Mikhail Chusavitin
ffb85d55a2
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 (commit559fc29), 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>
26 lines
808 B
Plaintext
26 lines
808 B
Plaintext
DEBIAN_VERSION=12
|
|
DEBIAN_KERNEL_ABI=auto
|
|
NVIDIA_DRIVER_VERSION=580.126.20
|
|
NVIDIA_FABRICMANAGER_VERSION=580.126.20-1
|
|
NCCL_VERSION=2.28.9-1
|
|
NCCL_CUDA_VERSION=13.0
|
|
NCCL_SHA256=2e6faafd2c19cffc7738d9283976a3200ea9db9895907f337f0c7e5a25563186
|
|
NCCL_TESTS_VERSION=2.13.10
|
|
NVCC_VERSION=12.8
|
|
CUBLAS_VERSION=13.1.1.3-1
|
|
CUDA_USERSPACE_VERSION=13.0.96-1
|
|
DCGM_VERSION=4.5.2-1
|
|
JOHN_JUMBO_COMMIT=67fcf9fe5a
|
|
ROCM_VERSION=6.3.4
|
|
ROCM_SMI_VERSION=7.4.0.60304-76~22.04
|
|
ROCM_BANDWIDTH_TEST_VERSION=1.4.0.60304-76~22.04
|
|
ROCM_VALIDATION_SUITE_VERSION=1.1.0.60304-76~22.04
|
|
ROCBLAS_VERSION=4.3.0.60304-76~22.04
|
|
ROCRAND_VERSION=3.2.0.60304-76~22.04
|
|
HIP_RUNTIME_AMD_VERSION=6.3.42134.60304-76~22.04
|
|
HIPBLASLT_VERSION=0.10.0.60304-76~22.04
|
|
COMGR_VERSION=2.8.0.60304-76~22.04
|
|
GO_VERSION=1.24.0
|
|
AUDIT_VERSION=1.0.0
|
|
MEMTEST_VERSION=6.10-4
|