reanimator/bee
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases 1 Wiki Activity

iso: pin NVIDIA driver/DCGM to DGX OS 7.5.0-validated combo

Browse Source 
590.48.01/DCGM 4.6.0-1 was never a deliberate compatibility choice — the
driver version traces back to a CDN 404 on the originally pinned
550.54.15 (commit 559fc29), and DCGM was bumped twice since purely
because NVIDIA pulled the previously-pinned core package from the repo
(6dee8f3, b2e177a). The 590 branch only has two point releases so far,
and its nvvs plugin fails dcgmi diag with "Detected driver major
version 0 is not between the required versions 346 and 2000" on H100 —
confirmed via a staging support bundle.

Switch to 580.126.20 / DCGM 4.5.2-1, the combo NVIDIA's own DGX OS 7.5.0
release notes validate for H100. Both packages confirmed present in the
NVIDIA CUDA debian12/x86_64 apt repo.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
This commit is contained in:
Mikhail Chusavitin
2026-07-06 13:33:52 +03:00
parent ee43a8e6ad
commit ffb85d55a2
3 changed files with 5 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files
iso/README.md
+1 -1
View File
@@ -48,7 +48,7 @@ sh iso/builder/build-in-container.sh --cache-dir /path/to/cache
- The builder image is automatically rebuilt if the local tag exists for the wrong architecture. - The builder image is automatically rebuilt if the local tag exists for the wrong architecture.
- The live ISO boots with Debian `live-boot` `toram`, so the read-only medium is copied into RAM during boot and the runtime no longer depends on the original USB/BMC virtual media staying present. - The live ISO boots with Debian `live-boot` `toram`, so the read-only medium is copied into RAM during boot and the runtime no longer depends on the original USB/BMC virtual media staying present.
- Target systems need enough RAM for the full compressed live medium plus normal runtime overhead, or boot may fail before reaching the TUI. - Target systems need enough RAM for the full compressed live medium plus normal runtime overhead, or boot may fail before reaching the TUI.
- The NVIDIA variant installs DCGM 4 packages matched to the CUDA user-mode driver major version. For driver branch `590` / CUDA `13.x`, the package family is `datacenter-gpu-manager-4-cuda13` rather than legacy `datacenter-gpu-manager`. - The NVIDIA variant installs DCGM 4 packages matched to the CUDA user-mode driver major version. For driver branch `580` / CUDA `13.x`, the package family is `datacenter-gpu-manager-4-cuda13` rather than legacy `datacenter-gpu-manager`.
- Override the container platform only if you know why: - Override the container platform only if you know why:
```sh ```sh
iso/builder/VERSIONS
+3 -3
View File
@@ -1,7 +1,7 @@
DEBIAN_VERSION=12 DEBIAN_VERSION=12
DEBIAN_KERNEL_ABI=auto DEBIAN_KERNEL_ABI=auto
NVIDIA_DRIVER_VERSION=590.48.01 NVIDIA_DRIVER_VERSION=580.126.20
NVIDIA_FABRICMANAGER_VERSION=590.48.01-1 NVIDIA_FABRICMANAGER_VERSION=580.126.20-1
NCCL_VERSION=2.28.9-1 NCCL_VERSION=2.28.9-1
NCCL_CUDA_VERSION=13.0 NCCL_CUDA_VERSION=13.0
NCCL_SHA256=2e6faafd2c19cffc7738d9283976a3200ea9db9895907f337f0c7e5a25563186 NCCL_SHA256=2e6faafd2c19cffc7738d9283976a3200ea9db9895907f337f0c7e5a25563186
@@ -9,7 +9,7 @@ NCCL_TESTS_VERSION=2.13.10
NVCC_VERSION=12.8 NVCC_VERSION=12.8
CUBLAS_VERSION=13.1.1.3-1 CUBLAS_VERSION=13.1.1.3-1
CUDA_USERSPACE_VERSION=13.0.96-1 CUDA_USERSPACE_VERSION=13.0.96-1
DCGM_VERSION=4.6.0-1 DCGM_VERSION=4.5.2-1
JOHN_JUMBO_COMMIT=67fcf9fe5a JOHN_JUMBO_COMMIT=67fcf9fe5a
ROCM_VERSION=6.3.4 ROCM_VERSION=6.3.4
ROCM_SMI_VERSION=7.4.0.60304-76~22.04 ROCM_SMI_VERSION=7.4.0.60304-76~22.04
iso/builder/config/package-lists/bee-nvidia.list.chroot
+1 -1
View File
@@ -2,7 +2,7 @@
# Validate uses dcgmi diagnostics; Burn uses dcgmproftester as the official # Validate uses dcgmi diagnostics; Burn uses dcgmproftester as the official
# NVIDIA max-compute recipe. The smoketest/runtime contract treats # NVIDIA max-compute recipe. The smoketest/runtime contract treats
# dcgmproftester as required in the LiveCD. # dcgmproftester as required in the LiveCD.
# DCGM 4 is packaged per CUDA major. The image ships NVIDIA driver 590 with # DCGM 4 is packaged per CUDA major. The image ships NVIDIA driver 580 with
# CUDA 13 userspace, so install the CUDA 13 build plus proprietary components # CUDA 13 userspace, so install the CUDA 13 build plus proprietary components
# explicitly. # explicitly.
nvtop nvtop