Skip to content

Latest commit

 

History

History
112 lines (69 loc) · 2.73 KB

benchmark.md

File metadata and controls

112 lines (69 loc) · 2.73 KB
Raw

Tengine Performance Report

Revision Record

Date Rev Change Description Author
2017-12-29 0.1 Initial version FeyaHan
2018-01-06 0.2 Add multi CPU performance HaoLuo
2018-06-14 0.3 Add ACL_GPU performance Chunying

Catalog

Test Environment

Test

Performance

Test Environment

  • Tengine : v0.3

  • Broad : ROCK960

  • CPU : Rockchip RK3399.

    • Dual-core Cortex-A72 up to 2.0GHz (real frequency is 1.8GHz);

    • Quad-core Cortex-A53 up to 1.5GHz (real frequency is 1.4GHz).

  • GPU : Mali T864 (800MHz).

  • Operating System : Ubuntu 16.04.

Test

Step1. install Tengine

For more information about the build of Tengine, please refer to the documentation of [install](install.md)

Step2. lock the cpu frequency at maximum

    #switch to root user
    > sudo su 

    #check which available policy, policy4 for A72, policy0 for A53
    > cat  /sys/devices/system/cpu/cpufreq/policy4/scaling_available_governors  

    #set performance policy
    > echo performance > /sys/devices/system/cpu/cpufreq/policy4/scaling_governor 
    
    #check cpu frequency
    > cat /sys/devices/system/cpu/cpufreq/policy4/cpuinfo_cur_freq

Step3: test bench_sqz, bench_mobilenet

  • get model

    You can get the models from Tengine model zoo,the pass word is 57vb. And then, put the "mobilenet.caffemodel" "mobilenet_deploy.prototxt" "squeezenet_v1.1.caffemodel" "sqz.prototxt" in ~/tengine/models

  • set device

    1. use ACL_GPU

      For how to build tengine with ACL_GPU, see acl_driver.md.
      You can run the test as

      ./build/tests/bin/bench_sqz -d acl_opencl
./build/tests/bin/bench_mobilenet -d acl_opencl

    2. use CPU: single-core/multi-cores

      To assign on different cpu core, there are two methods:

      • export TENGINE_CPU_LIST=0,1,2,3
      • tests/bin/bench_sqz –p 0,1,2,3

      For rk3399, cpu(0-3) are A53, cpu(4-5) are A72.

      • 1A72 tests/bin/bench_sqz –p 4
      • 2A72 tests/bin/bench_sqz –p 4,5
      • 1A53 tests/bin/bench_sqz –p 0
      • 4A53 tests/bin/bench_sqz –p 0,1,2,3

Performance

SqueezeNet(ms) Mobilenet (ms)
rk3399(1*A72) 91.2 122.1
rk3399(2*A72) 51.2 65.4
rk3399(1*A53) 232.5 323.6
rk3399(4*A53) 79.2 96.3
ACL(GPU) 61.4 95.9

Notes:
(1) We run N=100 times per test case.
(2) We take the average time of N repeats.