こちらのベンチマークを試してみます。ただし今の所GPU環境が無いためCPUのみです。
https://github.com/tensorflow/benchmarks
環境： MacbookPro '13 Core i5-8259U CPU 2.30GHz Python 3.6.6
CPU用に以下のコマンド引数を用います。
python tf_cnn_benchmarks.py --batch_size=512 --model=alexnet --data_format=NHWC --device=cpu
以下結果です。
TensorFlow: 1.14
Model: alexnet
Dataset: imagenet (synthetic)
Mode: training
SingleSess: False
Batch size: 512 global
512 per device
Num batches: 100
Num epochs: 0.04
Devices: ['/cpu:0']
NUMA bind: False
Data format: NHWC
Optimizer: sgd
Variables: parameter_server
==========
Step Img/sec total_loss
1 images/sec: 47.8 +/- 0.0 (jitter = 0.0) nan
10 images/sec: 47.7 +/- 0.1 (jitter = 0.4) nan
20 images/sec: 47.7 +/- 0.1 (jitter = 0.3) nan
30 images/sec: 47.7 +/- 0.1 (jitter = 0.3) nan
40 images/sec: 47.7 +/- 0.1 (jitter = 0.3) nan
50 images/sec: 47.7 +/- 0.0 (jitter = 0.3) nan
60 images/sec: 47.7 +/- 0.0 (jitter = 0.4) nan
70 images/sec: 47.7 +/- 0.0 (jitter = 0.4) nan
80 images/sec: 47.6 +/- 0.0 (jitter = 0.4) nan
90 images/sec: 47.5 +/- 0.1 (jitter = 0.5) nan
100 images/sec: 47.5 +/- 0.1 (jitter = 0.5) nan
----------------------------------------------------------------
total images/sec: 47.52
----------------------------------------------------------------