Inference time saving for 1024 * 1024

[phyllispeng123]

Thanks for your excellent work on Flux, I see the flops saved in the table, can you share the inference time saving in GPU in resolution=1024 * 1024 ?

[Huage001]

Please refer to Tab. 7 for the time comparison. Actually, CLEAR may not produce acceleration at 1024 resolution and is recommended to use for larger scales. Please refer to #1 for more discussion.