TensorFlow

This is a benchmark of the TensorFlow deep learning framework using the TensorFlow reference benchmarks (tensorflow/benchmarks with tf_cnn_benchmarks.py). Note with the Phoronix Test Suite there is also pts/tensorflow-lite for benchmarking the TensorFlow Lite binaries if desired for complementary metrics.

To run this test with the Phoronix Test Suite, the basic command is: phoronix-test-suite benchmark tensorflow.

Project Site

tensorflow.org

Source Repository

github.com

Test Created

6 February 2017

Last Updated

26 March 2024

Test Maintainer

Michael Larabel 

Test Type

System

Average Install Time

46 Seconds

Average Run Time

1 Minute, 51 Seconds

Test Dependencies

Python

Accolades

80k+ Downloads

Supported Platforms


Public Result Uploads *Reported Installs **Reported Test Completions **Test Profile Page Views ***OpenBenchmarking.orgEventsTensorflow Popularity Statisticspts/tensorflow2017.022017.052017.082017.112018.022018.052018.082018.112019.022019.052019.082019.112020.022020.052020.082020.112021.022021.052021.082021.112022.022022.052022.082022.112023.022023.052023.082023.112024.022024.056K12K18K24K30K
* Uploading of benchmark result data to OpenBenchmarking.org is always optional (opt-in) via the Phoronix Test Suite for users wishing to share their results publicly.
** Data based on those opting to upload their test results to OpenBenchmarking.org and users enabling the opt-in anonymous statistics reporting while running benchmarks from an Internet-connected platform.
*** Test profile page view reporting began March 2021.
Data updated weekly as of 7 June 2024.
3216.6%25617.7%121.1%6417.9%51215.2%1611.5%Batch Size Option PopularityOpenBenchmarking.org
GoogLeNet33.9%AlexNet32.4%ResNet-5033.7%Model Option PopularityOpenBenchmarking.org

Revision History

pts/tensorflow-2.2.0   [View Source]   Tue, 26 Mar 2024 13:32:06 GMT
Update against TensorFlow 2.16.1 upstream.

pts/tensorflow-2.1.1   [View Source]   Sat, 06 Jan 2024 17:48:46 GMT
Add NVIDIA GPU CUDA option.

pts/tensorflow-2.1.0   [View Source]   Tue, 28 Mar 2023 16:38:47 GMT
Update against TensorFlow 2.12 upstream to work with newer distros / Python.

pts/tensorflow-2.0.0   [View Source]   Fri, 07 Oct 2022 12:18:59 GMT
Update this, get it going against TF 2.10 upstream...

pts/tensorflow-1.1.0   [View Source]   Fri, 06 Mar 2020 07:23:10 GMT
Rework test, increase max steps to 1000

pts/tensorflow-1.0.0   [View Source]   Mon, 06 Feb 2017 18:08:28 GMT
Initial commit of TensorFlow benchmark

Suites Using This Test

Machine Learning

HPC - High Performance Computing

CPU Massive


Performance Metrics

Analyze Test Configuration:

TensorFlow 2.12

Device: CPU - Batch Size: 1 - Model: VGG-16

OpenBenchmarking.org metrics for this test profile configuration based on 151 public results since 7 January 2024 with the latest data as of 13 March 2024.

Below is an overview of the generalized performance for components where there is sufficient statistically significant data based upon user-uploaded results. It is important to keep in mind particularly in the Linux/open-source space there can be vastly different OS configurations, with this overview intended to offer just general guidance as to the performance expectations.

Component
Percentile Rank
# Compatible Public Results
images/sec (Average)
98th
4
12.07 +/- 0.21
93rd
3
9.76 +/- 0.01
91st
4
9.74 +/- 0.01
84th
6
5.77 +/- 0.01
81st
4
5.74 +/- 0.01
76th
9
5.63 +/- 0.35
Mid-Tier
75th
< 5.50
68th
6
5.04 +/- 0.49
58th
4
4.77 +/- 0.12
51st
10
4.54 +/- 0.18
Median
50th
4.41
49th
3
4.40 +/- 0.01
43rd
7
4.07 +/- 0.15
41st
4
3.78 +/- 0.03
39th
4
3.75 +/- 0.01
36th
3
3.70 +/- 0.01
35th
3
3.42 +/- 0.01
32nd
3
3.26 +/- 0.02
27th
3
2.69 +/- 0.02
Low-Tier
25th
< 2.69
25th
4
2.68 +/- 0.01
22nd
4
2.36 +/- 0.27
19th
3
2.01 +/- 0.01
7th
3
1.45 +/- 0.02
4th
3
1.08 +/- 0.01
OpenBenchmarking.orgDistribution Of Public Results - Device: CPU - Batch Size: 1 - Model: VGG-16151 Results Range From 0 To 13 images/sec246810121416182022241530456075

Based on OpenBenchmarking.org data, the selected test / test configuration (TensorFlow 2.12 - Device: CPU - Batch Size: 1 - Model: VGG-16) has an average run-time of 2 minutes. By default this test profile is set to run at least 3 times but may increase if the standard deviation exceeds pre-defined defaults or other calculations deem additional runs necessary for greater statistical accuracy of the result.

OpenBenchmarking.orgMinutesTime Required To Complete BenchmarkDevice: CPU - Batch Size: 1 - Model: VGG-16Run-Time246810Min: 1 / Avg: 1.57 / Max: 4

Tested CPU Architectures

This benchmark has been successfully tested on the below mentioned architectures. The CPU architectures listed is where successful OpenBenchmarking.org result uploads occurred, namely for helping to determine if a given test is compatible with various alternative CPU architectures.

CPU Architecture
Kernel Identifier
Verified On
Intel / AMD x86 64-bit
x86_64
(Many Processors)