This thread has been locked.
If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.
Tool/software:
Hello,
While looking at memory latencies in AM243x (SPRACV1B), I noticed DDR4 access time is unexpectedly high (and experiments with the TMDS243EVM confirmed this). As this document also mentions access time for "TCM of another Cortex-R5F" and I would like to minimize DDR4 usage, I would like to know if it is possible for one core to use the TCM of all other cores even if the cores themselves are not active.
Best regards,
Ricardo
Ricardo, you can access both TCMs within a cluster from one CPU. Or you can make both TCMs in a cluster look like one using the SINGLE_CORE bitfield in the ctrlmmr_sec_clstrx_cfg register. A lot of this is described in section 6.2.3.2.2 of the TRM.
I don't think you can access a TCM from another cluster from the R5 of a different cluster. I think it would have high latency anyway. Let me double check that.
Are you looking to optimize the DDR4 access time? Or is the theoretical access time still too slow for your needs?
REgards,
james
Hey Ricardo,
Chiming in on the "TCM of another Cortex-R5F" portion of the question:
You can use the TCM of another R5F core, but you will experience a higher latency (this is benchmarked in that document as TCM of another Cortex-R5F ). I believe the other core has to be powered on, but does not need to be running any code for that access to work. You may be able to place the other R5F into a WFI (Wait for Interrupt) state to reduce any active power consumption while still allowing access to it's TCMs.
Sincerely,
Lucas