Performance

This section covers the main performance levers for SGLang Diffusion: attention backends, caching acceleration, and profiling.

Overview

Optimization	Type	Description
Cache-DiT	Caching	Block-level caching with DBCache, TaylorSeer, and SCM
TeaCache	Caching	Timestep-level caching based on temporal similarity
Attention Backends	Kernel	Optimized attention implementations (FlashAttention, SageAttention, etc.)
Profiling	Diagnostics	PyTorch Profiler and Nsight Systems guidance

Start Here

• Use Attention Backends to choose the best backend for your model and hardware.

• Use Deployment Cookbook to choose CPU offload, FSDP, CFG parallelism, SP, and TP.

• Use Caching Acceleration to reduce denoising cost with Cache-DiT or TeaCache.

• Use Profiling when you need to diagnose a bottleneck rather than guess.

Caching at a Glance

• Cache-DiT is block-level caching for diffusers pipelines and higher speedup-oriented tuning.

• TeaCache is timestep-level caching built into SGLang model families.

• Attention Backends
• Deployment Cookbook
• Caching Acceleration
• Profiling Multimodal Generation

Current Baseline Snapshot

For Ring SP benchmark details, see:

• Ring SP Performance

References

• Cache-DiT Repository

• TeaCache Paper