HPL 2.3 from netlib compiled against OpenBLAS with a minimal
single-process MPI stub — no MPI package required in the ISO.
Matrix size is auto-sized to 80% of total RAM at runtime.
Build:
- VERSIONS: HPL_VERSION=2.3, HPL_SHA256=32c5c17d…
- build-hpl.sh: downloads HPL + OpenBLAS from Debian 12 repo,
compiles xhpl with a self-contained mpi_stub.c
- build.sh: step 80-hpl, injects xhpl + libopenblas into overlay
Runtime:
- bee-hpl: generates HPL.dat (N auto from /proc/meminfo, NB=256,
P=1 Q=1), runs xhpl, prints standard WR... Gflops output
- platform/hpl.go: RunHPL(), parses WR line → GFlops + PASSED/FAILED
- tasks.go: target "hpl"
- pages.go: LINPACK (HPL) card in validate/stress grid (stress-only)
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
All final artefacts for a given version now land in one place:
dist/easy-bee-v4.1/
easy-bee-nvidia-v4.1-amd64.iso
easy-bee-nvidia-v4.1-amd64.logs.tar.gz ← log archive
(logs dir deleted after archiving)
- Introduce OUT_DIR="${DIST_DIR}/easy-bee-v${ISO_VERSION_EFFECTIVE}"
- Move LOG_DIR, LOG_ARCHIVE, and ISO_OUT into OUT_DIR
- cleanup_build_log: use dirname(LOG_DIR) as tar -C base so the path is
correct regardless of where OUT_DIR lives; delete LOG_DIR after archiving
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- nvidia collector: read pcie.link.gen.current/max from nvidia-smi instead
of sysfs to avoid false Gen1 readings when GPU is in ASPM idle state
- build: remove bee-nccl-gpu-stress from rm -f list so shell script from
overlay is not silently dropped from the ISO
- smoketest: add explicit checks for bee-gpu-burn, bee-john-gpu-stress,
bee-nccl-gpu-stress, all_reduce_perf
- netconf: re-exec via sudo when not root to fix RTNETLINK/resolv.conf errors
- auto/config: reduce loglevel 7→3 to show clean systemd output on boot
- auto/config: blacklist snd_hda_intel and related audio modules (unused on servers)
- package-lists: remove firmware-intel-sound and firmware-amd-graphics from
base list; move firmware-amd-graphics to bee-amd variant only
- bible-local: mark memtest ADR resolved, document working solution
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
live-build picks up ALL .list.chroot files in config/package-lists/.
After rsync, bee-nvidia.list.chroot, bee-amd.list.chroot, and
bee-nogpu.list.chroot all end up in BUILD_WORK_DIR — causing lb to
try installing packages from every variant (and leaving version
placeholders unsubstituted in the unused lists).
Fix: after copying bee-${BEE_GPU_VENDOR}.list.chroot → bee-gpu.list.chroot,
delete all other bee-{nvidia,amd,nogpu}.list.chroot from BUILD_WORK_DIR.
Also includes nomsi boot mode changes (bee-nvidia-load + grub.cfg).
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- 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>
ROCM_BANDWIDTH_TEST_VERSION, ROCM_VALIDATION_SUITE_VERSION, ROCBLAS,
ROCRAND, HIP_RUNTIME_AMD, HIPBLASLT, COMGR were defined in VERSIONS and
in bee.list.chroot but the sed substitution block only covered 3 of them.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Move datacenter-gpu-manager and rocm-smi-lib from dynamic chroot hooks
into live-build's config/archives mechanism so lb caches the .deb files
in cache/packages.chroot/ between builds, eliminating repeated 900+ MB
downloads. Versions pinned via VERSIONS and substituted into package
lists at build time.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Switch squashfs compression from xz to zstd (3-5x faster compression,
~10-15% larger but decompresses faster at boot)
- Stop rm -rf BUILD_WORK_DIR on each build; rsync only config changes
so lb can reuse its chroot across builds (skips apt install step)
- Keep lb-packages cache in CACHE_ROOT as fallback if work dir is wiped
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Saves cache/packages.chroot before wiping BUILD_WORK_DIR and
restores it after, so apt packages are not re-downloaded on every
build. Cache lives in --cache-dir (same place as Go/NVIDIA cache).
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Without this, git refuses to read the bind-mounted repo (UID mismatch)
and describe returns empty, causing the version to fall back to iso/v1.0.20.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Plain v2.x tags are now the active tagging scheme; iso/v1.0.x tags
are legacy. Swap priority in resolve_iso_version so the ISO is named
bee-debian12-v2.x-amd64.iso instead of v1.0.x-N-gHASH.
Also tighten the v* pattern to v[0-9]* to avoid accidentally matching
other prefixed tags in both resolve functions.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
resolve_iso_version only matched iso/v* pattern; GUI release tags
(v2, v2.1 ... v2.6) were ignored, falling back to the old v1.0.20
annotated tag via resolve_audit_version.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Previously the ISO file was named after git describe --match 'audit/v*',
so a new iso/ tag produced names like v1.0.9-1-gXXXXXXX instead of v1.0.17.
Now build.sh has resolve_iso_version() that looks at iso/v* tags separately.
The bee binary inside the ISO still uses AUDIT_VERSION_EFFECTIVE.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Export detected DEBIAN_KERNEL_ABI as BEE_KERNEL_ABI from build.sh so
auto/config can pin linux-packages to the exact versioned package
(e.g. linux-image-6.1.0-31 + flavour amd64 = linux-image-6.1.0-31-amd64).
This prevents nvidia.ko vermagic mismatch if the linux-image-amd64
meta-package is updated between build start and lb build chroot step.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Dockerfile: linux-headers-amd64 meta-package instead of pinned ABI;
remove DEBIAN_KERNEL_ABI build-arg (no longer needed at image build time)
- build-in-container.sh: drop --build-arg DEBIAN_KERNEL_ABI
- build.sh: apt-get update + detect ABI from apt-cache at build time;
auto-install linux-headers-<ABI> if kernel changed since image build
Image rebuild is now needed only when changing Go version or lb tools,
not on every Debian kernel point release.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
DEBIAN_KERNEL_ABI=auto in VERSIONS — build.sh queries
apt-cache depends linux-image-amd64 to find the current ABI.
lb config now uses linux-image-amd64 meta-package.
This prevents build failures when Debian drops old kernel packages
from the repo (happens with every point release).
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
NVIDIA's CUDA repo for Debian 12 only has NCCL packages for cuda13.x,
not cuda12.x. Update to the latest available: 2.28.9-1+cuda13.0.
Also pass sha256 from VERSIONS into build-nccl.sh for integrity check.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Download libnccl2 .deb from NVIDIA's CUDA apt repo (Debian 12) during ISO
build, extract libnccl.so.* into the overlay at /usr/lib/ alongside
libnvidia-ml and libcuda. Version pinned in VERSIONS, reflected in
/etc/bee-release.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
The static KERNEL_PKG_VERSION pin was the root cause of nvidia-smi never
working: modules were compiled for pinned version (e.g. 6.12.76-r0) but
the ISO kernel was unpinned (latest from repo at build time). When Alpine
updated linux-lts, the two diverged silently.
Fix: both steps now use whatever linux-lts is current in Alpine 3.21 main
at build time. build-nvidia-module.sh uses `apk add --update linux-lts-dev`
(no version pin), mkimage gets the same package from the same mirror.
Module cache is still keyed by detected KVER so rebuilds remain fast.
Removed: KERNEL_VERSION, KERNEL_PKG_VERSION from VERSIONS, all pin references
from build.sh and build-nvidia-module.sh.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- bee-audit init.d: use --output file: so "audit output written" is logged
(stdout mode silently redirects, never emits the slog confirmation)
- build-nvidia-module.sh: use $KERNEL_SRC in find for .ko collection
(was hardcoded $EXTRACT_DIR/kernel, silent failure if path differs)
- smoketest: add bee-audit to required services (was never checked)
- smoketest: remove legacy bee-audit-debug from service list
- smoketest: internet ping → warn (live CD runs in isolated network, no internet)
- build.sh: auto-copy smoketest.sh → overlay/usr/local/bin/bee-smoketest
(removes manual sync hazard; smoketest.sh is now single source of truth)
- remove static overlay/usr/local/bin/bee-smoketest (generated by build.sh now)
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
