Hi,
I'm following the image_processing_openmp project with MCSDK and I have a question about how the executable is loaded. I am referencing instructions from the wiki for this demo at http://processors.wiki.ti.com/index.php/MCSDK_Image_Processing_Demonstration_Guide#Software_outline_of_the_OpenMP_demo.
They instruct you to build, connect Core 0, then load the .OUT file on Core 0 and run. In other OpenMP projects I've written (based on the OpenMP HelloWorld project), you connect as many cores as you intend to run, and you load the .OUT file on all connected cores and run from Core 0.
Why is this example different? If I try to run this with 8 cores, ( with OpenMP.setNumProcessors(8); in my CFG script), I attempt the following steps:
1) select all 8 cores and connect
2) select all 8 cores and load "...Release\image_processing_openmp_evmc6678l.out"
From here, I would select core 0 and Run, but it automatically runs on core 0 and suspends cores 1-7, the console output shows:
[C66xx_6] Error pushing completion in QMSS
[C66xx_7] Error pushing completion in QMSS
[C66xx_1] Error pushing completion in QMSS
[C66xx_2] Error pushing completion in QMSS
[C66xx_3] Error pushing completion in QMSS
[C66xx_4] Error pushing completion in QMSS
[C66xx_5] Error pushing completion in QMSS
What's going on, and how should this be done?
Thanks,
Ryan
