Issue about inference time in v1.3

[WangRongsheng]

When I run open sora plan (v1.3), for each video it takes about 11 minutes to generate it in 1*NVIDIA A100(80GB). Is it normal?

CUDA_VISIBLE_DEVICES=0 torchrun --nnodes=1 --nproc_per_node 1 --master_port 29514 \
    -m opensora.sample.sample \
    --model_path "/root/autodl-tmp/Open-Sora-Plan-1.3.1/Open-Sora-Plan-v1.3.0/any93x640x640" \
    --version v1_3 \
    --num_frames 93 \
    --height 640 \
    --width 640 \
    --cache_dir "../cache_dir" \
    --text_encoder_name_1 "/root/autodl-tmp/Open-Sora-Plan-1.3.1/model/hub/models--hf-models--mt5-xxl/snapshots/e07c395916dfbc315d4e5e48b4a54a1e8821b5c0" \
    --text_prompt "examples/medsora_prompt_output.txt" \
    --ae WFVAEModel_D8_4x8x8 \
    --ae_path "/root/autodl-tmp/Open-Sora-Plan-1.3.1/Open-Sora-Plan-v1.3.0/vae" \
    --save_img_path "./train_1_3_nomotion_fps18" \
    --fps 18 \
    --guidance_scale 7.5 \
    --num_sampling_steps 100 \
    --max_sequence_length 512 \
    --sample_method EulerAncestralDiscrete \
    --seed 42 \
    --num_samples_per_prompt 1 \
    --rescale_betas_zero_snr \
    --prediction_type "v_prediction"