app.py

import gradio as gr
import torch
from diffusers import AutoPipelineForText2Image
import time

USE_TORCH_COMPILE = False

dtype = torch.float16
device = torch.device("cuda:0")

pipeline = AutoPipelineForText2Image.from_pretrained("stabilityai/stable-diffusion-xl-base-1.0", variant="fp16", torch_dtype=torch.float16)
pipeline.to("cuda")


if USE_TORCH_COMPILE:
    pipeline.unet = torch.compile(pipeline.unet, mode="reduce-overhead", fullgraph=True)
    
def generate(prompt_len: int, num_images_per_prompt: int = 1):
    prompt = prompt_len * "a"
    num_inference_steps = 40
    start_time = time.time()
    pipeline(prompt, num_images_per_prompt=num_images_per_prompt, num_inference_steps=num_inference_steps).images
    end_time = time.time()


    print(f"For {num_inference_steps} steps", end_time - start_time)
    print("Avg per step", (end_time - start_time) / num_inference_steps)


with gr.Blocks(css="style.css") as demo:
    batch_size = gr.Slider(
        label="Batch size",
        minimum=0,
        maximum=8,
        step=1,
        value=0,
    )
    prompt_len = gr.Slider(
        label="Prompt len",
        minimum=1,
        maximum=77,
        step=20,
        value=1,
    )
    btn = gr.Button("Benchmark!").style(
        margin=False,
        rounded=(False, True, True, False),
        full_width=False,
    )

    btn.click(fn=generate, inputs=[batch_size, prompt_len])