원클릭으로 Manus에서 모든 스킬 실행

edge-fm-orin-docker-build

스타31

포크3

업데이트2026년 4월 8일 14:43

Build edge-fm for Jetson Orin inside a mounted Docker workspace using `nvcr.io/nvidia/l4t-jetpack:r36.4.0`. Use when Codex needs to validate Orin compilation, reproduce a JetPack 6.1 build environment, or produce a repeatable Docker-based Orin build command without relying on Thor.

설치

Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.

Manus에서 실행

출처

windog-labs

windog-labs/edge-fm-x

GitHub 저장소 열기 Creator 저장소 보기

다운로드

Manus에서 실행

edge-fm Orin Docker Build

这个 skill 只用于 Orin，不要把 thor 当成替代平台。默认平台固定为 -DPLATFORM=orin，镜像固定为 nvcr.io/nvidia/l4t-jetpack:r36.4.0。

什么时候用

用户要验证 edge-fm 能否在 Jetson Orin / JetPack 6.1 环境下编译通过。
用户要一个不污染宿主机的挂载式 Docker 构建流程。
用户要复现 Orin 的 CUDA / cuDNN / TensorRT 头文件和库路径。

工作流

先确认镜像已可用；如果本地没有，再执行：
```
docker pull nvcr.io/nvidia/l4t-jetpack:r36.4.0
```

使用挂载本地目录的方式进入容器，工作目录固定到仓库根目录：

export EDGE_FM_BUILD_JOBS="${EDGE_FM_BUILD_JOBS:-1}"
docker run --rm --platform linux/arm64 \
  -v "$PWD":/workspace/edge-fm \
  -w /workspace/edge-fm \
  nvcr.io/nvidia/l4t-jetpack:r36.4.0 \
  bash -lc 'apt-get update && DEBIAN_FRONTEND=noninteractive apt-get install -y cmake ninja-build && cmake -S . -B build-orin -G Ninja -DPLATFORM=orin -DCMAKE_BUILD_TYPE=Release && cmake --build build-orin --parallel "${EDGE_FM_BUILD_JOBS}"'

如果需要 Python 绑定安装产物，再补：

export EDGE_FM_BUILD_JOBS="${EDGE_FM_BUILD_JOBS:-1}"
docker run --rm --platform linux/arm64 \
  -v "$PWD":/workspace/edge-fm \
  -w /workspace/edge-fm \
  nvcr.io/nvidia/l4t-jetpack:r36.4.0 \
  bash -lc 'apt-get update && DEBIAN_FRONTEND=noninteractive apt-get install -y cmake ninja-build && cmake -S . -B build-orin -G Ninja -DPLATFORM=orin -DCMAKE_BUILD_TYPE=Release && cmake --build build-orin --parallel "${EDGE_FM_BUILD_JOBS}" && cmake --install build-orin'

实施要求

只使用挂载目录，不要把源码 COPY 进镜像。
只构建 orin，不要改成 thor，也不要用 thor 的编译选项代替 orin。
配置阶段必须显式传 -DPLATFORM=orin。
默认构建目录使用 build-orin/，避免污染通用 build/。
依赖安装只补容器内缺失的基础工具，至少包含 cmake 和 ninja-build。
默认并发建议从 EDGE_FM_BUILD_JOBS=1 起步；Jetson/JetPack 容器内某些 CUDA 模板单元内存开销很大，先保证稳定通过，再按机器内存上调。
如果需要排查 CUDA / cuDNN 发现问题，优先检查：
- /usr/local/cuda/targets/aarch64-linux/include
- /usr/lib/aarch64-linux-gnu
- /usr/include/aarch64-linux-gnu

验证

先看 cmake -S . -B build-orin -DPLATFORM=orin 是否通过。
再看 cmake --build build-orin --parallel 是否完成。
若用户要求安装产物，确认 build-orin/install/ 下出现 lib/ 与 python/。

이 저장소의 다른 Skills

같은 저장소

edge-fm-cuda-kernel-optimizer

windog-labs/edge-fm-x

NCU-guided CUDA/CUTLASS kernel optimization workflow for edge-fm-x, including short standalone/repo-in-place tuning and long Humanize + KernelPilot loops. Use when optimizing kernels under src/, tuning standalone CUDA/CUTLASS reproductions, or when the user asks to speed up an Edge-FM CUDA kernel and mentions Nsight Compute, Nsight Systems, NCU, benchmark plateaus, Humanize, KernelPilot, long-running kernel search, source provenance, ledgers, CUTLASS, latency regressions, or iterative optimization.

2026-05-1831

cuda-skill

windog-labs/edge-fm-x

Query NVIDIA PTX ISA 9.1, CUDA Runtime API 13.1, Driver API 13.1, Programming Guide v13.1, Best Practices Guide, Nsight Compute, Nsight Systems local documentation. Debug and optimize GPU kernels with nsys/ncu/compute-sanitizer workflows. Includes hardware quick guides for T4, A100, H100, A800, RTX 3060, and Jetson Orin. Use when writing, debugging, or optimizing CUDA code, GPU kernels, PTX instructions, inline PTX, TensorCore operations (WMMA, WGMMA, TMA, tcgen05), or when the user mentions CUDA API functions, error codes, device properties, memory management, profiling, GPU performance, compute capabilities, CUDA Graphs, Cooperative Groups, Unified Memory, dynamic parallelism, or CUDA programming model concepts.

2026-05-1531

cutlass-skill

windog-labs/edge-fm-x

Develop, debug, and optimize CUTLASS C++, CuTe C++, and CuTe DSL kernels using the local third_party/cutlass repository in this workspace. Use when the user mentions CUTLASS, CuTe, CuTe DSL, cute::Layout, cute::Tensor, TiledMMA, TiledCopy, CollectiveBuilder, CollectiveMainloop, CollectiveEpilogue, GEMM kernel, grouped GEMM, sparse GEMM, FMHA, StreamK, warp specialization, TMA, WGMMA, tcgen05, blockscaled GEMM, or asks about writing or understanding high-performance CUDA kernels with CUTLASS/CuTe.

2026-05-1531

edge-fm-benchmark-report

windog-labs/edge-fm-x

Generate detailed benchmark reports inside this edge-fm repo, covering configurable Qwen2.5 LLM/VLM model matrices on CUDA. The default fair comparison remains Transformers vs EdgeFM(cuda-graph) vs TRT-Edge-LLM for LLM, while VLM currently uses Transformers vs EdgeFM(cuda-graph). Use when the user asks for benchmark指标、性能对比、详细报告、cuda-graph vs TRT-Edge-LLM analysis, VLM benchmark reports, or wants the existing tests/engine/test_qwen2_generate.py benchmark flow wrapped into a repeatable local workflow.

2026-05-1531

humanize-kernel-agent-loop

windog-labs/edge-fm-x

Run an end-to-end Humanize Kernel Agent Loop for GPU kernel optimization: plan, refine, create a clean standalone repo, use kernel-knowledge, benchmark, profile with Nsight Compute, maintain lineage/ledgers, and start RLCR.

2026-05-1531

kernel-knowledge

windog-labs/edge-fm-x

Use when optimizing GPU kernels with Humanize or Codex and the agent should consult the local kernel knowledge pack before planning, choosing optimization directions, or doing plateau research. Covers code-first framework/topic lookup, source provenance, baseline-derived candidates, and user-directed kernel stacks.

2026-05-1531

name	edge-fm-orin-docker-build
description	Build edge-fm for Jetson Orin inside a mounted Docker workspace using `nvcr.io/nvidia/l4t-jetpack:r36.4.0`. Use when Codex needs to validate Orin compilation, reproduce a JetPack 6.1 build environment, or produce a repeatable Docker-based Orin build command without relying on Thor.

edge-fm Orin Docker Build

这个 skill 只用于 Orin，不要把 thor 当成替代平台。默认平台固定为 -DPLATFORM=orin，镜像固定为 nvcr.io/nvidia/l4t-jetpack:r36.4.0。

什么时候用

用户要验证 edge-fm 能否在 Jetson Orin / JetPack 6.1 环境下编译通过。
用户要一个不污染宿主机的挂载式 Docker 构建流程。
用户要复现 Orin 的 CUDA / cuDNN / TensorRT 头文件和库路径。

工作流

先确认镜像已可用；如果本地没有，再执行：
```
docker pull nvcr.io/nvidia/l4t-jetpack:r36.4.0
```

使用挂载本地目录的方式进入容器，工作目录固定到仓库根目录：

export EDGE_FM_BUILD_JOBS="${EDGE_FM_BUILD_JOBS:-1}"
docker run --rm --platform linux/arm64 \
  -v "$PWD":/workspace/edge-fm \
  -w /workspace/edge-fm \
  nvcr.io/nvidia/l4t-jetpack:r36.4.0 \
  bash -lc 'apt-get update && DEBIAN_FRONTEND=noninteractive apt-get install -y cmake ninja-build && cmake -S . -B build-orin -G Ninja -DPLATFORM=orin -DCMAKE_BUILD_TYPE=Release && cmake --build build-orin --parallel "${EDGE_FM_BUILD_JOBS}"'

如果需要 Python 绑定安装产物，再补：

export EDGE_FM_BUILD_JOBS="${EDGE_FM_BUILD_JOBS:-1}"
docker run --rm --platform linux/arm64 \
  -v "$PWD":/workspace/edge-fm \
  -w /workspace/edge-fm \
  nvcr.io/nvidia/l4t-jetpack:r36.4.0 \
  bash -lc 'apt-get update && DEBIAN_FRONTEND=noninteractive apt-get install -y cmake ninja-build && cmake -S . -B build-orin -G Ninja -DPLATFORM=orin -DCMAKE_BUILD_TYPE=Release && cmake --build build-orin --parallel "${EDGE_FM_BUILD_JOBS}" && cmake --install build-orin'

实施要求

只使用挂载目录，不要把源码 COPY 进镜像。
只构建 orin，不要改成 thor，也不要用 thor 的编译选项代替 orin。
配置阶段必须显式传 -DPLATFORM=orin。
默认构建目录使用 build-orin/，避免污染通用 build/。
依赖安装只补容器内缺失的基础工具，至少包含 cmake 和 ninja-build。
默认并发建议从 EDGE_FM_BUILD_JOBS=1 起步；Jetson/JetPack 容器内某些 CUDA 模板单元内存开销很大，先保证稳定通过，再按机器内存上调。
如果需要排查 CUDA / cuDNN 发现问题，优先检查：
- /usr/local/cuda/targets/aarch64-linux/include
- /usr/lib/aarch64-linux-gnu
- /usr/include/aarch64-linux-gnu

验证

先看 cmake -S . -B build-orin -DPLATFORM=orin 是否通过。
再看 cmake --build build-orin --parallel 是否完成。
若用户要求安装产物，确认 build-orin/install/ 下出现 lib/ 与 python/。