Deepstream Profile Pipeline

What it does

When your agent uses it

• The user asks for an efficient
• Profiled pipeline —

Example prompts

• “/deepstream-profile-pipeline”

Requirements

• Python 3
• Compatibility (from SKILL.md): DeepStream SDK 9.0 on Ubuntu 22.04 or 24.04, run from the `nvcr.io/nvidia/deepstream:9.0-triton-multiarch` container (the dev image; the slimmer `samples-multiarch` variant strips the nsys NVTX injector and produces empty per-plugin NVTX traces — do not use it for profiling). Requires `nsys` (Nsight Systems 2024+) and `nvidia-smi` on PATH. No GUI dependency — the skill runs fully headless and uses only `nsys profile` + `nsys stats`.

What it can do on your machine

Read from SKILL.md and the folder at commit 67a13c0. It shows what the files ask for, not the result of running them.

• Tool permissions

  Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

  From allowed-tools in the SKILL.md frontmatter.

• Runs code

  Ships 1 file in scripts/ (Python), which the agent can run.

  From the folder's file list and the shell code blocks in SKILL.md.

• Network

  No URLs in SKILL.md.

  From URLs in SKILL.md, links to its own repository left out.

• Credentials

  Names no API keys, tokens, secrets or passwords.

  From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

• Compatibility

  DeepStream SDK 9.0 on Ubuntu 22.04 or 24.04, run from the `nvcr.io/nvidia/deepstream:9.0-triton-multiarch` container (the dev image; the slimmer `samples-multiarch` variant strips the nsys NVTX injector and produces empty per-plugin NVTX traces — do not use it for profiling). Requires `nsys` (Nsight Systems 2024+) and `nvidia-smi` on PATH. No GUI dependency — the skill runs fully headless and uses only `nsys profile` + `nsys stats`.

  From compatibility in the SKILL.md frontmatter.

Context cost

Deepstream Profile Pipeline loads about 4.1k tokens when it runs, and up to ~13k if it reads all its reference files. Until then it costs about 59 tokens; SKILL.md has 1,467 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it

~59

When it runs · the whole SKILL.md, loaded when a task matches

~4.1k

With references · SKILL.md plus every file in references/, read only if the agent opens them

~13k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); the scripts in this folder are not scanned.

SKILL.md

The full file from NVIDIA/skills at commit 67a13c0, republished under its Apache-2.0 licence (© NVIDIA). 1,467 words, ~4,066 tokens.

name

deepstream-profile-pipeline

description

Profile a DeepStream pipeline with Nsight Systems and derive its configs from the measurement. Use when the user asks for an efficient, performant, or profiled pipeline — or to benchmark, tune, or measure FPS.

compatibility

DeepStream SDK 9.0 on Ubuntu 22.04 or 24.04, run from the `nvcr.io/nvidia/deepstream:9.0-triton-multiarch` container (the dev image; the slimmer `samples-multiarch` variant strips the nsys NVTX injector and produces empty per-plugin NVTX traces — do not use it for profiling). Requires `nsys` (Nsight Systems 2024+) and `nvidia-smi` on PATH. No GUI dependency — the skill runs fully headless and uses only `nsys profile` + `nsys stats`.

metadata.author

NVIDIA CORPORATION

metadata.tags

deepstream, profiling, nsight-systems, nvtx, nvidia-smi, benchmarking

metadata.languages

bash, python, yaml

metadata.domain

video-analytics

metadata.team

deepstream-sdk

owner

NVIDIA CORPORATION

service

deepstream

version

0.1.0

reviewed

2026-04-24

license

CC-BY-4.0 AND Apache-2.0

nvurisrcbin       COVERED
nvstreammux       COVERED
nvinfer           COVERED
nvtracker         COVERED
queue_src         UNINSTRUMENTED — not re-tuned
fakesink          UNINSTRUMENTED — not re-tuned

# Identity + memory + compute
nvidia-smi --query-gpu=name,compute_cap,memory.total,memory.free,\
clocks.max.sm,clocks.max.memory,utilization.gpu \
--format=csv,noheader,nounits

# NVDEC / NVENC utilization (per-engine)
nvidia-smi --query-gpu=utilization.decoder,utilization.encoder \
--format=csv,noheader,nounits

# PCIe link width/gen (for H2D memcpy ceiling)
nvidia-smi --query-gpu=pcie.link.gen.current,pcie.link.width.current \
--format=csv,noheader,nounits

TS=$(date +%Y%m%d_%H%M%S)
OUT=/tmp/ds_profile_${TS}
nsys profile \
  --trace=cuda,nvtx,osrt \
  --gpu-metrics-devices=all \
  --cuda-memory-usage=true \
  --force-overwrite=true \
  --duration=30 \
  --output=${OUT} \
  <your-pipeline-launch-command>

# Per-kernel GPU time (top 10)
nsys stats --report cuda_gpu_kern_sum --format csv ${OUT}.nsys-rep | head -20

# Per-NVTX-range time (top 10) — this is the DS per-plugin breakdown
nsys stats --report nvtx_sum --format csv ${OUT}.nsys-rep | head -20

# Memcpy totals
nsys stats --report cuda_gpu_mem_time_sum --format csv ${OUT}.nsys-rep

# GPU metrics (SM activity, DRAM throughput) — requires --gpu-metrics-devices
nsys stats --report gpu_metric_gpu_util_sum --format csv ${OUT}.nsys-rep

## Profile summary

**Hardware**: <name>, <mem_total> GB, SM x<sm>, NVDEC x<nvdec>, PCIe Gen<g> x<w>
**Ceilings**: decode <X> fps, compute ~<Y> TOPS @ INT8, memory <Z> GB/s

**Inference plateau**: batch=<B>, peak=<F> fps per batch → <F × B> fps aggregate

**E2E measured**: <actual> fps  (=<pct>% of inference plateau)

### Per-plugin time (from NVTX) — only for plugins emitting NVTX in this build

| Plugin          | Share of wall time | GPU / CPU | Notes |
|-----------------|--------------------|-----------|-------|
| nvinfer         | <pct>%             | GPU       | (always emitted; if absent, NVTX injection is broken) |
| nvdsosd         | <pct>%             | GPU       | (when in pipeline) |
| ...             | ...                | ...       | (other plugins as the verification probe shows) |

(Numbers above are illustrative — fill in from `nsys stats --report nvtx_sum`. Plugins
that don't emit NVTX in your DS / image combo simply don't appear; that's not a bug, it's
the limit of what NVTX captures here. See `references/nvtx-coverage.md`.)

### Applied configs (sample shape; values come from R1–R6 + Stage 3 measurements)

- `nvstreammux.batch-size = <plateau_batch>`
- `nvinfer.network-mode = 1 (INT8)` if calibration available, else `2 (FP16)`
- decoder `num-extra-surfaces = min(plateau_batch, 5)`
- queue between source and pgie, `max-size-buffers = plateau_batch × 4`
- ... (full list per the user's pipeline shape)

### Uninstrumented (skipped re-tune)

List the elements that didn't emit NVTX in this build (typically the closed-source binary
plugins — see `references/nvtx-coverage.md`) plus plain GStreamer-core helpers. Report
them so the user knows what wasn't directly measurable.

<!-- Signing refresh marker. -->

© NVIDIA, Apache-2.0. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

SKILL.md and 15 other files (scripts, references) in skills/deepstream-profile-pipeline of NVIDIA/skills.

• SKILL.md
• BENCHMARK.md
• README.md
• evals/evals.json
• references/boundedness-rules.md
• references/config-derivation-rules.md
• references/hw-ceiling-formulas.md
• references/nsys-cli-recipes.md
• references/nvtx-coverage.md
• requirements.txt
• scripts/capacity_report.py
• skill-card.md
• skill.oms.sig
• tests/README.md
• tests/__init__.py
• tests/test_capacity_report.py

Open the folder on GitHubat commit 67a13c0