Tensorflow / PyTorch GPU 세팅

텐서플로와 파이토치에서 GPU 사용 확인과 셋업 참고 사항 기록
텐서 2.11 버전부터는 윈도우에서 GPU 사용을 지원하지 않는다고 함!

Tensorflow and PyTorch Setup

• CUDA: 11.2
• cuDNN : 8.1
• TF: tensorflow-gpu==2.10.0
• Torch: conda install pytorch==1.12.1 torchvision==0.13.1 torchaudio==0.12.1 cudatoolkit=11.3 -c pytorch
• Python: 3.7

CUDA Setup

• C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.2\bin
• C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.2\lib
• C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.2\include
• nvcc --version
• nvidia-smi

Ref

Tensorflow CUDA 버전 확인 PyTorch 버전 확인

import tensorflow as tf
import torch

tf.__version__, torch.__version__

('2.10.0', '1.12.1')

# TF GPU
from tensorflow.python.client import device_lib

device_lib.list_local_devices()

[name: "/device:CPU:0"
 device_type: "CPU"
 memory_limit: 268435456
 locality {
 }
 incarnation: 8541440801171234456
 xla_global_id: -1,
 name: "/device:GPU:0"
 device_type: "GPU"
 memory_limit: 6254755840
 locality {
   bus_id: 1
   links {
   }
 }
 incarnation: 17163750858690062226
 physical_device_desc: "device: 0, name: NVIDIA GeForce RTX 2070 SUPER, pci bus id: 0000:09:00.0, compute capability: 7.5"
 xla_global_id: 416903419]

# Torch GPU
"CUDA" if torch.cuda.is_available() else "CPU"

'CUDA'