+ j)dRt^RIHtHtHtHt.ROt]FtR]!4],n K AR#)aAnnotate kernels captured in CUDA graphs for profiler traces. During CUDA graph capture, :func:`mark_kernels` tags the GPU work nodes (kernels, memcpys, memsets) captured within its scope with user metadata. The recorded annotations are keyed by the same ``graph node id`` values that CUPTI-based profilers attach to kernel events during graph replay, so the annotations can later be joined against a profiler trace to identify which kernels belong to which region of the captured workload. Annotation recording is enabled per capture via the ``enable_annotations`` argument of :class:`torch.cuda.graph`:: import torch from torch.cuda.graph_annotations import mark_kernels, get_kernel_annotations g = torch.cuda.CUDAGraph() with torch.cuda.graph(g, enable_annotations=True): with mark_kernels("phase_A"): y = workload_a(x) with mark_kernels({"name": "phase_B", "dtype": "bf16"}): z = workload_b(y) annotations = get_kernel_annotations() The annotation mapping is typically pickled and joined against a profiler trace offline (matching each kernel event's ``graph node id`` field). Requires the ``cuda-bindings`` package and a CUDA driver that supports ``cudaGraphNodeGetToolsId`` (CUDA >= 13.1, or an equivalent cuda-compat package). When unavailable, recording silently degrades to a no-op; use :func:`is_available` to check support programmatically. .. warning:: This API is in prototype and may change in future releases. )clear_kernel_annotationsget_kernel_annotations is_available mark_kernelsztorch.cuda.graph_annotationsN) __doc__torch.cuda._graph_annotationsrrrr__all___nameglobals __module__]/home/luca/screen-checking/.venv/lib/python3.14/site-packages/torch/cuda/graph_annotations.pyrs:"H E"@GIe r