Difference between revisions of "Benchmarks:AI"

From Define Wiki
Jump to navigation Jump to search
(Created page with "== Training == ==== TensorFlow ==== TensorFlow provides scripts to run training benchmarks with different models. The scripts are hosted on GitHub [https://github.com/tensorfl...")
 
Line 7: Line 7:
 
To simplify the setup I have created a Dockerfile to pull the image and download the scripts. To use this first create a directory to hold your Dockerfiles.
 
To simplify the setup I have created a Dockerfile to pull the image and download the scripts. To use this first create a directory to hold your Dockerfiles.
  
 +
<syntaxhighlight>
 
mkdir ~/Dockerfiles
 
mkdir ~/Dockerfiles
 +
</syntaxhighlight>
  
 
Then create a file in this directory and add the following
 
Then create a file in this directory and add the following
  
 +
<syntaxhighlight>
 
FROM nvcr.io/nvidia/tensorflow:18.10-py3
 
FROM nvcr.io/nvidia/tensorflow:18.10-py3
 
RUN apt-get update && apt-get install git && git clone -b cnn_tf_v1.10_compatible https://github.com/tensorflow/benchmarks
 
RUN apt-get update && apt-get install git && git clone -b cnn_tf_v1.10_compatible https://github.com/tensorflow/benchmarks
 
ENTRYPOINT bash
 
ENTRYPOINT bash
 +
</syntaxhighlight>
  
 
To build the image run  
 
To build the image run  
  
docker build -f dockerfiles/tf_bench .
+
<syntaxhighlight>
 +
docker build -f ~/Dockerfiles/tf_bench -t tf_bench .
 +
</syntaxhighlight>
 +
 
 +
The best way to run the container is in interactive mode as this allows multiple runs to be performed in quick succession. To start the container run
 +
 
 +
<syntaxhighlight>
 +
docker run --runtime=nvidia -it tf_bench
 +
</syntaxhighlight>
 +
 
 +
The benchmark scripts are located in /workspace/scripts/tf_cnn_benchmarks.
 +
 
 +
To run the benchmark using synthetic data execute
 +
 
 +
<syntaxhighlight>
 +
python tf_cnn_benchmarks.py --num_gpus=1 --batch_size=32 --model=resnet50 --variable_update=parameter_server
 +
</syntaxhighlight>
 +
 
 +
The benchmark can be run with different models. Current supported models are resnet50, resnet152, inception3 and vgg16.

Revision as of 12:28, 31 January 2019

Training

TensorFlow

TensorFlow provides scripts to run training benchmarks with different models. The scripts are hosted on GitHub here.

It is recommended to run the scripts using nvidia-docker2 and the TensorFlow docker image obtained from NGC.

To simplify the setup I have created a Dockerfile to pull the image and download the scripts. To use this first create a directory to hold your Dockerfiles.

mkdir ~/Dockerfiles

Then create a file in this directory and add the following

FROM nvcr.io/nvidia/tensorflow:18.10-py3
RUN apt-get update && apt-get install git && git clone -b cnn_tf_v1.10_compatible https://github.com/tensorflow/benchmarks
ENTRYPOINT bash

To build the image run

docker build -f ~/Dockerfiles/tf_bench -t tf_bench .

The best way to run the container is in interactive mode as this allows multiple runs to be performed in quick succession. To start the container run

docker run --runtime=nvidia -it tf_bench

The benchmark scripts are located in /workspace/scripts/tf_cnn_benchmarks.

To run the benchmark using synthetic data execute

python tf_cnn_benchmarks.py --num_gpus=1 --batch_size=32 --model=resnet50 --variable_update=parameter_server

The benchmark can be run with different models. Current supported models are resnet50, resnet152, inception3 and vgg16.