Nodes/ComfyUI-TorchCompileSpeed/Torch Compile Speed Settings
ComfyUI Node

Torch Compile Speed Settings

torch.compile settings for maximum speed optimization. Speed Mode Features: - Uses inductor backend with max-autotune-no-cudagraphs - Enables dynamic compilation for better cache reuse - Disables CUDA graphs for flexibility - Enables all Triton autotune optimizations - First run: comprehensive autotune (slower) - Second run: cached execution (extremely fast) Author: eddy

By eddyhhlure1Eddy·Created 10 months ago·Updated 10 months ago· 24
Torch Compile Speed Settings
    • torch_compile_args
    backendinductor
    fullgraphfalse
    modespeed
    dynamicfalse
    dynamo_cache_size_limit64
    compile_transformer_blocks_onlytrue
    reuse_if_similartrue
    experimental_ptxfalse
    ptx_fast_mathtrue
    warmup_runs1
    ptx_cache_dir
    dynamo_recompile_limit128
    Categoryoptimization

    Inputs (12)

    NameTypeDefaultDescription
    backendCOMBOinductor2 options: inductor, cudagraphs
    fullgraphBOOLEANfalseEnable full graph mode
    modeCOMBOspeed5 options: default, max-autotune, max-autotune-no-cudagraphs, reduce-overhead, speed
    dynamicBOOLEANfalseEnable dynamic mode
    dynamo_cache_size_limitINT640–1024torch._dynamo.config.cache_size_limit
    compile_transformer_blocks_onlyBOOLEANtrueCompile only transformer blocks
    reuse_if_similarBOOLEANtrueReuse compiled artifacts when similar
    experimental_ptxBOOLEANfalseEnable experimental PTX acceleration
    ptx_fast_mathBOOLEANtrueEnable fast math if available
    warmup_runsINT10–5
    ptx_cache_diroptSTRING
    dynamo_recompile_limitoptINT1280–1024torch._dynamo.config.recompile_limit

    Outputs (1)

    NameTypeDescription
    torch_compile_argsWANCOMPILEARGS