環境が微妙に変わった？というような印象もあり、メモしておきます。
目次
Google Colab Proのそれぞれの使いどころ
- T4が一番性能が低い → 消費ユニット数を抑えたいなら、まずはT4で確認しながら進める
- A100はメモリと性能のバランスが良い → 本番はA100が良いかも
- L4 は性能が高い → 性能重視の処理であれば L4も選択肢に入る
- TPU v2-8はTensorFlowに最適化されたプロセッサ→ 行列演算・テンソル演算がやりたい時
基本環境
GPUなし状態で実施
Python バージョン：3.11.11
!python --version
Python 3.11.11CUDA： 12.5
!nvcc --version
nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2024 NVIDIA Corporation
Built on Thu_Jun__6_02:18:23_PDT_2024
Cuda compilation tools, release 12.5, V12.5.82
Build cuda_12.5.r12.5/compiler.34385749_0Pytorch : 2.6.0+cu124
!pip show torch
!pip show tensorflow
Name: torch
Version: 2.6.0+cu124
Summary: Tensors and Dynamic neural networks in Python with strong GPU acceleration
Home-page: https://pytorch.org/
Author: PyTorch Team
Author-email: packages@pytorch.org
License: BSD-3-Clause
Name: tensorflow
Version: 2.18.0
Summary: TensorFlow is an open source machine learning framework for everyone.
Home-page: https://www.tensorflow.org/
Author: Google Inc.
Author-email: packages@tensorflow.org
License: Apache 2.0A100選択時
1時間あたり約7.62 ユニットを消費
+-----------------------------------------------------------------------------------------+
| NVIDIA-SMI 550.54.15 Driver Version: 550.54.15 CUDA Version: 12.4 |
|-----------------------------------------+------------------------+----------------------+
| GPU Name Persistence-M | Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap | Memory-Usage | GPU-Util Compute M. |
| | | MIG M. |
|=========================================+========================+======================|
| 0 NVIDIA A100-SXM4-40GB Off | 00000000:00:04.0 Off | 0 |
| N/A 33C P0 45W / 400W | 0MiB / 40960MiB | 0% Default |
| | | Disabled |
+-----------------------------------------+------------------------+----------------------+
+-----------------------------------------------------------------------------------------+
| Processes: |
| GPU GI CI PID Type Process name GPU Memory |
| ID ID Usage |
|=========================================================================================|
| No running processes found |
+-----------------------------------------------------------------------------------------+L4選択時
1時間あたり約2.09ユニットを消費
!nvidia-smi
+-----------------------------------------------------------------------------------------+
| NVIDIA-SMI 550.54.15 Driver Version: 550.54.15 CUDA Version: 12.4 |
|-----------------------------------------+------------------------+----------------------+
| GPU Name Persistence-M | Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap | Memory-Usage | GPU-Util Compute M. |
| | | MIG M. |
|=========================================+========================+======================|
| 0 NVIDIA L4 Off | 00000000:00:03.0 Off | 0 |
| N/A 51C P8 12W / 72W | 0MiB / 23034MiB | 0% Default |
| | | N/A |
+-----------------------------------------+------------------------+----------------------+
+-----------------------------------------------------------------------------------------+
| Processes: |
| GPU GI CI PID Type Process name GPU Memory |
| ID ID Usage |
|=========================================================================================|
| No running processes found |
+-----------------------------------------------------------------------------------------+T4 選択時
1時間あたり 約1.67ユニットを消費
+-----------------------------------------------------------------------------------------+
| NVIDIA-SMI 550.54.15 Driver Version: 550.54.15 CUDA Version: 12.4 |
|-----------------------------------------+------------------------+----------------------+
| GPU Name Persistence-M | Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap | Memory-Usage | GPU-Util Compute M. |
| | | MIG M. |
|=========================================+========================+======================|
| 0 Tesla T4 Off | 00000000:00:04.0 Off | 0 |
| N/A 49C P8 9W / 70W | 0MiB / 15360MiB | 0% Default |
| | | N/A |
+-----------------------------------------+------------------------+----------------------+
+-----------------------------------------------------------------------------------------+
| Processes: |
| GPU GI CI PID Type Process name GPU Memory |
| ID ID Usage |
|=========================================================================================|
| No running processes found |
+-----------------------------------------------------------------------------------------+v2-8 TPU 選択時
1時間あたり約1.76ユニットを消費
import torch_xla.core.xla_model as xm
# TPU デバイスのメモリ情報を取得
mem_info = xm.get_memory_info('xla:0')
# メモリ情報の表示
print(f"TPU メモリ情報: {mem_info}")
#----
TPU メモリ情報: {'bytes_used': 32768, 'bytes_limit': 8034189312, 'peak_bytes_used': 32768}
# メモリ上限（bytes_limit）が 8034189312 bytes ≒ 7.48 GB ＝ TPU 1コアあたりの上限。これ×8コア
