Read NVLink and GPU-to-adapter distance from nvidia-smi topo

The release binary is built without cgo, so without NVML, and NVML was
the only source of NVLink peers: on every NVLink machine the portable
agent reported none, and tensor-parallel groups were chosen by GPU index.
Discovery now reads `nvidia-smi topo -m` after either path, for the
NVLink peers (kept from NVML where it gave them), each card's NUMA node,
and how far each network adapter is from it - PIX, PXB, PHB, NODE or
SYS - which travels to the control plane as nic_paths for placement to
use. Both the NICn-with-legend layout of newer drivers and the
adapter-named columns of older ones are read.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
This commit is contained in:
Ilker ManapandClaude Opus 5.5 committed 2026-09-29 00:09:26 +02:00
1 parent 22bb098247
commit bf2d171e89
8 files changed
+363 -30

No files matched your search

+2 -2
View File
@@ -40,8 +40,8 @@ const mandatorySMIFields = 10
//
// This is the cgo-free path (plan §3.2): the NVML bindings need cgo, which
// would force the whole binary — server included — to be built with a C
// toolchain. nvidia-smi gives everything except NVLink topology, which the
// NVML build adds.
// toolchain. The query gives everything except topology, which Discover
// reads from `nvidia-smi topo -m` afterwards.
func DiscoverGPUsSMI(ctx context.Context, binary string) ([]GPU, error) {
if binary == "" {
// Not configured and not installed is a node without an NVIDIA card,