Persistent Straggler Rank from Single-GPU Thermal Throttling
One rank consistently 2-4x slower drags every synchronous collective. When the cause is a failed fan/blocked airflow on exactly one GPU, the fleet-wide symptom (slow allreduce everywhere) hides a single-node hardware root cause.
One rank consistently 2-4x slower drags every synchronous collective.
What this failure is
Persistent Straggler Rank from Single-GPU Thermal Throttling is a Performance failure seen during ML training runs. One rank consistently 2-4x slower drags every synchronous collective. When the cause is a failed fan/blocked airflow on exactly one GPU, the fleet-wide symptom (slow allreduce everywhere) hides a single-node hardware root cause. Common tags: Straggler, Thermal, Hw Slowdown, Slow Rank.
Is this what broke your run? Paste your log.
You're reading about Persistent Straggler Rank from Single-GPU Thermal Throttling. Paste your own crash log or traceback below and get the real root cause for YOUR run, not this generic entry. No account, no card. Logs are masked at ingress and never saved to account history.
Want 14 days on the Scale plan?
Request an evaluation code. A verified workplace organization activates up to 50 diagnoses a day, alerts, history, and follow-up questions. No credit card or automatic subscription.
Why it happens (the mechanism)
Failed chassis fan / blocked inlet on one GPU position. Degraded thermal paste or cold-plate contact on one SXM site. BMC fan-policy misconfiguration on one node. Taken together, these mechanisms explain why the failure is reproducible, why it tends to surface on specific workloads or scales, and why generic mitigation attempts often fall short without addressing the underlying cause.
What you'll observe
- collective time dominated by one rank (straggler detector flags the same rank every window)
- nvidia-smi on that GPU: HW Thermal Slowdown Active, temperature 15-30C above fleet
Common symptoms and what they mean
| Symptom | Why it happens |
|---|---|
| step_time p50 healthy, p99 = one rank at 3x | failed chassis fan / blocked inlet on one GPU position |
| clocks on the hot GPU pinned far below application clocks | degraded thermal paste or cold-plate contact on one SXM site |
| problem follows the node, not the rank id | BMC fan-policy misconfiguration on one node |
Which systems are affected
- dense DGX/HGX chassis with chassis-fan failures
- air-cooled retrofits
- any synchronous data-parallel training
How to confirm this is the problem
Use this checklist to test the hypothesis against a small reproduction. No single line proves the root cause, so preserve the preceding events and compare one variable at a time.
- ✓Reproduce the failure from a clean checkpoint/seed: the symptom must appear without warm-up state from a previous run.
- ✓Verified signal present: step_time p50 healthy, p99 = one rank at 3x
- ✓Verified signal present: clocks on the hot GPU pinned far below application clocks
- ✓Verified signal present: problem follows the node, not the rank id
- ✓A targeted fix from the "How to fix it" section eliminates or substantially reduces the symptom within one validation pass.
The fix and the prevention pattern
The root cause is on this page and stays free. A free account adds the exact remediation steps, saved history, and the fix on every entry in the encyclopedia.
Sign up free. Unlock the full analysisNo credit card. Daily allowance follows verified trust tier. Instant access.
Diagnose this failure in VS Code
Select the traceback or open the failed terminal, then run Denpex locally to see the initiating rank, collateral failures, exact fix, and verification command without uploading the log.
Install the free VS Code extensionRelated failures to investigate next
Root cause
- failed chassis fan / blocked inlet on one GPU position
- degraded thermal paste or cold-plate contact on one SXM site
- BMC fan-policy misconfiguration on one node
The fix and how to prevent it
Evaluate Denpex on your own logs
Request a Scale evaluation code. A verified workplace organization activates 14 days with up to 50 diagnoses a day. Every account keeps its current diagnosis allowance and gets a verification path. No card or automatic subscription.
Don't just read the fix, diagnose your run
The encyclopedia tells you what went wrong. Denpex tells you what went wrong in YOUR training run. With your logs, your config, and your stack.