AM6442: MSRAM requirements for Profinet on AM64xx

Part Number: AM6442

Tool/software:

Hello,

we are evaluating Industrial Ethernet Profinet example from ind_comms_sdk_am64x_09_02_00_09 on multi-core AM6442 processor.

So far we managed to run the Profinet example on R50_0 core and also managed to extend it with IPC communication to R50_1 core.

I looks like Profinet example uses MSRAM very generously, for example heap is placed in MSRAM.

We need to estimate how much MSRAM can be available to cores that are not running Profinet but processing other device tasks, like R50_1 core  in our IPC extended Profinet example, or maybe running linux.

We know of this post https://e2e.ti.com/support/processors-group/processors/f/processors-forum/1378695/am6442-am64x-evm-msram-for-indistrial-protocols?tisearch=e2e-sitesearch&keymatch=MSRAM%20profinet#

But looking at .map file of the example is not sufficient. We think the Profinet example is not optimized in the regard of MSRAM usage.

We need to know which sections absolutely need to be placed in MSRAM to have correct Profinet operation and how big they can get.

Since there are many shared resources in this complex multi-core device, something like resource usage with minimum/maximum figures is absolutely needed for Profinet stack datasheet in general.

For now an approximate estimation would be OK.

Best regards

Rudolf

  • Hello Rudolf,

    Would you please share your linker.cmd and .map file?

    By Profinet example I assume you mean "Profinet stack + Profinet demo application". For the time being, memory usage can't be estimated out of the box because we have dependencies on some configuration options like maximum number of slot/subslots needed. Also in order to have a generic solution that works on EVM boards as well as on Launchpads, we can't always use DDR to save some MSRAM resources. 

    On the other hand, our demo application is just meant to provide a usage example so you can modify it according to your needs and manipulate the linker script accordingly to achieve the required memory utilization.

    Kind regards,
    Kamil