Hi all,
I'am working with TMS320C6670 and I want to use an TCP3e for encoding LTE data.
After reading the UG for TCP3e everything seems to be similar with VCP2: You need to configure the EDMA3 transfers, prepare the input data and start the co-processor, after that just wait for the last EDMA3 transfer with output data. So as I understand for TCP3e there are 3 mandatory EDMA3 transfers: config for the code block, input data and output data.
But when I get to the examples from the PDK I come up with a lot of questions, maybe someone could clarify some of them or point me to the explicit document.
Questions I have:
1) "With Sync Buffer", "Pair-wise", "Without sync Buffer" these are the chaining modes as I understand? What are the difference between them?
2) Why there are 7 EDMA transfers? WEVT EDMA, REVT EDMA is input and output data as I understand, but what the SYNC EDMA, TRIG SYNC EDMA, Mgz0 EDMA, Mgz1 EDMA and Queue EDMA stand for?
3) Why we are using QMSS for data if we have the EDMA3? Is there a job that EDMA3 can't handle? This is all that I can get from the code : "By default, we use the Qmss API to get the address."
4) At the end of the testing application all three methods show Effective Throughput about 6.4438 Mbps, which is not so good for LTE, is there any chances to get higher speed?
5) In our application we also using a VCP2 for the decoding Wi-Fi signal field and there CSL EDMA3 functions are used instead of the EDMA3 driver. Could it cause some troubles in the future if we implement the TCp3e like in example?
6) As I understood from the comments only TC0 can be used for the correct TCP3e driver work, other just fails.
7) is the L2 cache disabling mandatory actions or not?
Sorry for a lot of questions, but I just can't get clear answers for that anywhere.
Best regards, Pavlo!