TDA4VM: CCS Load Analysis tool problem

Hi Sergei,

Are you using SK-TDA4VM or TDA4VM?

See the following user guides for CCS setup and troubleshooting:

https://software-dl.ti.com/jacinto7/esd/processor-sdk-rtos-jacinto7/10_00_00_05/exports/docs/psdk_rtos/docs/user_guide/ccs_setup_j721e.html

https://software-dl.ti.com/ccs/esd/documents/users_guide/ccs_troubleshooting.html#rtos-analyzer-issues

Thank you,

Fabiana

Hi Fabiana,

Thank you for your response. 

I'm using TDA4VM.

I tried two CCS target configurations: with and without loading initialization script of  DMSC_Cortex_M3_0  (ccs_base/emulation/gel/J721E_DRA829_TDA4VM/gel/J721E.gel). 

I successfully ran /packages/ti/drv/sciclient/tools/ccsLoadDmsc/j721e/launch.js script from Scripting Console.

I still cannot use the above mentioned tools (RTOS Analyzer, System Analyzer, Hardware Trace Analyzer). Nothing happens whenever an attempt

to run any of those tools.

Actually I need to measure load of C7x and C66x. What would be the right way to do this?

Best regards, Sergei

Hi Sergei,

The RTOS analyzer is designed for TI RTOS only. Is this the OS you are working with?

Thank you,

Fabiana

Hi Fabiana,

I'm working with Linux. Could you please advice how can I measure C7x and C66x utilization in this case?

Thanks, Sergei

May be via PMU?

It looks like it is not enabled for C7x DSP:

root@j721e-evm:/proc/device-tree# dmesg | grep -i pmu
[ 0.586587] hw perfevents: enabled with armv8_cortex_a72 PMU driver, 7 counters available
root@j721e-evm:/proc/device-tree#

If it is the way, how can I enable it for C7x and C66x?


Thanks, Sergei

Hi Sergei,

Only A72 running Linux is supported. Running Linux is not supported on C6x and C7x. We support running RTOS binaries on them.

Please take a look at the CCS setup information to understand the dependencies and requirements for the CCS tools: https://software-dl.ti.com/jacinto7/esd/processor-sdk-rtos-jacinto7/10_00_00_05/exports/docs/psdk_rtos/docs/user_guide/ccs_setup_j721e.html#ccs-setup-for-platform

Thank you,

Fabiana

Hi Fabiana,

I'm sorry for the confusion. Yes, the main program is running on the A72. However, it runs some code on the DSPs and uses hardware accelerators.

I execute the edgeai_demo, which uses deep learning and hardware accelerators, what would be the correct way to measure the utilization of the DSPs and hardware accelerators?

The board was flashed with image built with sdk 9.02 (ti-processor-sdk-linux-edgeai-j721e-evm-09_02_00_05-Linux-x86-Install.bin) as per instruction https://software-dl.ti.com/jacinto7/esd/processor-sdk-linux-jacinto7/09_01_00_06/exports/docs/linux/Overview_Building_the_SDK.html.

I assume C7x and C6x run without an OS (bare-metal), where the application code directly controls the DSP without OS-level abstractions.

Can I measure how edgeai_demo utilizes DSPs?

Best regards, Sergei

Hi Fabiana,

Oh... This looks like what I needed. Thank you very much for pointing it out. I'm not sure how I missed that...

Best regards, Sergei