---
name: "ascend-memory-profiling"
description: "Attribute vLLM serving HBM usage on Ascend to weights, KV cache, HCCL, activations and runtime using measured evidence. Use for 显存归因, 显存 profiling, or 内存各部分拆分. A quick current memory-usage or idle-card lookup uses the fleet monitor; kernel latency analysis uses profiling-analysis."
---

<!-- Generated from .agents/skills/ascend-memory-profiling/SKILL.md. Do not edit. -->

# ascend-memory-profiling

Read `.agents/skills/ascend-memory-profiling/SKILL.md` and only the references needed
for the current task. The canonical skill owns the workflow.