+ source /admin/home/sayak/.bashrc
++ alias 'idlenodes=sinfo --Node --long | grep idle | wc -l'
++ alias 'devnode=srun --nodes=1 --cpus-per-task=96 --gres=gpu:8 --exclusive --partition=dev-cluster --pty bash'
++ alias 'myjobs=squeue --format="%.18i %.9P %.50j %.8u %.8T %.10M %.9l %.6D %R" --me'
++ export PATH=/opt/slurm/bin:/opt/slurm/sbin:/admin/home/sayak/miniconda3/envs/diffusers/bin:/opt/slurm/bin:/opt/slurm/sbin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/snap/bin
++ PATH=/opt/slurm/bin:/opt/slurm/sbin:/admin/home/sayak/miniconda3/envs/diffusers/bin:/opt/slurm/bin:/opt/slurm/sbin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/snap/bin
++ case $- in
++ return
+ source /admin/home/sayak/miniconda3/etc/profile.d/conda.sh
++ export CONDA_EXE=/admin/home/sayak/miniconda3/bin/conda
++ CONDA_EXE=/admin/home/sayak/miniconda3/bin/conda
++ export _CE_M=
++ _CE_M=
++ export _CE_CONDA=
++ _CE_CONDA=
++ export CONDA_PYTHON_EXE=/admin/home/sayak/miniconda3/bin/python
++ CONDA_PYTHON_EXE=/admin/home/sayak/miniconda3/bin/python
++ '[' -z x ']'
+ conda activate /fsx/sayak
+ local cmd=activate
+ case "$cmd" in
+ __conda_activate activate /fsx/sayak
+ '[' -n '' ']'
+ local ask_conda
++ PS1=
++ __conda_exe shell.posix activate /fsx/sayak
++ /admin/home/sayak/miniconda3/bin/conda shell.posix activate /fsx/sayak
+ ask_conda='PS1='\''(/fsx/sayak) '\''
export PATH='\''/opt/slurm/bin:/opt/slurm/sbin:/fsx/sayak/bin:/opt/slurm/bin:/opt/slurm/sbin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/snap/bin'\''
export CONDA_PREFIX='\''/fsx/sayak'\''
export CONDA_SHLVL='\''4'\''
export CONDA_DEFAULT_ENV='\''/fsx/sayak'\''
export CONDA_PROMPT_MODIFIER='\''(/fsx/sayak) '\''
export CONDA_PREFIX_3='\''/admin/home/sayak/miniconda3/envs/diffusers'\''
export CONDA_EXE='\''/admin/home/sayak/miniconda3/bin/conda'\''
export _CE_M='\'''\''
export _CE_CONDA='\'''\''
export CONDA_PYTHON_EXE='\''/admin/home/sayak/miniconda3/bin/python'\'''
+ eval 'PS1='\''(/fsx/sayak) '\''
export PATH='\''/opt/slurm/bin:/opt/slurm/sbin:/fsx/sayak/bin:/opt/slurm/bin:/opt/slurm/sbin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/snap/bin'\''
export CONDA_PREFIX='\''/fsx/sayak'\''
export CONDA_SHLVL='\''4'\''
export CONDA_DEFAULT_ENV='\''/fsx/sayak'\''
export CONDA_PROMPT_MODIFIER='\''(/fsx/sayak) '\''
export CONDA_PREFIX_3='\''/admin/home/sayak/miniconda3/envs/diffusers'\''
export CONDA_EXE='\''/admin/home/sayak/miniconda3/bin/conda'\''
export _CE_M='\'''\''
export _CE_CONDA='\'''\''
export CONDA_PYTHON_EXE='\''/admin/home/sayak/miniconda3/bin/python'\'''
++ PS1='(/fsx/sayak) '
++ export PATH=/opt/slurm/bin:/opt/slurm/sbin:/fsx/sayak/bin:/opt/slurm/bin:/opt/slurm/sbin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/snap/bin
++ PATH=/opt/slurm/bin:/opt/slurm/sbin:/fsx/sayak/bin:/opt/slurm/bin:/opt/slurm/sbin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/snap/bin
++ export CONDA_PREFIX=/fsx/sayak
++ CONDA_PREFIX=/fsx/sayak
++ export CONDA_SHLVL=4
++ CONDA_SHLVL=4
++ export CONDA_DEFAULT_ENV=/fsx/sayak
++ CONDA_DEFAULT_ENV=/fsx/sayak
++ export 'CONDA_PROMPT_MODIFIER=(/fsx/sayak) '
++ CONDA_PROMPT_MODIFIER='(/fsx/sayak) '
++ export CONDA_PREFIX_3=/admin/home/sayak/miniconda3/envs/diffusers
++ CONDA_PREFIX_3=/admin/home/sayak/miniconda3/envs/diffusers
++ export CONDA_EXE=/admin/home/sayak/miniconda3/bin/conda
++ CONDA_EXE=/admin/home/sayak/miniconda3/bin/conda
++ export _CE_M=
++ _CE_M=
++ export _CE_CONDA=
++ _CE_CONDA=
++ export CONDA_PYTHON_EXE=/admin/home/sayak/miniconda3/bin/python
++ CONDA_PYTHON_EXE=/admin/home/sayak/miniconda3/bin/python
+ __conda_hashr
+ '[' -n '' ']'
+ '[' -n '' ']'
+ hash -r
++ date
+ echo 'START TIME: Wed Dec 20 11:30:51 UTC 2023'
START TIME: Wed Dec 20 11:30:51 UTC 2023
+ REPO=/fsx/sayak/sdxl-fast
+ OUTPUT_DIR=/fsx/sayak/benchmarking
+ LOG_PATH=/fsx/sayak/benchmarking/main_log.txt
+ mkdir -p /fsx/sayak/benchmarking
+ touch /fsx/sayak/benchmarking/main_log.txt
+ pushd /fsx/sayak/sdxl-fast
/fsx/sayak/sdxl-fast /fsx/sayak
+ PROGRAM=run_specific.sh
+ export 'LAUNCHER=CUDA_VISIBLE_DEVICES=0 bash'
+ LAUNCHER='CUDA_VISIBLE_DEVICES=0 bash'
+ export 'CMD=CUDA_VISIBLE_DEVICES=0 bash run_specific.sh'
+ CMD='CUDA_VISIBLE_DEVICES=0 bash run_specific.sh'
+ echo CUDA_VISIBLE_DEVICES=0 bash run_specific.sh
CUDA_VISIBLE_DEVICES=0 bash run_specific.sh
+ export NCCL_PROTO=simple
+ NCCL_PROTO=simple
+ export RDMAV_FORK_SAFE=1
+ RDMAV_FORK_SAFE=1
+ export FI_EFA_FORK_SAFE=1
+ FI_EFA_FORK_SAFE=1
+ export FI_EFA_USE_DEVICE_RDMA=1
+ FI_EFA_USE_DEVICE_RDMA=1
+ export FI_PROVIDER=efa
+ FI_PROVIDER=efa
+ export FI_LOG_LEVEL=1
+ FI_LOG_LEVEL=1
+ export NCCL_IB_DISABLE=1
+ NCCL_IB_DISABLE=1
+ export NCCL_SOCKET_IFNAME=ens
+ NCCL_SOCKET_IFNAME=ens
+ clear
+ srun --wait=60 --jobid 938570 bash -c 'CUDA_VISIBLE_DEVICES=0 bash run_specific.sh'
+ tee /fsx/sayak/benchmarking/main_log.txt
/fsx/sayak/lib/python3.9/site-packages/transformers/utils/generic.py:441: UserWarning: torch.utils._pytree._register_pytree_node is deprecated. Please use torch.utils._pytree.register_pytree_node instead.
  _torch_pytree._register_pytree_node(
/fsx/sayak/lib/python3.9/site-packages/transformers/utils/generic.py:309: UserWarning: torch.utils._pytree._register_pytree_node is deprecated. Please use torch.utils._pytree.register_pytree_node instead.
  _torch_pytree._register_pytree_node(
/fsx/sayak/diffusers/src/diffusers/utils/outputs.py:63: UserWarning: torch.utils._pytree._register_pytree_node is deprecated. Please use torch.utils._pytree.register_pytree_node instead.
  torch.utils._pytree._register_pytree_node(
Namespace(no_bf16=False, no_sdpa=False, batch_size=1, num_inference_steps=30, enable_fused_projections=True, upcast_vae=False, compile_unet=True, compile_vae=True, compile_mode='max-autotune', change_comp_config=True, do_quant='int8dynamic', tag='branch12-1-select')
Using dtype: torch.bfloat16

Loading pipeline components...:   0%|          | 0/7 [00:00<?, ?it/s]
Loading pipeline components...:  14%|█▍        | 1/7 [00:00<00:01,  5.95it/s]
Loading pipeline components...:  43%|████▎     | 3/7 [00:02<00:03,  1.29it/s]
Loading pipeline components...:  71%|███████▏  | 5/7 [00:16<00:08,  4.19s/it]
Loading pipeline components...:  86%|████████▌ | 6/7 [01:16<00:19, 19.13s/it]
Loading pipeline components...: 100%|██████████| 7/7 [01:19<00:00, 14.62s/it]
Loading pipeline components...: 100%|██████████| 7/7 [01:19<00:00, 11.38s/it]
Enabling fused QKV projections for both UNet and VAE.
Compile UNet
Apply quantization to UNet
Compile VAE
Apply quantization to VAE
AUTOTUNE addmm(8192x640, 8192x320, 320x640)
  triton_mm_130 0.0330 ms 100.0%
  triton_mm_129 0.0339 ms 97.2%
  triton_mm_132 0.0358 ms 92.0%
  bias_addmm 0.0364 ms 90.5%
  triton_mm_131 0.0374 ms 88.1%
  triton_mm_128 0.0412 ms 80.0%
  triton_mm_136 0.0415 ms 79.4%
  triton_mm_135 0.0470 ms 70.1%
  addmm 0.0531 ms 62.0%
  triton_mm_138 0.0633 ms 52.1%
SingleProcess AUTOTUNE takes 5.7408 seconds
AUTOTUNE addmm(2048x1280, 2048x640, 640x1280)
  triton_mm_579 0.0281 ms 100.0%
  triton_mm_580 0.0287 ms 98.1%
  bias_addmm 0.0288 ms 97.7%
  triton_mm_582 0.0309 ms 90.9%
  triton_mm_581 0.0318 ms 88.5%
  triton_mm_586 0.0353 ms 79.6%
  addmm 0.0385 ms 73.1%
  triton_mm_578 0.0396 ms 71.1%
  triton_mm_585 0.0478 ms 58.8%
  triton_mm_588 0.0509 ms 55.2%
SingleProcess AUTOTUNE takes 4.5000 seconds
AUTOTUNE addmm(2048x1280, 2048x2560, 2560x1280)
  bias_addmm 0.0736 ms 100.0%
  triton_mm_3098 0.0792 ms 92.9%
  triton_mm_3097 0.0797 ms 92.4%
  addmm 0.0858 ms 85.8%
  triton_mm_3100 0.0897 ms 82.1%
  triton_mm_3099 0.0899 ms 81.9%
  triton_mm_3104 0.1010 ms 72.9%
  triton_mm_3096 0.1226 ms 60.0%
  triton_mm_3103 0.1541 ms 47.8%
  triton_mm_3106 0.1731 ms 42.5%
SingleProcess AUTOTUNE takes 4.7145 seconds
AUTOTUNE addmm(2048x1280, 2048x1920, 1920x1280)
  bias_addmm 0.0587 ms 100.0%
  triton_mm_4776 0.0610 ms 96.2%
  triton_mm_4775 0.0612 ms 95.8%
  addmm 0.0692 ms 84.8%
  triton_mm_4778 0.0703 ms 83.4%
  triton_mm_4777 0.0704 ms 83.4%
  triton_mm_4782 0.0787 ms 74.6%
  triton_mm_4774 0.0946 ms 62.0%
  triton_mm_4781 0.1177 ms 49.9%
  triton_mm_4784 0.1301 ms 45.1%
SingleProcess AUTOTUNE takes 4.5449 seconds
AUTOTUNE addmm(8192x640, 8192x1920, 1920x640)
  bias_addmm 0.1110 ms 100.0%
  triton_mm_5623 0.1157 ms 95.9%
  triton_mm_5622 0.1205 ms 92.1%
  triton_mm_5624 0.1285 ms 86.4%
  triton_mm_5625 0.1298 ms 85.5%
  addmm 0.1319 ms 84.1%
  triton_mm_5629 0.1490 ms 74.5%
  triton_mm_5621 0.1710 ms 64.9%
  triton_mm_5628 0.1872 ms 59.3%
  triton_mm_5630 0.2678 ms 41.5%
SingleProcess AUTOTUNE takes 4.6234 seconds
AUTOTUNE addmm(8192x640, 8192x1280, 1280x640)
  bias_addmm 0.0811 ms 100.0%
  triton_mm_5851 0.0831 ms 97.5%
  triton_mm_5850 0.0834 ms 97.2%
  triton_mm_5852 0.0912 ms 88.9%
  triton_mm_5853 0.0921 ms 88.0%
  addmm 0.1013 ms 80.0%
  triton_mm_5857 0.1049 ms 77.2%
  triton_mm_5849 0.1151 ms 70.4%
  triton_mm_5856 0.1329 ms 61.0%
  triton_mm_5858 0.1820 ms 44.5%
SingleProcess AUTOTUNE takes 4.6530 seconds
AUTOTUNE addmm(8192x640, 8192x960, 960x640)
  triton_mm_6079 0.0652 ms 100.0%
  triton_mm_6078 0.0655 ms 99.6%
  bias_addmm 0.0662 ms 98.6%
  triton_mm_6080 0.0712 ms 91.6%
  triton_mm_6081 0.0723 ms 90.2%
  triton_mm_6085 0.0829 ms 78.7%
  addmm 0.0844 ms 77.3%
  triton_mm_6077 0.0911 ms 71.6%
  triton_mm_6084 0.1045 ms 62.4%
  triton_mm_6086 0.1430 ms 45.6%
SingleProcess AUTOTUNE takes 4.7932 seconds
AUTOTUNE addmm(32768x320, 32768x960, 960x320)
  bias_addmm 0.1209 ms 100.0%
  triton_mm_6313 0.1262 ms 95.8%
  triton_mm_6315 0.1382 ms 87.5%
  triton_mm_6312 0.1467 ms 82.4%
  triton_mm_6319 0.1561 ms 77.4%
  triton_mm_6314 0.1634 ms 74.0%
  addmm 0.1645 ms 73.5%
  triton_mm_6311 0.1660 ms 72.8%
  triton_mm_6318 0.2311 ms 52.3%
  triton_mm_6321 0.2683 ms 45.0%
SingleProcess AUTOTUNE takes 4.8578 seconds
AUTOTUNE addmm(32768x320, 32768x640, 640x320)
  bias_addmm 0.0897 ms 100.0%
  triton_mm_6350 0.0931 ms 96.4%
  triton_mm_6352 0.1026 ms 87.5%
  triton_mm_6349 0.1099 ms 81.7%
  triton_mm_6356 0.1154 ms 77.8%
  triton_mm_6348 0.1200 ms 74.8%
  triton_mm_6351 0.1215 ms 73.8%
  addmm 0.1336 ms 67.2%
  triton_mm_6355 0.1651 ms 54.3%
  triton_mm_6357 0.1983 ms 45.3%
SingleProcess AUTOTUNE takes 4.5428 seconds
AUTOTUNE addmm(16384x4, 16384x4, 4x4)
  triton_mm_6411 0.0082 ms 100.0%
  triton_mm_6415 0.0082 ms 100.0%
  triton_mm_6412 0.0084 ms 97.7%
  triton_mm_6414 0.0084 ms 97.3%
  triton_mm_6410 0.0087 ms 94.1%
  triton_mm_6418 0.0088 ms 92.4%
  triton_mm_6416 0.0089 ms 91.4%
  triton_mm_6413 0.0090 ms 91.1%
  triton_mm_6419 0.0091 ms 89.5%
  triton_mm_6417 0.0092 ms 88.9%
SingleProcess AUTOTUNE takes 2.7235 seconds
AUTOTUNE convolution(1x4x128x128, 512x4x3x3)
  triton_convolution_6420 0.0340 ms 100.0%
  triton_convolution_6421 0.0362 ms 94.2%
  triton_convolution_6425 0.0363 ms 93.7%
  triton_convolution_6426 0.0369 ms 92.3%
  triton_convolution_6423 0.0370 ms 92.0%
  triton_convolution_6424 0.0412 ms 82.6%
  triton_convolution_6422 0.0754 ms 45.2%
  convolution 0.0967 ms 35.2%
SingleProcess AUTOTUNE takes 3.2605 seconds
AUTOTUNE addmm(262144x256, 262144x512, 512x256)
  bias_addmm 0.3396 ms 100.0%
  triton_mm_6586 0.4330 ms 78.4%
  triton_mm_6585 0.4331 ms 78.4%
  triton_mm_6588 0.4779 ms 71.1%
  triton_mm_6587 0.4850 ms 70.0%
  addmm 0.5526 ms 61.5%
  triton_mm_6592 0.5604 ms 60.6%
  triton_mm_6584 0.5750 ms 59.1%
  triton_mm_6591 0.6858 ms 49.5%
  triton_mm_6594 0.8957 ms 37.9%
SingleProcess AUTOTUNE takes 4.6345 seconds
/fsx/sayak/lib/python3.9/site-packages/transformers/utils/generic.py:441: UserWarning: torch.utils._pytree._register_pytree_node is deprecated. Please use torch.utils._pytree.register_pytree_node instead.
  _torch_pytree._register_pytree_node(
/fsx/sayak/lib/python3.9/site-packages/transformers/utils/generic.py:309: UserWarning: torch.utils._pytree._register_pytree_node is deprecated. Please use torch.utils._pytree.register_pytree_node instead.
  _torch_pytree._register_pytree_node(
/fsx/sayak/diffusers/src/diffusers/utils/outputs.py:63: UserWarning: torch.utils._pytree._register_pytree_node is deprecated. Please use torch.utils._pytree.register_pytree_node instead.
  torch.utils._pytree._register_pytree_node(
Namespace(no_bf16=False, no_sdpa=False, batch_size=1, num_inference_steps=30, enable_fused_projections=True, upcast_vae=False, compile_unet=True, compile_vae=True, compile_mode='max-autotune', change_comp_config=True, do_quant=None, tag='branch12-1-select')
Using dtype: torch.bfloat16

Loading pipeline components...:   0%|          | 0/7 [00:00<?, ?it/s]
Loading pipeline components...:  14%|█▍        | 1/7 [00:01<00:11,  1.98s/it]
Loading pipeline components...:  29%|██▊       | 2/7 [00:02<00:04,  1.09it/s]
Loading pipeline components...:  57%|█████▋    | 4/7 [00:18<00:16,  5.53s/it]
Loading pipeline components...:  86%|████████▌ | 6/7 [00:21<00:03,  3.64s/it]
Loading pipeline components...: 100%|██████████| 7/7 [01:22<00:00, 18.01s/it]
Loading pipeline components...: 100%|██████████| 7/7 [01:22<00:00, 11.77s/it]
Enabling fused QKV projections for both UNet and VAE.
Compile UNet
Compile VAE
python: can't open file '/fsx/sayak/sdxl-fast/prepare_plot.py': [Errno 2] No such file or directory
srun: error: ip-26-0-154-212: task 0: Exited with exit code 2
++ date
+ echo 'END TIME: Wed Dec 20 11:54:30 UTC 2023'
END TIME: Wed Dec 20 11:54:30 UTC 2023