized_keys] [--variant nvidia|amd|all]" >&2
            exit 1
            ;;
    esac
 done

+case "$VARIANT" in
+    nvidia|amd|all) ;;
+    *) echo "unknown variant: $VARIANT (expected nvidia, amd, or all)" >&2; exit 1 ;;
+esac
+
 if [ "$CLEAN_CACHE" = "1" ]; then
    echo "=== cleaning build cache: ${CACHE_DIR} ==="
    rm -rf "${CACHE_DIR:?}/go-build" \
@@ -49,8 +59,9 @@ if [ "$CLEAN_CACHE" = "1" ]; then
           "${CACHE_DIR:?}/tmp" \
           "${CACHE_DIR:?}/bee" \
           "${CACHE_DIR:?}/lb-packages"
-    echo "=== cleaning live-build work dir: ${REPO_ROOT}/dist/live-build-work ==="
-    rm -rf "${REPO_ROOT}/dist/live-build-work"
+    echo "=== cleaning live-build work dirs ==="
+    rm -rf "${REPO_ROOT}/dist/live-build-work-nvidia"
+    rm -rf "${REPO_ROOT}/dist/live-build-work-amd"
    echo "=== caches cleared, proceeding with build ==="
 fi

@@ -108,34 +119,71 @@ else
    echo "=== using existing builder image ${IMAGE_REF} (${BUILDER_PLATFORM}) ==="
 fi

-set -- \
-    run --rm --privileged \
-    --platform "${BUILDER_PLATFORM}" \
-    -v "${REPO_ROOT}:/work" \
-    -v "${CACHE_DIR}:/cache" \
-    -e BEE_CONTAINER_BUILD=1 \
-    -e GOCACHE=/cache/go-build \
-    -e GOMODCACHE=/cache/go-mod \
-    -e TMPDIR=/cache/tmp \
-    -e BEE_CACHE_DIR=/cache/bee \
-    -w /work \
-    "${IMAGE_REF}" \
-    sh /work/iso/builder/build.sh
-
-if [ -n "$AUTH_KEYS" ]; then
-    set -- run --rm --privileged \
-        --platform "${BUILDER_PLATFORM}" \
-        -v "${REPO_ROOT}:/work" \
-        -v "${CACHE_DIR}:/cache" \
-        -v "${AUTH_KEYS_DIR}:/tmp/bee-authkeys:ro" \
+# Build base docker run args (without --authorized-keys)
+build_run_args() {
+    _variant="$1"
+    _auth_arg=""
+    if [ -n "$AUTH_KEYS" ]; then
+        _auth_arg="--authorized-keys /tmp/bee-authkeys/${AUTH_KEYS_BASE}"
+    fi
+    echo "run --rm --privileged \
+        --platform ${BUILDER_PLATFORM} \
+        -v ${REPO_ROOT}:/work \
+        -v ${CACHE_DIR}:/cache \
+        ${AUTH_KEYS:+-v ${AUTH_KEYS_DIR}:/tmp/bee-authkeys:ro} \
        -e BEE_CONTAINER_BUILD=1 \
        -e GOCACHE=/cache/go-build \
        -e GOMODCACHE=/cache/go-mod \
        -e TMPDIR=/cache/tmp \
        -e BEE_CACHE_DIR=/cache/bee \
        -w /work \
-        "${IMAGE_REF}" \
-        sh /work/iso/builder/build.sh --authorized-keys "/tmp/bee-authkeys/${AUTH_KEYS_BASE}"
-fi
+        ${IMAGE_REF} \
+        sh /work/iso/builder/build.sh --variant ${_variant} ${_auth_arg}"
+}

-"$CONTAINER_TOOL" "$@"
+run_variant() {
+    _v="$1"
+    echo "=== building variant: ${_v} ==="
+    if [ -n "$AUTH_KEYS" ]; then
+        "$CONTAINER_TOOL" run --rm --privileged \
+            --platform "${BUILDER_PLATFORM}" \
+            -v "${REPO_ROOT}:/work" \
+            -v "${CACHE_DIR}:/cache" \
+            -v "${AUTH_KEYS_DIR}:/tmp/bee-authkeys:ro" \
+            -e BEE_CONTAINER_BUILD=1 \
+            -e GOCACHE=/cache/go-build \
+            -e GOMODCACHE=/cache/go-mod \
+            -e TMPDIR=/cache/tmp \
+            -e BEE_CACHE_DIR=/cache/bee \
+            -w /work \
+            "${IMAGE_REF}" \
+            sh /work/iso/builder/build.sh --variant "${_v}" \
+                --authorized-keys "/tmp/bee-authkeys/${AUTH_KEYS_BASE}"
+    else
+        "$CONTAINER_TOOL" run --rm --privileged \
+            --platform "${BUILDER_PLATFORM}" \
+            -v "${REPO_ROOT}:/work" \
+            -v "${CACHE_DIR}:/cache" \
+            -e BEE_CONTAINER_BUILD=1 \
+            -e GOCACHE=/cache/go-build \
+            -e GOMODCACHE=/cache/go-mod \
+            -e TMPDIR=/cache/tmp \
+            -e BEE_CACHE_DIR=/cache/bee \
+            -w /work \
+            "${IMAGE_REF}" \
+            sh /work/iso/builder/build.sh --variant "${_v}"
+    fi
+}
+
+case "$VARIANT" in
+    nvidia)
+        run_variant nvidia
+        ;;
+    amd)
+        run_variant amd
+        ;;
+    all)
+        run_variant nvidia
+        run_variant amd
+        ;;
+esac
--- a/iso/builder/build.sh
+++ b/iso/builder/build.sh
@@ -13,19 +13,29 @@ BUILDER_DIR="${REPO_ROOT}/iso/builder"
 OVERLAY_DIR="${REPO_ROOT}/iso/overlay"
 DIST_DIR="${REPO_ROOT}/dist"
 VENDOR_DIR="${REPO_ROOT}/iso/vendor"
-BUILD_WORK_DIR="${DIST_DIR}/live-build-work"
-OVERLAY_STAGE_DIR="${DIST_DIR}/overlay-stage"
 CACHE_ROOT="${BEE_CACHE_DIR:-${DIST_DIR}/cache}"
 AUTH_KEYS=""
+BEE_GPU_VENDOR="nvidia"

 # parse args
 while [ $# -gt 0 ]; do
    case "$1" in
        --authorized-keys) AUTH_KEYS="$2"; shift 2 ;;
+        --variant) BEE_GPU_VENDOR="$2"; shift 2 ;;
        *) echo "unknown arg: $1"; exit 1 ;;
    esac
 done

+case "$BEE_GPU_VENDOR" in
+    nvidia|amd) ;;
+    *) echo "unknown variant: $BEE_GPU_VENDOR (expected nvidia or amd)" >&2; exit 1 ;;
+esac
+
+BUILD_WORK_DIR="${DIST_DIR}/live-build-work-${BEE_GPU_VENDOR}"
+OVERLAY_STAGE_DIR="${DIST_DIR}/overlay-stage-${BEE_GPU_VENDOR}"
+
+export BEE_GPU_VENDOR
+
 . "${BUILDER_DIR}/VERSIONS"
 export PATH="$PATH:/usr/local/go/bin"

@@ -132,7 +142,7 @@ if [ ! -d "/usr/src/linux-headers-${KVER}" ]; then
    apt-get install -y "linux-headers-${KVER}"
 fi

-echo "=== bee ISO build ==="
+echo "=== bee ISO build (variant: ${BEE_GPU_VENDOR}) ==="
 echo "Debian: ${DEBIAN_VERSION}, Kernel ABI: ${DEBIAN_KERNEL_ABI}, Go: ${GO_VERSION}"
 echo "Audit version: ${AUDIT_VERSION_EFFECTIVE}, ISO version: ${ISO_VERSION_EFFECTIVE}"
 echo ""
@@ -141,8 +151,8 @@ echo "=== syncing git submodules ==="
 git -C "${REPO_ROOT}" submodule update --init --recursive

 # --- compile bee binary (static, Linux amd64) ---
+# Shared between variants — built once, reused on second pass.
 BEE_BIN="${DIST_DIR}/bee-linux-amd64"
-GPU_STRESS_BIN="${DIST_DIR}/bee-gpu-stress-linux-amd64"
 NEED_BUILD=1
 if [ -f "$BEE_BIN" ]; then
    NEWEST_SRC=$(find "${REPO_ROOT}/audit" -name '*.go' -newer "$BEE_BIN" | head -1)
@@ -172,37 +182,41 @@ else
    echo "=== bee binary up to date, skipping build ==="
 fi

-echo ""
-echo "=== downloading cuBLAS/cuBLASLt/cudart ${NCCL_CUDA_VERSION} userspace ==="
-sh "${BUILDER_DIR}/build-cublas.sh" \
-    "${CUBLAS_VERSION}" \
-    "${CUDA_USERSPACE_VERSION}" \
-    "${NCCL_CUDA_VERSION}" \
-    "${DIST_DIR}"
+# --- NVIDIA-only build steps ---
+GPU_STRESS_BIN="${DIST_DIR}/bee-gpu-stress-linux-amd64"
+if [ "$BEE_GPU_VENDOR" = "nvidia" ]; then
+    echo ""
+    echo "=== downloading cuBLAS/cuBLASLt/cudart ${NCCL_CUDA_VERSION} userspace ==="
+    sh "${BUILDER_DIR}/build-cublas.sh" \
+        "${CUBLAS_VERSION}" \
+        "${CUDA_USERSPACE_VERSION}" \
+        "${NCCL_CUDA_VERSION}" \
+        "${DIST_DIR}"

-CUBLAS_CACHE="${DIST_DIR}/cublas-${CUBLAS_VERSION}+cuda${NCCL_CUDA_VERSION}"
+    CUBLAS_CACHE="${DIST_DIR}/cublas-${CUBLAS_VERSION}+cuda${NCCL_CUDA_VERSION}"

-GPU_STRESS_NEED_BUILD=1
-if [ -f "$GPU_STRESS_BIN" ] && [ "${BUILDER_DIR}/bee-gpu-stress.c" -ot "$GPU_STRESS_BIN" ]; then
-    GPU_STRESS_NEED_BUILD=0
+    GPU_STRESS_NEED_BUILD=1
+    if [ -f "$GPU_STRESS_BIN" ] && [ "${BUILDER_DIR}/bee-gpu-stress.c" -ot "$GPU_STRESS_BIN" ]; then
+        GPU_STRESS_NEED_BUILD=0
+    fi
+
+    if [ "$GPU_STRESS_NEED_BUILD" = "1" ]; then
+        echo "=== building bee-gpu-stress ==="
+        gcc -O2 -s -Wall -Wextra \
+            -I"${CUBLAS_CACHE}/include" \
+            -o "$GPU_STRESS_BIN" \
+            "${BUILDER_DIR}/bee-gpu-stress.c" \
+            -ldl -lm
+        echo "binary: $GPU_STRESS_BIN"
+    else
+        echo "=== bee-gpu-stress up to date, skipping build ==="
+    fi
 fi

-if [ "$GPU_STRESS_NEED_BUILD" = "1" ]; then
-    echo "=== building bee-gpu-stress ==="
-    gcc -O2 -s -Wall -Wextra \
-        -I"${CUBLAS_CACHE}/include" \
-        -o "$GPU_STRESS_BIN" \
-        "${BUILDER_DIR}/bee-gpu-stress.c" \
-        -ldl -lm
-    echo "binary: $GPU_STRESS_BIN"
-else
-    echo "=== bee-gpu-stress up to date, skipping build ==="
-fi
-
-echo "=== preparing staged overlay ==="
-# Sync builder config into work dir, preserving lb cache (chroot + packages).
-# We do NOT rm -rf BUILD_WORK_DIR so lb can reuse its chroot on repeat builds.
+echo "=== preparing staged overlay (${BEE_GPU_VENDOR}) ==="
 mkdir -p "${BUILD_WORK_DIR}" "${OVERLAY_STAGE_DIR}"
+
+# Sync builder config into variant work dir, preserving lb cache.
 rsync -a --delete \
    --exclude='cache/' \
    --exclude='chroot/' \
@@ -212,7 +226,10 @@ rsync -a --delete \
    --exclude='*.contents' \
    --exclude='*.files' \
    "${BUILDER_DIR}/" "${BUILD_WORK_DIR}/"
-# Also persist package cache to CACHE_ROOT so it survives a manual wipe of BUILD_WORK_DIR.
+
+# Share deb package cache across variants.
+# Restore: populate work dir cache from shared cache before build.
+# Persist: sync back after build (done after lb build below).
 LB_PKG_CACHE="${CACHE_ROOT}/lb-packages"
 mkdir -p "${LB_PKG_CACHE}"
 if [ -d "${BUILD_WORK_DIR}/cache/packages.chroot" ]; then
@@ -221,6 +238,7 @@ elif [ -d "${LB_PKG_CACHE}" ] && [ "$(ls -A "${LB_PKG_CACHE}" 2>/dev/null)" ]; t
    mkdir -p "${BUILD_WORK_DIR}/cache/packages.chroot"
    rsync -a "${LB_PKG_CACHE}/" "${BUILD_WORK_DIR}/cache/packages.chroot/"
 fi
+
 rsync -a "${OVERLAY_DIR}/" "${OVERLAY_STAGE_DIR}/"
 rm -f \
    "${OVERLAY_STAGE_DIR}/etc/bee-ssh-password-fallback" \
@@ -231,6 +249,12 @@ rm -f \
    "${OVERLAY_STAGE_DIR}/usr/local/bin/bee-smoketest" \
    "${OVERLAY_STAGE_DIR}/usr/local/bin/all_reduce_perf"

+# Remove NVIDIA-specific overlay files for AMD variant
+if [ "$BEE_GPU_VENDOR" = "amd" ]; then
+    rm -f "${OVERLAY_STAGE_DIR}/usr/local/bin/bee-nvidia-load"
+    rm -f "${OVERLAY_STAGE_DIR}/etc/systemd/system/bee-nvidia.service"
+fi
+
 # --- inject authorized_keys for SSH access ---
 AUTHORIZED_KEYS_FILE="${OVERLAY_STAGE_DIR}/root/.ssh/authorized_keys"
 mkdir -p "${OVERLAY_STAGE_DIR}/root/.ssh"
@@ -268,8 +292,11 @@ fi
 mkdir -p "${OVERLAY_STAGE_DIR}/usr/local/bin"
 cp "${DIST_DIR}/bee-linux-amd64" "${OVERLAY_STAGE_DIR}/usr/local/bin/bee"
 chmod +x "${OVERLAY_STAGE_DIR}/usr/local/bin/bee"
-cp "${GPU_STRESS_BIN}" "${OVERLAY_STAGE_DIR}/usr/local/bin/bee-gpu-stress"
-chmod +x "${OVERLAY_STAGE_DIR}/usr/local/bin/bee-gpu-stress"
+
+if [ "$BEE_GPU_VENDOR" = "nvidia" ] && [ -f "$GPU_STRESS_BIN" ]; then
+    cp "${GPU_STRESS_BIN}" "${OVERLAY_STAGE_DIR}/usr/local/bin/bee-gpu-stress"
+    chmod +x "${OVERLAY_STAGE_DIR}/usr/local/bin/bee-gpu-stress"
+fi

 # --- inject smoketest into overlay so it runs directly on the live CD ---
 cp "${BUILDER_DIR}/smoketest.sh" "${OVERLAY_STAGE_DIR}/usr/local/bin/bee-smoketest"
@@ -286,107 +313,143 @@ for tool in storcli64 sas2ircu sas3ircu arcconf ssacli; do
    fi
 done

-# --- build NVIDIA kernel modules ---
-echo ""
-echo "=== building NVIDIA ${NVIDIA_DRIVER_VERSION} modules ==="
-sh "${BUILDER_DIR}/build-nvidia-module.sh" "${NVIDIA_DRIVER_VERSION}" "${DIST_DIR}" "${DEBIAN_KERNEL_ABI}"
+# --- NVIDIA kernel modules and userspace libs ---
+if [ "$BEE_GPU_VENDOR" = "nvidia" ]; then
+    echo ""
+    echo "=== building NVIDIA ${NVIDIA_DRIVER_VERSION} modules ==="
+    sh "${BUILDER_DIR}/build-nvidia-module.sh" "${NVIDIA_DRIVER_VERSION}" "${DIST_DIR}" "${DEBIAN_KERNEL_ABI}"

-KVER="${DEBIAN_KERNEL_ABI}-amd64"
-NVIDIA_CACHE="${DIST_DIR}/nvidia-${NVIDIA_DRIVER_VERSION}-${KVER}"
+    KVER="${DEBIAN_KERNEL_ABI}-amd64"
+    NVIDIA_CACHE="${DIST_DIR}/nvidia-${NVIDIA_DRIVER_VERSION}-${KVER}"

-# Inject .ko files into overlay at /usr/local/lib/nvidia/
-OVERLAY_KMOD_DIR="${OVERLAY_DIR}/usr/local/lib/nvidia"
-OVERLAY_KMOD_DIR="${OVERLAY_STAGE_DIR}/usr/local/lib/nvidia"
-mkdir -p "${OVERLAY_KMOD_DIR}"
-cp "${NVIDIA_CACHE}/modules/"*.ko "${OVERLAY_KMOD_DIR}/"
+    # Inject .ko files into overlay at /usr/local/lib/nvidia/
+    OVERLAY_KMOD_DIR="${OVERLAY_STAGE_DIR}/usr/local/lib/nvidia"
+    mkdir -p "${OVERLAY_KMOD_DIR}"
+    cp "${NVIDIA_CACHE}/modules/"*.ko "${OVERLAY_KMOD_DIR}/"

-# Inject nvidia-smi and libnvidia-ml
-mkdir -p "${OVERLAY_STAGE_DIR}/usr/local/bin" "${OVERLAY_STAGE_DIR}/usr/lib"
-cp "${NVIDIA_CACHE}/bin/nvidia-smi" "${OVERLAY_STAGE_DIR}/usr/local/bin/"
-chmod +x "${OVERLAY_STAGE_DIR}/usr/local/bin/nvidia-smi"
-cp "${NVIDIA_CACHE}/bin/nvidia-bug-report.sh" "${OVERLAY_STAGE_DIR}/usr/local/bin/" 2>/dev/null || true
-chmod +x "${OVERLAY_STAGE_DIR}/usr/local/bin/nvidia-bug-report.sh" 2>/dev/null || true
-cp "${NVIDIA_CACHE}/lib/"* "${OVERLAY_STAGE_DIR}/usr/lib/" 2>/dev/null || true
+    # Inject nvidia-smi and libnvidia-ml
+    mkdir -p "${OVERLAY_STAGE_DIR}/usr/local/bin" "${OVERLAY_STAGE_DIR}/usr/lib"
+    cp "${NVIDIA_CACHE}/bin/nvidia-smi" "${OVERLAY_STAGE_DIR}/usr/local/bin/"
+    chmod +x "${OVERLAY_STAGE_DIR}/usr/local/bin/nvidia-smi"
+    cp "${NVIDIA_CACHE}/bin/nvidia-bug-report.sh" "${OVERLAY_STAGE_DIR}/usr/local/bin/" 2>/dev/null || true
+    chmod +x "${OVERLAY_STAGE_DIR}/usr/local/bin/nvidia-bug-report.sh" 2>/dev/null || true
+    cp "${NVIDIA_CACHE}/lib/"* "${OVERLAY_STAGE_DIR}/usr/lib/" 2>/dev/null || true

-# Inject GSP firmware into /lib/firmware/nvidia/<version>/
-if [ -d "${NVIDIA_CACHE}/firmware" ] && [ "$(ls -A "${NVIDIA_CACHE}/firmware" 2>/dev/null)" ]; then
-    mkdir -p "${OVERLAY_STAGE_DIR}/lib/firmware/nvidia/${NVIDIA_DRIVER_VERSION}"
-    cp "${NVIDIA_CACHE}/firmware/"* "${OVERLAY_STAGE_DIR}/lib/firmware/nvidia/${NVIDIA_DRIVER_VERSION}/"
-    echo "=== firmware: $(ls "${OVERLAY_STAGE_DIR}/lib/firmware/nvidia/${NVIDIA_DRIVER_VERSION}/" | wc -l) files injected ==="
+    # Inject GSP firmware into /lib/firmware/nvidia/<version>/
+    if [ -d "${NVIDIA_CACHE}/firmware" ] && [ "$(ls -A "${NVIDIA_CACHE}/firmware" 2>/dev/null)" ]; then
+        mkdir -p "${OVERLAY_STAGE_DIR}/lib/firmware/nvidia/${NVIDIA_DRIVER_VERSION}"
+        cp "${NVIDIA_CACHE}/firmware/"* "${OVERLAY_STAGE_DIR}/lib/firmware/nvidia/${NVIDIA_DRIVER_VERSION}/"
+        echo "=== firmware: $(ls "${OVERLAY_STAGE_DIR}/lib/firmware/nvidia/${NVIDIA_DRIVER_VERSION}/" | wc -l) files injected ==="
+    fi
+
+    # --- build / download NCCL ---
+    echo ""
+    echo "=== downloading NCCL ${NCCL_VERSION}+cuda${NCCL_CUDA_VERSION} ==="
+    sh "${BUILDER_DIR}/build-nccl.sh" "${NCCL_VERSION}" "${NCCL_CUDA_VERSION}" "${DIST_DIR}" "${NCCL_SHA256:-}"
+
+    NCCL_CACHE="${DIST_DIR}/nccl-${NCCL_VERSION}+cuda${NCCL_CUDA_VERSION}"
+
+    # Inject libnccl.so.* into overlay alongside other NVIDIA userspace libs
+    cp "${NCCL_CACHE}/lib/"* "${OVERLAY_STAGE_DIR}/usr/lib/"
+    echo "=== NCCL: $(ls "${NCCL_CACHE}/lib/" | wc -l) files injected into /usr/lib/ ==="
+
+    # Inject cuBLAS/cuBLASLt/cudart runtime libs used by bee-gpu-stress tensor-core GEMM path
+    cp "${CUBLAS_CACHE}/lib/"* "${OVERLAY_STAGE_DIR}/usr/lib/"
+    echo "=== cuBLAS: $(ls "${CUBLAS_CACHE}/lib/" | wc -l) files injected into /usr/lib/ ==="
+
+    # --- build nccl-tests ---
+    echo ""
+    echo "=== building nccl-tests ${NCCL_TESTS_VERSION} ==="
+    sh "${BUILDER_DIR}/build-nccl-tests.sh" \
+        "${NCCL_TESTS_VERSION}" \
+        "${NCCL_VERSION}" \
+        "${NCCL_CUDA_VERSION}" \
+        "${DIST_DIR}" \
+        "${NVCC_VERSION}" \
+        "${DEBIAN_VERSION}"
+
+    NCCL_TESTS_CACHE="${DIST_DIR}/nccl-tests-${NCCL_TESTS_VERSION}"
+    cp "${NCCL_TESTS_CACHE}/bin/all_reduce_perf" "${OVERLAY_STAGE_DIR}/usr/local/bin/all_reduce_perf"
+    chmod +x "${OVERLAY_STAGE_DIR}/usr/local/bin/all_reduce_perf"
+    echo "=== all_reduce_perf injected ==="
 fi

-# --- build / download NCCL ---
-echo ""
-echo "=== downloading NCCL ${NCCL_VERSION}+cuda${NCCL_CUDA_VERSION} ==="
-sh "${BUILDER_DIR}/build-nccl.sh" "${NCCL_VERSION}" "${NCCL_CUDA_VERSION}" "${DIST_DIR}" "${NCCL_SHA256:-}"
-
-NCCL_CACHE="${DIST_DIR}/nccl-${NCCL_VERSION}+cuda${NCCL_CUDA_VERSION}"
-
-# Inject libnccl.so.* into overlay alongside other NVIDIA userspace libs
-cp "${NCCL_CACHE}/lib/"* "${OVERLAY_STAGE_DIR}/usr/lib/"
-echo "=== NCCL: $(ls "${NCCL_CACHE}/lib/" | wc -l) files injected into /usr/lib/ ==="
-
-# Inject cuBLAS/cuBLASLt/cudart runtime libs used by bee-gpu-stress tensor-core GEMM path
-cp "${CUBLAS_CACHE}/lib/"* "${OVERLAY_STAGE_DIR}/usr/lib/"
-echo "=== cuBLAS: $(ls "${CUBLAS_CACHE}/lib/" | wc -l) files injected into /usr/lib/ ==="
-
-# --- build nccl-tests ---
-echo ""
-echo "=== building nccl-tests ${NCCL_TESTS_VERSION} ==="
-sh "${BUILDER_DIR}/build-nccl-tests.sh" \
-    "${NCCL_TESTS_VERSION}" \
-    "${NCCL_VERSION}" \
-    "${NCCL_CUDA_VERSION}" \
-    "${DIST_DIR}" \
-    "${NVCC_VERSION}" \
-    "${DEBIAN_VERSION}"
-
-NCCL_TESTS_CACHE="${DIST_DIR}/nccl-tests-${NCCL_TESTS_VERSION}"
-cp "${NCCL_TESTS_CACHE}/bin/all_reduce_perf" "${OVERLAY_STAGE_DIR}/usr/local/bin/all_reduce_perf"
-chmod +x "${OVERLAY_STAGE_DIR}/usr/local/bin/all_reduce_perf"
-echo "=== all_reduce_perf injected ==="
-
 # --- embed build metadata ---
 mkdir -p "${OVERLAY_STAGE_DIR}/etc"
 BUILD_DATE="$(date +%Y-%m-%d)"
 GIT_COMMIT="$(git -C "${REPO_ROOT}" rev-parse --short HEAD 2>/dev/null || echo unknown)"
-cat > "${OVERLAY_STAGE_DIR}/etc/bee-release" <<EOF
-BEE_ISO_VERSION=${ISO_VERSION_EFFECTIVE}
-BEE_AUDIT_VERSION=${AUDIT_VERSION_EFFECTIVE}
-BUILD_DATE=${BUILD_DATE}
-GIT_COMMIT=${GIT_COMMIT}
-DEBIAN_VERSION=${DEBIAN_VERSION}
-DEBIAN_KERNEL_ABI=${DEBIAN_KERNEL_ABI}
-NVIDIA_DRIVER_VERSION=${NVIDIA_DRIVER_VERSION}
+
+if [ "$BEE_GPU_VENDOR" = "nvidia" ]; then
+    GPU_VERSION_LINE="NVIDIA_DRIVER_VERSION=${NVIDIA_DRIVER_VERSION}
 NCCL_VERSION=${NCCL_VERSION}
 NCCL_CUDA_VERSION=${NCCL_CUDA_VERSION}
 CUBLAS_VERSION=${CUBLAS_VERSION}
 CUDA_USERSPACE_VERSION=${CUDA_USERSPACE_VERSION}
-NCCL_TESTS_VERSION=${NCCL_TESTS_VERSION}
+NCCL_TESTS_VERSION=${NCCL_TESTS_VERSION}"
+    GPU_BUILD_INFO="nvidia:${NVIDIA_DRIVER_VERSION}"
+else
+    GPU_VERSION_LINE="ROCM_VERSION=${ROCM_VERSION}"
+    GPU_BUILD_INFO="rocm:${ROCM_VERSION}"
+fi
+
+cat > "${OVERLAY_STAGE_DIR}/etc/bee-release" <<EOF
+BEE_ISO_VERSION=${ISO_VERSION_EFFECTIVE}
+BEE_AUDIT_VERSION=${AUDIT_VERSION_EFFECTIVE}
+BEE_GPU_VENDOR=${BEE_GPU_VENDOR}
+BUILD_DATE=${BUILD_DATE}
+GIT_COMMIT=${GIT_COMMIT}
+DEBIAN_VERSION=${DEBIAN_VERSION}
+DEBIAN_KERNEL_ABI=${DEBIAN_KERNEL_ABI}
+${GPU_VERSION_LINE}
 EOF

+# Write GPU vendor marker for hooks
+echo "${BEE_GPU_VENDOR}" > "${OVERLAY_STAGE_DIR}/etc/bee-gpu-vendor"
+
 # Patch motd with build info
-BEE_BUILD_INFO="${BUILD_DATE} git:${GIT_COMMIT} debian:${DEBIAN_VERSION} nvidia:${NVIDIA_DRIVER_VERSION}"
+BEE_BUILD_INFO="${BUILD_DATE} git:${GIT_COMMIT} debian:${DEBIAN_VERSION} ${GPU_BUILD_INFO}"
 if [ -f "${OVERLAY_STAGE_DIR}/etc/motd" ]; then
    sed "s/%%BUILD_INFO%%/${BEE_BUILD_INFO}/" "${OVERLAY_STAGE_DIR}/etc/motd" \
        > "${OVERLAY_STAGE_DIR}/etc/motd.patched"
    mv "${OVERLAY_STAGE_DIR}/etc/motd.patched" "${OVERLAY_STAGE_DIR}/etc/motd"
 fi

-# --- substitute version placeholders in package list ---
-sed -i \
-    -e "s/%%DCGM_VERSION%%/${DCGM_VERSION}/g" \
-    -e "s/%%ROCM_VERSION%%/${ROCM_VERSION}/g" \
-    -e "s/%%ROCM_SMI_VERSION%%/${ROCM_SMI_VERSION}/g" \
-    -e "s/%%ROCM_BANDWIDTH_TEST_VERSION%%/${ROCM_BANDWIDTH_TEST_VERSION}/g" \
-    -e "s/%%ROCM_VALIDATION_SUITE_VERSION%%/${ROCM_VALIDATION_SUITE_VERSION}/g" \
-    -e "s/%%ROCBLAS_VERSION%%/${ROCBLAS_VERSION}/g" \
-    -e "s/%%ROCRAND_VERSION%%/${ROCRAND_VERSION}/g" \
-    -e "s/%%HIP_RUNTIME_AMD_VERSION%%/${HIP_RUNTIME_AMD_VERSION}/g" \
-    -e "s/%%HIPBLASLT_VERSION%%/${HIPBLASLT_VERSION}/g" \
-    -e "s/%%COMGR_VERSION%%/${COMGR_VERSION}/g" \
-    "${BUILD_WORK_DIR}/config/package-lists/bee.list.chroot" \
-    "${BUILD_WORK_DIR}/config/archives/rocm.list.chroot"
+# --- copy variant-specific package list into work dir ---
+cp "${BUILD_WORK_DIR}/config/package-lists/bee-${BEE_GPU_VENDOR}.list.chroot" \
+   "${BUILD_WORK_DIR}/config/package-lists/bee-gpu.list.chroot"
+
+# --- remove archives for the other vendor ---
+if [ "$BEE_GPU_VENDOR" = "nvidia" ]; then
+    rm -f "${BUILD_WORK_DIR}/config/archives/rocm.list.chroot" \
+          "${BUILD_WORK_DIR}/config/archives/rocm.key.chroot"
+else
+    rm -f "${BUILD_WORK_DIR}/config/archives/nvidia-cuda.list.chroot" \
+          "${BUILD_WORK_DIR}/config/archives/nvidia-cuda.key.chroot"
+fi
+
+# --- substitute version placeholders in package list and archive ---
+if [ "$BEE_GPU_VENDOR" = "nvidia" ]; then
+    sed -i \
+        -e "s/%%DCGM_VERSION%%/${DCGM_VERSION}/g" \
+        "${BUILD_WORK_DIR}/config/package-lists/bee-gpu.list.chroot"
+else
+    sed -i \
+        -e "s/%%ROCM_VERSION%%/${ROCM_VERSION}/g" \
+        -e "s/%%ROCM_SMI_VERSION%%/${ROCM_SMI_VERSION}/g" \
+        -e "s/%%ROCM_BANDWIDTH_TEST_VERSION%%/${ROCM_BANDWIDTH_TEST_VERSION}/g" \
+        -e "s/%%ROCM_VALIDATION_SUITE_VERSION%%/${ROCM_VALIDATION_SUITE_VERSION}/g" \
+        -e "s/%%ROCBLAS_VERSION%%/${ROCBLAS_VERSION}/g" \
+        -e "s/%%ROCRAND_VERSION%%/${ROCRAND_VERSION}/g" \
+        -e "s/%%HIP_RUNTIME_AMD_VERSION%%/${HIP_RUNTIME_AMD_VERSION}/g" \
+        -e "s/%%HIPBLASLT_VERSION%%/${HIPBLASLT_VERSION}/g" \
+        -e "s/%%COMGR_VERSION%%/${COMGR_VERSION}/g" \
+        "${BUILD_WORK_DIR}/config/package-lists/bee-gpu.list.chroot"
+    if [ -f "${BUILD_WORK_DIR}/config/archives/rocm.list.chroot" ]; then
+        sed -i \
+            -e "s/%%ROCM_VERSION%%/${ROCM_VERSION}/g" \
+            "${BUILD_WORK_DIR}/config/archives/rocm.list.chroot"
+    fi
+fi

 # --- sync overlay into live-build includes.chroot ---
 LB_DIR="${BUILD_WORK_DIR}"
@@ -402,20 +465,31 @@ fi

 # --- build ISO using live-build ---
 echo ""
-echo "=== building ISO (live-build) ==="
+echo "=== building ISO (live-build, variant: ${BEE_GPU_VENDOR}) ==="
+
+# Export for auto/config
+BEE_GPU_VENDOR_UPPER="$(echo "${BEE_GPU_VENDOR}" | tr 'a-z' 'A-Z')"
+export BEE_GPU_VENDOR_UPPER

 cd "${LB_DIR}"
 lb clean 2>&1 | tail -3
 lb config 2>&1 | tail -5
 lb build 2>&1

+# --- persist deb package cache back to shared location ---
+# This allows the second variant to reuse all downloaded packages.
+if [ -d "${BUILD_WORK_DIR}/cache/packages.chroot" ]; then
+    rsync -a "${BUILD_WORK_DIR}/cache/packages.chroot/" "${LB_PKG_CACHE}/"
+    echo "=== package cache synced to ${LB_PKG_CACHE} ==="
+fi
+
 # live-build outputs live-image-amd64.hybrid.iso in LB_DIR
 ISO_RAW="${LB_DIR}/live-image-amd64.hybrid.iso"
-ISO_OUT="${DIST_DIR}/bee-debian${DEBIAN_VERSION}-v${ISO_VERSION_EFFECTIVE}-amd64.iso"
+ISO_OUT="${DIST_DIR}/easy-bee-${BEE_GPU_VENDOR}-v${ISO_VERSION_EFFECTIVE}-amd64.iso"
 if [ -f "$ISO_RAW" ]; then
    cp "$ISO_RAW" "$ISO_OUT"
    echo ""
-    echo "=== done ==="
+    echo "=== done (${BEE_GPU_VENDOR}) ==="
    echo "ISO: $ISO_OUT"
    if command -v stat >/dev/null 2>&1; then
        ISO_SIZE_BYTES="$(stat -c '%s' "$ISO_OUT" 2>/dev/null || stat -f '%z' "$ISO_OUT")"
--- a/iso/builder/config/bootloaders/grub-pc/grub.cfg
+++ b/iso/builder/config/bootloaders/grub-pc/grub.cfg
@@ -10,12 +10,12 @@ echo "  ╚══════╝╚═╝  ╚═╝╚══════╝
 echo ""

 menuentry "EASY-BEE" {
-    linux   @KERNEL_LIVE@ @APPEND_LIVE@ bee.nvidia.mode=normal net.ifnames=0 biosdevname=0 mitigations=off transparent_hugepage=always numa_balancing=disable nowatchdog nosoftlockup
+    linux   @KERNEL_LIVE@ @APPEND_LIVE@ nomodeset bee.nvidia.mode=normal net.ifnames=0 biosdevname=0 mitigations=off transparent_hugepage=always numa_balancing=disable nowatchdog nosoftlockup
    initrd  @INITRD_LIVE@
 }

 menuentry "EASY-BEE (load to RAM)" {
-    linux   @KERNEL_LIVE@ @APPEND_LIVE@ toram bee.nvidia.mode=normal net.ifnames=0 biosdevname=0 mitigations=off transparent_hugepage=always numa_balancing=disable nowatchdog nosoftlockup
+    linux   @KERNEL_LIVE@ @APPEND_LIVE@ toram nomodeset bee.nvidia.mode=normal net.ifnames=0 biosdevname=0 mitigations=off transparent_hugepage=always numa_balancing=disable nowatchdog nosoftlockup
    initrd  @INITRD_LIVE@
 }

--- a/iso/builder/config/hooks/normal/9000-bee-setup.hook.chroot
+++ b/iso/builder/config/hooks/normal/9000-bee-setup.hook.chroot
@@ -5,6 +5,9 @@ set -e

 echo "=== bee chroot setup ==="

+GPU_VENDOR=$(cat /etc/bee-gpu-vendor 2>/dev/null || echo nvidia)
+echo "=== GPU vendor: ${GPU_VENDOR} ==="
+
 ensure_bee_console_user() {
    if id bee >/dev/null 2>&1; then
        usermod -d /home/bee -s /bin/bash bee 2>/dev/null || true
@@ -21,10 +24,8 @@ ensure_bee_console_user() {

 ensure_bee_console_user

-# Enable bee services
-systemctl enable nvidia-dcgm.service 2>/dev/null || true
+# Enable common bee services
 systemctl enable bee-network.service
-systemctl enable bee-nvidia.service
 systemctl enable bee-preflight.service
 systemctl enable bee-audit.service
 systemctl enable bee-web.service
@@ -36,25 +37,33 @@ systemctl enable serial-getty@ttyS0.service 2>/dev/null || true
 systemctl enable serial-getty@ttyS1.service 2>/dev/null || true
 systemctl enable bee-journal-mirror@ttyS1.service 2>/dev/null || true

+# Enable GPU-vendor specific services
+if [ "$GPU_VENDOR" = "nvidia" ]; then
+    systemctl enable nvidia-dcgm.service 2>/dev/null || true
+    systemctl enable bee-nvidia.service
+elif [ "$GPU_VENDOR" = "amd" ]; then
+    # ROCm symlinks (packages install to /opt/rocm-*/bin/)
+    for tool in rocm-smi rocm-bandwidth-test rvs; do
+        if [ ! -e /usr/local/bin/${tool} ]; then
+            bin_path="$(find /opt -path "*/bin/${tool}" -type f 2>/dev/null | sort | tail -1)"
+            [ -n "${bin_path}" ] && ln -sf "${bin_path}" /usr/local/bin/${tool}
+        fi
+    done
+fi
+
 # Ensure scripts are executable
 chmod +x /usr/local/bin/bee-network.sh  2>/dev/null || true
-chmod +x /usr/local/bin/bee-nvidia-load 2>/dev/null || true
 chmod +x /usr/local/bin/bee-sshsetup   2>/dev/null || true
 chmod +x /usr/local/bin/bee-smoketest  2>/dev/null || true
 chmod +x /usr/local/bin/bee            2>/dev/null || true
 chmod +x /usr/local/bin/bee-log-run    2>/dev/null || true
+if [ "$GPU_VENDOR" = "nvidia" ]; then
+    chmod +x /usr/local/bin/bee-nvidia-load 2>/dev/null || true
+fi

 # Reload udev rules
 udevadm control --reload-rules 2>/dev/null || true

-# rocm symlinks (packages install to /opt/rocm-*/bin/)
-for tool in rocm-smi rocm-bandwidth-test rvs; do
-    if [ ! -e /usr/local/bin/${tool} ]; then
-        bin_path="$(find /opt -path "*/bin/${tool}" -type f 2>/dev/null | sort | tail -1)"
-        [ -n "${bin_path}" ] && ln -sf "${bin_path}" /usr/local/bin/${tool}
-    fi
-done
-
 # Create export directory
 mkdir -p /appdata/bee/export

@@ -62,4 +71,4 @@ if [ -f /etc/sudoers.d/bee ]; then
    chmod 0440 /etc/sudoers.d/bee
 fi

-echo "=== bee chroot setup complete ==="
+echo "=== bee chroot setup complete (${GPU_VENDOR}) ==="
--- a/iso/builder/config/hooks/normal/9100-memtest.hook.binary
+++ b/iso/builder/config/hooks/normal/9100-memtest.hook.binary
@@ -4,6 +4,9 @@
 # not inside the squashfs).
 set -e

+echo "memtest: scanning chroot/boot/ for memtest files:"
+ls chroot/boot/memtest* 2>/dev/null || echo "memtest: WARNING: no memtest files found in chroot/boot/"
+
 for f in memtest86+x64.bin memtest86+x64.efi memtest86+ia32.bin memtest86+ia32.efi; do
    src="chroot/boot/${f}"
    if [ -f "${src}" ]; then
--- a/iso/builder/config/package-lists/bee-amd.list.chroot
+++ b/iso/builder/config/package-lists/bee-amd.list.chroot
@@ -0,0 +1,9 @@
+# AMD ROCm — GPU monitoring, bandwidth test, and compute stress (RVS GST)
+rocm-smi-lib=%%ROCM_SMI_VERSION%%
+rocm-bandwidth-test=%%ROCM_BANDWIDTH_TEST_VERSION%%
+rocm-validation-suite=%%ROCM_VALIDATION_SUITE_VERSION%%
+rocblas=%%ROCBLAS_VERSION%%
+rocrand=%%ROCRAND_VERSION%%
+hip-runtime-amd=%%HIP_RUNTIME_AMD_VERSION%%
+hipblaslt=%%HIPBLASLT_VERSION%%
+comgr=%%COMGR_VERSION%%
--- a/iso/builder/config/package-lists/bee-nvidia.list.chroot
+++ b/iso/builder/config/package-lists/bee-nvidia.list.chroot
@@ -0,0 +1,2 @@
+# NVIDIA DCGM (Data Center GPU Manager) — dcgmi diag for acceptance testing
+datacenter-gpu-manager=1:%%DCGM_VERSION%%
--- a/iso/builder/config/package-lists/bee.list.chroot
+++ b/iso/builder/config/package-lists/bee.list.chroot
@@ -72,18 +72,5 @@ firmware-bnx2x
 firmware-cavium
 firmware-qlogic

-# NVIDIA DCGM (Data Center GPU Manager) — dcgmi diag for acceptance testing
-datacenter-gpu-manager=1:%%DCGM_VERSION%%
-
-# AMD ROCm — GPU monitoring, bandwidth test, and compute stress (RVS GST)
-rocm-smi-lib=%%ROCM_SMI_VERSION%%
-rocm-bandwidth-test=%%ROCM_BANDWIDTH_TEST_VERSION%%
-rocm-validation-suite=%%ROCM_VALIDATION_SUITE_VERSION%%
-rocblas=%%ROCBLAS_VERSION%%
-rocrand=%%ROCRAND_VERSION%%
-hip-runtime-amd=%%HIP_RUNTIME_AMD_VERSION%%
-hipblaslt=%%HIPBLASLT_VERSION%%
-comgr=%%COMGR_VERSION%%
-
 # glibc compat helpers (for any external binaries that need it)
 libc6
--- a/iso/builder/smoketest.sh
+++ b/iso/builder/smoketest.sh
@@ -39,7 +39,7 @@ info "nvidia boot mode: ${NVIDIA_BOOT_MODE}"
 # --- PATH & binaries ---
 echo "-- PATH & binaries --"
 for tool in dmidecode smartctl nvme ipmitool lspci bee; do
-    if p=$(PATH="/usr/local/bin:$PATH" command -v "$tool" 2>/dev/null); then
+    if p=$(PATH="/usr/local/bin:/usr/sbin:/sbin:$PATH" command -v "$tool" 2>/dev/null); then
        ok "$tool found: $p"
    else
        fail "$tool: NOT FOUND"
--- a/iso/overlay/etc/profile.d/bee.sh
+++ b/iso/overlay/etc/profile.d/bee.sh
@@ -1,4 +1,4 @@
-export PATH="$PATH:/usr/local/bin:/opt/rocm/bin:/opt/rocm/sbin"
+export PATH="$PATH:/usr/local/bin:/usr/sbin:/sbin:/opt/rocm/bin:/opt/rocm/sbin"

 # Print web UI URLs on the local console at login.
 if [ -z "${SSH_CONNECTION:-}" ] \