17 lines
462 B
Markdown
17 lines
462 B
Markdown
|
# Quantization for Diffusion Transformers
|
||
|
|
|
||
|
|
vLLM-Omni supports quantization of DiT linear layers to reduce memory usage and accelerate inference.
|
||
|
|
|
||
|
|
## Supported Methods
|
||
|
|
|
||
|
|
| Method | Guide |
|
||
|
|
|--------|-------|
|
||
|
|
| FP8 | [FP8](fp8.md) |
|
||
|
|
|
||
|
|
## Device Compatibility
|
||
|
|
|
||
|
|
| GPU Generation | Example GPUs | FP8 Mode |
|
||
|
|
|---------------|-------------------|----------|
|
||
|
|
| Ada/Hopper (SM 89+) | RTX 4090, H100, H200 | Full W8A8 with native hardware |
|
||
|
|
|
||
|
|
Kernel selection is automatic.
|