+ All Categories
Home > Technology > Q2.12: Benchmarking Techniques

Q2.12: Benchmarking Techniques

Date post: 28-Nov-2014
Category:
Upload: linaro
View: 299 times
Download: 0 times
Share this document with a friend
Description:
Resource: Q2.12 Name: Benchmarking Techniques Date: 30-05-2012 Speaker: Michael Hope
24
Benchmarking Techniques Michael Hope <[email protected]> bzr branch lp:~michaelh1/+junk/benchmarking-techniques r6
Transcript
Page 1: Q2.12: Benchmarking Techniques

Benchmarking Techniques

Michael Hope <[email protected]>

bzr branch lp:~michaelh1/+junk/benchmarking-techniques r6

Page 2: Q2.12: Benchmarking Techniques

2

Why benchmark?

Page 3: Q2.12: Benchmarking Techniques

3

Issues are

RelevanceAccuracy

Repeatability

Page 4: Q2.12: Benchmarking Techniques

4

Picking relevant benchmarks:

ProfileWorkloadFeatures

We use SPEC 2000 and EEMBC

We'd like shareable benchmarks

Page 5: Q2.12: Benchmarking Techniques

5

Test Platform

Build / test / benchmark via Linux / web / commodity hardware

Page 6: Q2.12: Benchmarking Techniques

6

Measuring

Page 7: Q2.12: Benchmarking Techniques

7

Realtime timers

See 'man clock_gettime'

Page 8: Q2.12: Benchmarking Techniques

8

<comparison of different timers>

See https://wiki.linaro.org/WorkingGroups/ToolChain/Benchmarks/TimerAccuracy

Page 9: Q2.12: Benchmarking Techniques

9

Page 10: Q2.12: Benchmarking Techniques

10

Page 11: Q2.12: Benchmarking Techniques

11

External influences

Page 12: Q2.12: Benchmarking Techniques

12

Other features to be wary of

schedulergovernor

cpuidle, power management, thermal limiting

SMPbugs, like core lockdown

NEON startup

See “Understanding the Linux Kernel” ch10:http://oreilly.com/catalog/linuxkernel/chapter/ch10.html

Page 13: Q2.12: Benchmarking Techniques

13

Putting things together and running

We use:timer built into the app

run five timescollect everything

post process

Page 14: Q2.12: Benchmarking Techniques

14

Statistics

Page 15: Q2.12: Benchmarking Techniques

15

Page 16: Q2.12: Benchmarking Techniques

16

Page 17: Q2.12: Benchmarking Techniques

17

Standard deviationDispersion / coefficient of variance

t-scorest-test

Page 18: Q2.12: Benchmarking Techniques

18

t=X̄ 1− X̄ 2

√ s12

N 1

+s22

N 2

See http://en.wikipedia.org/wiki/Welch%27s_t_test

Page 19: Q2.12: Benchmarking Techniques

19

Compiler Mean Std CVgcc-4.6.2 1.00 134u 134ugcc-linaro-4.6-2011.11 4.25 5035u 1178u

t = 30,300

Page 20: Q2.12: Benchmarking Techniques

21

Variant Mean Std CVPlain 1.00 320u 320uWith SMS 1.01 430u 426u

t = 118

Page 21: Q2.12: Benchmarking Techniques

22

Variant Mean Std CVPlain 1.00 320u 320uWith vectoriser 1.001 404u 404u

t = 8.27 - significant

Page 22: Q2.12: Benchmarking Techniques

23

Our tools

perfdifftestbetterstabulatePython

LibreOffice!

Other statistical tools like scipy.stat, R, Judge, and ministat

Page 23: Q2.12: Benchmarking Techniques

24http://help.libreoffice.org/Calc/Applying_AutoFilter

Page 24: Q2.12: Benchmarking Techniques

25

www.linaro.org / wiki.linaro.org

people.linaro.org/~michaelh/presentations


Recommended