优化3D扩散模型在不同GPU上的运行效率，提升速度百倍。
Performance Analysis and Optimization of 3D Generative Diffusion Models across GPU Architectures
- 分析Med-DDPM在NVIDIA多代GPU的底层性能瓶颈
- 通过布局优化与精度提升，实现100倍加速
- 适合关注医疗图像生成与GPU优化的研究者
扩散模型已成为高保真3D MRI合成的关键技术，但其部署受限于每样本需数百次U-Net计算带来的巨大GPU资源消耗，以及高度异构的内核行为。本文针对最先进的医学扩散模型Med-DDPM，对三代NVIDIA GPU架构进行全方位性能分析，涵盖内核级运行时分解、指令混合特征、内存系统利用率、线程组级活动及剖析器优先级评分。结果表明，训练过程主要由cuDNN卷积和隐式GEMM内核主导，效率低下源于内存访问模式、张量布局转换以及Tensor Core利用不足。基于这些洞察，我们评估了两种架构感知优化：TF32 Tensor Core激活与3D channels-last布局。实验显示，该方法可使SM周期减少达100倍，动态指令数降低100倍，Tensor Core利用率从1.45提升至9.98倍，A100上IPC提升7%，且不损失合成质量。
原文摘要 · Abstract (English)
Diffusion models have become essential for high-fidelity 3D MRI synthesis, yet their deployment remains constrained by substantial GPU resource demands arising from hundreds of U-Net evaluations per sample and a highly heterogeneous kernel behavior. This paper performs a comprehensive performance analysis of the state-of-the-art medical diffusion model, Med-DDPM, across three generations of NVIDIA architectures to study kernel-level runtime breakdowns, instruction-mix characteristics, memory system utilization, warp-level activities, and profiler priority-score estimates. We show that training is overwhelmingly dominated by cuDNN convolution and implicit-GEMM kernels, with inefficiencies arising from memory-access patterns, tensor-layout conversions, and limited Tensor Core utilization. Guided by these insights, we evaluate two architecture-aware optimizations TF32 Tensor Core activation and a 3D channels-last layout and demonstrate that they reduce SM cycles by up to 100x, cut dynamic instructions by 100x, raise Tensor Core utilization from 1.45 to 9.98x, and increase IPC by 7% on A100, all without degrading synthesis quality.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。