ComfyUI Node: INT8 Lazy Torch Compile

Authored by SparknightLLC

Created

Updated

44 stars

Run ComfyUI workflows without the setup

No installs, no CUDA version roulette, no GPU sitting idle on your bill. Bring a workflow and run it in the browser.

Category

loaders

Inputs

model MODEL
backend
  • inductor
  • cudagraphs
fullgraph BOOLEAN
mode
  • default
  • max-autotune
  • max-autotune-no-cudagraphs
  • reduce-overhead
dynamic
  • auto
  • true
  • false
compile_transformer_blocks_only BOOLEAN
dynamo_cache_size_limit INT
use_guard_filter BOOLEAN
disable_dynamic_vram BOOLEAN
log_compile BOOLEAN

Outputs

MODEL

Extension: ComfyUI-INT8-Fast-Fork

Fork of node to load models in INT8 for 1.5~2X Speed gains on 30 series cards. Contains additional fixes and performance improvements.

Authored by SparknightLLC

Looking for a different node?

Run ComfyUI workflows without the setup

No installs, no CUDA version roulette, no GPU sitting idle on your bill. Bring a workflow and run it in the browser.

Learn more