ApexAI / performance_test

**This project is deprecated** Go to https://gitlab.com/ApexAI/performance_test
64 stars 41 forks source link

Compute cpu load % for running experiments and log in the result csv and generated plots #99

Closed daggarwa closed 5 years ago

daggarwa commented 5 years ago

This PR contains changes to :

  1. Compute the %CPU load for the experiments running as part of performance test
  2. Add a new column in performance test results csv called cpu usage(%) logging the cpu load.
  3. Plot the CPU usage along with printing the average load seen during the experiment as part of the pdf generated.

The new column added can be seen in the screenshot below with the load values logged throughout the experiment run: image

The cpu usage(%) is seen as part of the generated pdf on the usage graph on the right plotted as a blue line and also a new field called cpu usage(%) is seen as part of average results section of the report.

image


This change is Reviewable

daggarwa commented 5 years ago

@deeplearningrobotics I have implemented the changes to compute and log the CPU load % in performance test results and report. Can you please review these changes. Thanks!

daggarwa commented 5 years ago

@monidzik Can you please review this PR. Thanks!

monidzik commented 5 years ago

@daggarwa One more comment, I run this with -DPERFORMANCE_TEST_ODB_FOR_SQL_ENABLED=ON to see if the new cpu_load value is correctly stored in the database - and I got error in analysis_result.hpp, in AnalysisResult() {} because m_cpu_load is not initialized. So to resolve this please change line 163 to const float_t m_cpu_load = {}; Thanks!

daggarwa commented 5 years ago

@daggarwa One more comment, I run this with -DPERFORMANCE_TEST_ODB_FOR_SQL_ENABLED=ON to see if the new cpu_load value is correctly stored in the database - and I got error in analysis_result.hpp, in AnalysisResult() {} because m_cpu_load is not initialized. So to resolve this please change line 163 to const float_t m_cpu_load = {}; Thanks!

@monidzik Okay thank you for that catch. I will change it accordingly.

daggarwa commented 5 years ago

@monidzik I have addressed all your comments. @deeplearningrobotics Can you please do the final review. Thanks!

daggarwa commented 5 years ago

@deeplearningrobotics I have addressed all your review comments. Can you please review it once? Thanks

daggarwa commented 5 years ago

@deeplearningrobotics I have addressed all your review comments. Please review once. Thanks!