Timing within OpenMP on C6678

Viktor Grytsenko

Hello,

I tried to measure the performance of C6678 with the simple one-loop code within OpenMP
and got next result:
[C66xx_0] ...on 1 core...
[C66xx_0] delta_time_point[01] = 7269 [cycles]
[C66xx_0] ...on 4 cores...
[C66xx_0] delta_time_point_1[01] = 7279 [cycles]

I expected to get about 4 times less cycles quantity on 4 cores,
but got almost the same value.

Please explain me what I did incorrectly?

Project is in the attachment.
I used:
Code Composer Studio
Version: 5.2.1.00018

mcsdk_2_01_02_06

TMDXEVM6678L EVM Hardware

Best regards,

over 9 years ago

0 Viktor Grytsenko over 9 years ago

Expert 1095 points

3005.4530.Proba_OpenMP.zip

0 Raja over 9 years ago in reply to Viktor Grytsenko

TI__Guru* 81335 points

Hi Viktor,

Moved this thread to correct(Keystone) forum.

Thanks.

0 Viktor Grytsenko over 9 years ago in reply to Raja

Expert 1095 points

Hi Rajasekaran,

This is Keystone Multicore Forum.

Did you suggest me to move my post to another forum?

Could you answer once more question.

Why did I get about 500 times slower result on 4 cores when I checked "Enable Support for OpenMP 3.0" in the Project's Properties ?

Thank you.

0 Raja over 9 years ago in reply to Viktor Grytsenko

TI__Guru* 81335 points

Hi Viktor,

No. I have moved this thread to Keystone Multicore forum.

I have requested my team members to answer your query, you shall watch this thread for answer.

Thanks.

0 Viktor Grytsenko over 9 years ago in reply to Raja

Expert 1095 points

Hi Rajasekaran,

Thank you. I'll watch this thread.

Processors

Processors forum

Timing within OpenMP on C6678