Runware Accelerator Options

Runware Accelerator Options

Configure Runware accelerator options for faster image generation. Supports TeaCache (transformer models like Flux/SD3) and DeepCache (UNet models like SDXL/SD1.5) with fine-grained control over caching parameters and step ranges. Note: These caching methods work best with deterministic schedulers (Euler, DDIM) rather than stochastic schedulers.

Pack: Runware.ai ComfyUI Inference API Integration

custom_nodes.ComfyUI-Runware

Inputs (18)

NameTypeRequired
fbcacheBOOLEANoptional
useCacheDistanceBOOLEANoptional
cacheDistanceFLOAToptional
teaCacheBOOLEANoptional
useTeaCacheDistanceBOOLEANoptional
teaCacheDistanceFLOAToptional
deepCacheBOOLEANoptional
useDeepCacheOptionsBOOLEANoptional
deepCacheIntervalINToptional
deepCacheBranchIdINToptional
useCacheStepsBOOLEANoptional
cacheStartStepINToptional
cacheEndStepINToptional
useCachePercentageStepsBOOLEANoptional
cacheStartStepPercentageINToptional
cacheEndStepPercentageINToptional
useCacheMaxConsecutiveStepsBOOLEANoptional
cacheMaxConsecutiveStepsINToptional

Outputs (1)

NameType
Runware AcceleratorRUNWAREACCELERATOR