Hi
I am trying to run the OMP UIA example project on TMS320C6670 EVM using instructions from http://processors.wiki.ti.com/index.php/SystemAnalyzerTutorial7.
I am able to build and run the project. In Live Session view, I can even find"OMP PARALLEL FOR...." events. But the Duration view of CCS is blank.
So, I used the filter in the live session view to filter the required events only and computed manually (substracting the stop time from start time). The results didn't make any sense.
| 40655851262 | C66xx_0 | Start: Reference time start |
| 40774170418 | C66xx_0 | Stop: Reference time stop |
| 40774176260 | C66xx_0 | Start: OMP PARALLEL FOR start, using 1 cores |
| 40892555187 | C66xx_0 | Stop: OMP PARALLEL FOR stop, using 1 cores |
| 40892556608 | C66xx_0 | Start: OMP PARALLEL FOR start, using 2 cores |
| 47044820977 | C66xx_0 | Stop: OMP PARALLEL FOR stop, using 2 cores |
| 47044822384 | C66xx_0 | Start: OMP PARALLEL FOR start, using 3 cores |
| 49220830945 | C66xx_0 | Stop: OMP PARALLEL FOR stop, using 3 cores |
| 49220832380 | C66xx_0 | Start: OMP PARALLEL FOR start, using 4 cores |
| 53700382053 | C66xx_0 | Stop: OMP PARALLEL FOR stop, using 4 cores |
| Event | Duration | Duration/ref |
| Reference time | 118319156 | |
| Using 1 Core | 118378927 | 1.000505168 |
| Using 2 Core | 6152264369 | 51.9971962 |
| Using 3 Core | 2176008561 | 18.39100814 |
| Using 4 Core | 4479549673 | 37.85988528 |
So, it seems that using more cores have somehow decreased the performance. Am I doing something wrong here?
System info:
CCS v5.5.0.00077, Compiler v7.4.4, XDCtools v3.25.3.72, IPC v1.24.3.32, MCSDK PDK TMS320C6670 v1.1.2.6, OpenMP BIOS runtime library v1.1.3.02, SYS/BIOS v6.35.4.50, System Analyzer (UIA Target) v1.3.1.08
--Amanullah Ghazi
