CUDA graph replay detected invalidated memory buffer

A CUDA graph was replayed after the memory it captured was freed or reallocated. Graphs record raw device pointers, so any tensor whose storage moved between capture and replay leaves the graph reading memory that is no longer its own. This entry explains how to confirm the cause, apply the fix, and separate it from adjacent cuda-graphs failures.

The literal signature is "CUDA graph replay detected invalidated memory buffer". It is a environment failure associated with CUDA Graphs, custom kernels, and vision pipelines. The line identifies the failing operation or subsystem, while the surrounding evidence decides whether it is the initiating fault or a downstream symptom.

A CUDA graph was replayed after the memory it captured was freed or reallocated. Graphs record raw device pointers, so any tensor whose storage moved between capture and replay leaves the graph reading memory that is no longer its own. The failure becomes visible at this call site because the operation first requires the missing resource, valid state, healthy peer, or correct result. Earlier log lines and a known-good control carry more causal value than the final wrapper exception.

• The workload stops or loses forward progress after emitting "CUDA graph replay detected invalidated memory buffer".
• A retry on the same configuration reproduces the failure because the causal state has not changed.
• The outer framework exception can hide the rank, node, allocation, or dependency that failed first.
• Increasing timeouts or reducing workload size can suppress the symptom without correcting the cause.

• CUDA Graphs, custom kernels, and vision pipelines
• production-shaped multi-accelerator workloads
• containerized and bare-metal deployments of the same stack

Apply the following checklist to a small reproduction: each box below is a positive signal that you are looking at this exact failure rather than a sibling in the same taxonomy.

• ✓Find the first occurrence of "CUDA graph replay detected invalidated memory buffer" and preserve at least 100 lines before it.
• ✓Identify which rank, node, device, or process emitted the first related warning.
• ✓anything that reallocates between capture and replay breaks this, a changing batch shape, torch.empty inside the step, or an allocator flush (torch.cuda.empty_cache()). Confirm shapes are identical every iteration.
• ✓Repeat the same input after the targeted change and require the signature to disappear.
• ✓Resume from latest checkpoint only after the control passes.

CUDA graph replay detected invalidated memory buffer

Timestamps and exact values vary across runs, but the pattern. An info-level start, an early WARN, an ERROR carrying the symptom. Is the actual fingerprint you should alert on. The Denpex platform flags this combination automatically.

make the input and output buffers STATIC. Allocate them once, copy new data into them in place (static_input.copy_(new_data)), and replay, never rebind the graph to freshly allocated tensors. This changes the condition in the causal diagnosis instead of hiding the outer exception. The repeated control proves ownership before recovery from latest checkpoint.

# Preserve evidence before restarting
rg -n -i 'error|exception|timeout|failed' <log-file>
nvidia-smi
python -m torch.utils.collect_env

# Find the exact signature in the complete log
rg -n -F -- "CUDA graph replay detected invalidated memory buffer" <log-file>

Adapt the snippet to your framework. The same pattern holds for PyTorch Lightning, Hugging Face Trainer, DeepSpeed, Megatron-LM, and vLLM training wrappers. Where the wrapper exposes a config flag (for examplelr_scheduler_type in Trainer), prefer the flag over the imperative API to keep the schedule declarative and reproducible.

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.

CUDA reports errors asynchronously, so the traceback usually points at whatever line synchronised next rather than the one at fault. The hub covers every common CUDA error and how to make it report honestly.

• A CUDA graph was replayed after the memory it captured was freed or reallocated. Graphs record raw device pointers, so any tensor whose storage moved between capture and replay leaves the graph reading memory that is no longer its own.
• The decisive evidence is the first log line that precedes "CUDA graph replay detected invalidated memory buffer" and differs from a healthy run.
• A nearby failure remains a competing hypothesis until a control separates configuration, capacity, transport, and hardware causes.