📖 前言
随着 NVIDIA Blackwell 架构的问世，DGX Spark (Personal AI Supercomputer)将桌面级 AI 算力推向了新的巅峰。这台怪兽级设备搭载了GB200/GB10级别的 GPU 和NVIDIA Grace CPU (ARM64)，并运行在最新的CUDA 13环境下。
然而，“最强硬件"往往伴随着"最难环境”。由于 Grace CPU 采用ARM (aarch64)架构，且 CUDA 13 过于前沿，传统的 PyTorch 安装方法极易失败。
本文将手把手教你如何在这台超级计算机上部署Stable Diffusion 3.5 Large，并利用其128GB 海量显存开启高性能模式，实现秒级、零延迟的创作体验。
🛠️ 硬件环境规格 (DGX Spark Specs)
本次部署基于 NVIDIA 最新一代个人 AI 超级计算机：
- Platform: NVIDIA DGX Spark (Personal AI Supercomputer)
- Architecture: NVIDIA Blackwell (GB10 / GB200 Tensor Core GPU)
- VRAM:128GB HBM3e(Unified Memory Architecture)
- CPU:NVIDIA Grace CPU(144-core ARM64)
- Software Stack: NVIDIA AI Enterprise
- CUDA Version:CUDA 13.0
💡 核心挑战与优势：
- 挑战：标准 PyPI 源的 PyTorch 通常只适配 CUDA 11/12，直接安装会导致找不到 GPU。
- 优势：128GB 显存允许我们将 SD3.5 的 20GB 模型 + 10GB T5 编码器永久驻留显存，消除模型加载时间。
📦 第一步：构建兼容 CUDA 13 的基础环境
1.1 创建 Conda 环境
推荐使用 Python 3.11，它在 ARM 架构下的库支持最完善。
conda create-nsd-runtimepython=3.11-yconda activate sd-runtime1.2 安装适配 ARM + CUDA 13 的 PyTorch
这是最关键的一步。由于 CUDA 13 非常新，必须使用NVIDIA 官方 PyPI 索引来获取正确的 wheel 包。
# 1. 确保 pip 是最新的pipinstall--upgradepip# 2. 从 NVIDIA 官方源安装 PyTorch# 注意：NVIDIA 的 index 通常会包含对最新 CUDA 版本的兼容包pipinstalltorch torchvision torchaudio --index-url https://pypi.nvidia.com1.3 验证环境 (关键)
安装完成后，务必验证 PyTorch 是否成功链接到了 Blackwell GPU。
python-c"import torch; print(f'PyTorch Version: {torch.__version__}'); print(f'CUDA Available: {torch.cuda.is_available()}'); print(f'Device Name: {torch.cuda.get_device_name(0)}'); print(f'CUDA Version: {torch.version.cuda}')"- 预期输出：
CUDA Available:TrueDevice Name:NVIDIA GB10(或相关 Blackwell 代号)CUDA Version:13.x
📥 第二步：下载 Stable Diffusion 3.5 Large
SD3.5 是门控模型，请确保您已在 Hugging Face 签署协议并持有 Access Token。
# 1. 安装下载工具pipinstallhuggingface_hub# 2. 配置环境变量 (使用国内镜像加速)exportHF_ENDPOINT=https://hf-mirror.comexportHF_HUB_ENABLE_HF_TRANSFER=1exportHF_TOKEN="你的_HF_Token_粘贴在这里"# 3. 创建目录mkdir-p/home/david/workspaces/models/stabilityai/# 4. 高速下载 (包含 Checkpoint 和 T5/CLIP 编码器)huggingface-cli download stabilityai/stable-diffusion-3.5-large\--token$HF_TOKEN\--local-dir /home/david/workspaces/models/stabilityai/stable-diffusion-3.5-large🖥️ 第三步：部署 ComfyUI & 修复 ARM 依赖
3.1 克隆与安装
cd/home/david/workspacesgitclone https://github.com/comfyanonymous/ComfyUI.gitcdComfyUI# 安装常规依赖 (使用 --no-deps 防止破坏 Torch 环境)pipinstall-rrequirements.txt --no-deps# 🔥 核心修复：补齐 ARM 架构下缺失的特殊库# 如果缺少这些，ComfyUI 启动时会报错 "No module named kornia_rs" 或 "trampoline"pipinstalltrampoline kornia_rs mako comfyui-workflow-templates sqlalchemy alembic requests3.2 挂载模型 (软链接方案)
为了利用 DGX Spark 的存储性能，我们使用软链接映射模型。
# 1. 链接主模型ln-s/home/david/workspaces/models/stabilityai/stable-diffusion-3.5-large/sd3.5_large.safetensors\./models/checkpoints/# 2. 链接 CLIP 模型 (SD3 需要 3 个 CLIP)# 注意：我们直接使用 FP16 精度的 T5，因为显存足够大ln-s/home/david/workspaces/models/stabilityai/stable-diffusion-3.5-large/text_encoders/clip_l.safetensors ./models/clip/ln-s/home/david/workspaces/models/stabilityai/stable-diffusion-3.5-large/text_encoders/clip_g.safetensors ./models/clip/ln-s/home/david/workspaces/models/stabilityai/stable-diffusion-3.5-large/text_encoders/t5xxl_fp16.safetensors ./models/clip/🚀 第四步：高性能启动 (High VRAM Mode)
针对128GB 显存，我们开启"疯狗模式"——让所有模型永久驻留显存，拒绝反复加载。
创建启动脚本run_comfy.sh：
#!/bin/bashWORKDIR="/home/david/workspaces/ComfyUI"PYTHON_PATH="/root/miniconda3/envs/sd-runtime/bin/python"# 日志管理LOG_DIR="$WORKDIR/logs"mkdir-p"$LOG_DIR"LOG_FILE="$LOG_DIR/server_$(date"+%Y-%m-%d_%H-%M-%S").log"cd"$WORKDIR"||exitecho"🚀 Starting ComfyUI on DGX Spark (Blackwell)..."echo"⚡ Mode: High VRAM (Force Resident)"# --highvram: 强制模型常驻显存nohup"$PYTHON_PATH"main.py\--listen0.0.0.0\--port8188\--highvram\>"$LOG_FILE"2>&1&echo"✅ Service PID:$!"启动服务：chmod +x run_comfy.sh && ./run_comfy.sh
🎨 第五步：导入 SD3.5 标准工作流 (JSON)
SD3.5 采用了分离式文本编码器架构（CLIP-L + CLIP-G + T5），默认工作流无法正常出图。
请复制下方完整的 JSON 代码，保存为sd3_workflow.json，直接拖入 ComfyUI 浏览器界面即可使用。此工作流已针对 1024x1024 分辨率和 DGX Spark 进行了优化。
{"last_node_id":10,"last_link_id":16,"nodes":[{"id":1,"type":"KSampler","pos":[960,190],"size":{"0":315,"1":262},"flags":{},"order":0,"mode":0,"inputs":[{"name":"model","type":"MODEL","link":1},{"name":"positive","type":"CONDITIONING","link":2},{"name":"negative","type":"CONDITIONING","link":3},{"name":"latent_image","type":"LATENT","link":4}],"outputs":[{"name":"LATENT","type":"LATENT","links":[5],"slot_index":0}],"properties":{"Node name for S&R":"KSampler"},"widgets_values":[1234567890,"randomize",30,4.5,"euler","sgm_uniform",1]},{"id":2,"type":"CheckpointLoaderSimple","pos":[50,50],"size":{"0":315,"1":98},"flags":{},"order":1,"mode":0,"outputs":[{"name":"MODEL","type":"MODEL","links":[1],"slot_index":0},{"name":"CLIP","type":"CLIP","links":[],"slot_index":1},{"name":"VAE","type":"VAE","links":[6],"slot_index":2}],"properties":{"Node name for S&R":"CheckpointLoaderSimple"},"widgets_values":["sd3.5_large.safetensors"]},{"id":3,"type":"TripleCLIPLoader","pos":[50,200],"size":{"0":315,"1":150},"flags":{},"order":2,"mode":0,"outputs":[{"name":"CLIP","type":"CLIP","links":[7,8],"slot_index":0}],"properties":{"Node name for S&R":"TripleCLIPLoader"},"widgets_values":["clip_l.safetensors","clip_g.safetensors","t5xxl_fp16.safetensors"]},{"id":4,"type":"CLIPTextEncode","pos":[450,100],"size":{"0":425,"1":180},"flags":{},"order":3,"mode":0,"inputs":[{"name":"clip","type":"CLIP","link":7}],"outputs":[{"name":"CONDITIONING","type":"CONDITIONING","links":[2],"slot_index":0}],"properties":{"Node name for S&R":"CLIPTextEncode"},"widgets_values":["a highly detailed cyberpunk city, neon lights, 8k resolution, cinematic lighting, masterpiece, sharp focus, ray tracing, nvidia blackwell style"],"color":"#233","bgcolor":"#353"},{"id":5,"type":"CLIPTextEncode","pos":[450,350],"size":{"0":425,"1":180},"flags":{},"order":4,"mode":0,"inputs":[{"name":"clip","type":"CLIP","link":8}],"outputs":[{"name":"CONDITIONING","type":"CONDITIONING","links":[3],"slot_index":0}],"properties":{"Node name for S&R":"CLIPTextEncode"},"widgets_values":["low quality, blurry, ugly, bad anatomy, watermark, text, jpeg artifacts"],"color":"#332","bgcolor":"#533"},{"id":6,"type":"EmptyLatentImage","pos":[50,450],"size":{"0":315,"1":106},"flags":{},"order":5,"mode":0,"outputs":[{"name":"LATENT","type":"LATENT","links":[4],"slot_index":0}],"properties":{"Node name for S&R":"EmptyLatentImage"},"widgets_values":[1024,1024,1]},{"id":7,"type":"VAEDecode","pos":[1300,190],"size":{"0":210,"1":46},"flags":{},"order":6,"mode":0,"inputs":[{"name":"samples","type":"LATENT","link":5},{"name":"vae","type":"VAE","link":6}],"outputs":[{"name":"IMAGE","type":"IMAGE","links":[9],"slot_index":0}],"properties":{"Node name for S&R":"VAEDecode"}},{"id":8,"type":"SaveImage","pos":[1550,190],"size":{"0":500,"1":600},"flags":{},"order":7,"mode":0,"inputs":[{"name":"images","type":"IMAGE","link":9}],"properties":{"Node name for S&R":"SaveImage"},"widgets_values":["ComfyUI"]}],"links":[[1,2,0,1,0,"MODEL"],[2,4,0,1,1,"CONDITIONING"],[3,5,0,1,2,"CONDITIONING"],[4,6,0,1,3,"LATENT"],[5,1,0,7,0,"LATENT"],[6,2,2,7,1,"VAE"],[7,3,0,4,0,"CLIP"],[8,3,0,5,0,"CLIP"],[9,7,0,8,0,"IMAGE"]],"groups":[],"config":{},"extra":{},"version":0.4}流程如下：
效果如下：
💡 工作流配置说明：
- TripleCLIPLoader: 这是核心。我们明确指定了
clip_l,clip_g和t5xxl_fp16。如果这三个没选对，生成的图片将无法理解你的 Prompt。 - Scheduler: sgm_uniform: Stable Diffusion 3 必须使用 SGM Uniform 调度器，否则画面会充满噪点。
- Resolution: 默认设为 1024x1024。在 Blackwell 上，你可以尝试将其改为 2048x2048 进行原生 4K 生成。
📊 性能总结
在DGX Spark上完成上述部署后，性能表现如下：
- 模型加载: 首次启动后，模型将永久驻留显存(High VRAM Mode)。
- 生成速度: 1024x1024 分辨率下，实现秒级出图。
- 并发能力: 得益于 128GB 显存，你可以同时开启
Batch Size = 4甚至更高，或者在后台同时运行一个 70B 的大语言模型而不发生 OOM。
资源占用：
