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.

Relative performance between SM32C6713B and SM320C6727B

Other Parts Discussed in Thread: TMS320F28335, SM320C6727B

We have a control application where a TMS320F28335 is not providing enough computation throughput.  It is an extended temperature application so we are looking at SM32C6713B or SM320C6727B processors. The SM320C6727B is only available up to 250MHz in the extended temperature range while the SM32C6713B is 300MHz.

A few questions:

1) Any idea how much faster on average the x6700 family would be over the 150MHz '335? It does have a clock rate advantage plus a significantly wider computational path.

2) The SM32C6727B seems to have an improved memory and cache architecture. Would this be enough to offset it being 50MHz slower in our temperature range?

  • Tom,

    Welcome to the TI E2E forum. I hope you will find many good answers here and in the TI.com documents and in the TI Wiki Pages. Be sure to search those for helpful information and to browse for the questions others may have asked on similar topics.

    Which temperature range do you require? We seem to have a confusing array of answers on which parts are available at which temperature ranges.

    The only way to get a valid idea of relative performance is to run an example algorithm on the simulator. You can use CCSv5 and its simulator for any of these devices without requiring a license. We want to make it easy for you to do these evaluations.

    Each step in the architecture is done with performance improvements in mind. The C6727 adds some instruction enhancements, but it also adds an improved DMA engine, dMAX. The external memory interface performance will also be a place where you can get a big gain from one device to the next.

    Regards,
    RandyP

  • Hello Tom,

    Tom Clary said:
    1) Any idea how much faster on average the x6700 family would be over the 150MHz '335? It does have a clock rate advantage plus a significantly wider computational path.

    The C67x DSPLIB software suite has source code and collateral for optimized signal processing kernels. You can get the cycle counts of these operations (by running code in CCS as Randy mentioned or by paper calculations from collateral) and compare against the similar functions you may have implemented on the C2000. You can access the DSPLIB here: http://www.ti.com/tool/sprc121

    Tom Clary said:
    2) The SM32C6727B seems to have an improved memory and cache architecture. Would this be enough to offset it being 50MHz slower in our temperature range?

    Once you get a broad idea about the performance you would need on the C67x by the above analysis, you might be able to answer this based on the complete system-level requirements (data throughput from certain peripherals, data bandwidth through the system interconnect etc)  that you and other engineers in your team will have. There is no way to make a recommendation without knowing the complete requirements.

    Regards,

    Sunil Kamath