• State
  • Locked Locked
  • Replies 2 replies
  • Subscribers 98 subscribers
  • Views 247 views
  • Users 0 members are here
Support feedback
Options
Options
Related

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.

C6747 USB speed

Lopi
Intellectual 470 points

 

Does anyone know what the bottleneck is in the USB speed for the C6747 when operated as a device?  From the BIOSUSB Datasheet p.15,  I should expect speeds of 8.5MB/sec for file writes from an XP host to the DSP and 11.3 MB/sec for file reads.  This is in fact close to what I am observing.  Is there any way to speed it up?  Is the limitation in speed caused by the Jungo code or the part itself?

Thanks,

Lori

  • 0
    TI__Expert 7435 points

    Lori,

    Can you provide more information on

    • your usb use case
    • whether you are using the BIOSUSB stack and developing applications on top of it or porting/developing your own custom usb stack.

    C6747 usb per say is not limited to the published numbers but the throughput typically is limited by the way the usb application works/implemented.  For ex. USB MSC (for which the numbers quoted are published) imposes significant processing overhead and hence the associated throughput numbers.

    If on the other hand your use case involves writing custom applications then you can definitely achieve higher throughput numbers (than that is quoted for MSC transactions).

    regards

    swami

  • 0 in reply to swami
    Intellectual 470 points

    swami,

    Thanks for the quick reply.  Our ultimate goal is to feed ADC data into EMIFA, process the data, store the results in SDRAM and output via USB to a PC.  The PC will be running Windows with a custom API that displays the data.  I find the idea of writing a custom USB stack quite daunting.  What I've been trying to do is adapt msc_blkmedia_glue.c  to suit my needs.  So for example, in the function MSG_readLun instead of calling the blkmedia driver to copy the data to the USB buffer, I just copy it straight from SDRAM.  It seems to work, although I've run into unresolved problems when I try to enable cacheability for the SDRAM (i.e. the PC occasionally loses connection with the DSP if I make SDRAM cacheable).   The speed that I've seen is about 10 MB/s (similar to the BIOSUSB doc).  I had hoped that by bypassing the blkmedia driver that I would already be running a little faster.  It's frustrating that we don't have access to the Jungo source. 

    Lori