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.

Am18X Upp's Wait line

Other Parts Discussed in Thread: AM1808, OMAPL138

Dears;

for AM1808; I read in a post here that on the UPP port; the Wait line is used for debug and in normal use; it can not signal the sender to wait  while the receiver is processing the data.

but I read the datasheet and the UPP's manual several times; it doesn't mention anything about the Wait not being usable during normal transfers.

Can someone please clarify this.

Perhaps; if it is not for nomral use; then we should use a GPIO pin for that purpose.?

thank you.

 

  • That's correct.  The WAIT signal is not used by UPP in receive mode for flow control.  UPP will assert an overflow error if it cannot keep up with the receive data.  If your system is able to detect that its receive is falling behind, you should be able to use GPIO to signal the transmitter to stop.

  • Thanks;

    Do you think a Gpio is better or if a constant wait delay time from the sender side.

    I mean to do a constant wait time between each transmit.

    That however reduces the total bandwidth heavily!

    The Idea of UPP is great for inter-processors communication; but I am worried that it doesn't have enough signalings for controlling it.

    Also; it seams that it's dropped on all new processors after am18xx; like am37xx; am38 and etc.

    I hope I made the right choice of using upp for high-speed bus!

    thanks.

     

  • You should not need to add any explicit delays or WAIT signals on UPP.  The only thing to look out for is that UPP is given a high enough priority in the system to service the real-time deadlines.  This would be like any of the other peripherals without flow-control (McASP, McBSP, etc).

  • asad1 said:

    Also; it seams that it's dropped on all new processors after am18xx; like am37xx; am38 and etc.

    I hope I made the right choice of using upp for high-speed bus!

    It is in "consideration" for some of the upcoming roadmap devices, but it will depend on application use case and what other inter processor communication peripherals the device has to offer. For AM18xx and OMAPL138 you certainly made a good choice using uPP.

    If you can , can you share your exact use-case on how you are using uPP and what kind of bandwidth you expect from this interface in your application.

    Regards

    Mukul

     

  • Hi Mukul.

    Thanks for the info. I hope to hear more on the new ICs that use UPP soon.

    for some high volume projects we wanted a line of CPU products that we can build different models.

    then to choose the peripherals so that we can move from a low end to high end models.

    we had chosen marvell Cpus at first. but after lots of analysis; we are prototyping and testing TI's for 450, 600, 1000, 1500 Mhz line.

    am18xx, am35xx,am37xx,am38xx.

    Our competitor is using Intel Atom and Marvells chips. I think our designs will beat them hands down.

    that will give us 4 models which is very good.

    However in our designs communication among cpus is very important.

    so upp is a great choice for high data volume while low clock speed and less external support ICs to run it than Pcie, usb, and others.

    the only issue is that upp is only on the first item: 450Mhz cpu.

    for the 600 and 1000, we have to choose usb which is only 480mhz; while upp is about 1.6 Ghz.

    we can also drop the 600 and 1000 and go with 1500 mhz model; but the cpu cost is about $40 which may be too much since the

    Atom's similar speeds is about $29 to $35.

    the Ti's chips are more for smart phone, video, lcd type. while we'll use them for Data processing only.

    Power usage and Sata on am18xx was the main reason for choosing. without sata and upp; the atmel's would have been better.

    so as you see lots and lots of technical issues has to get analyzed. specially that 60% of the peripherals on the cpu we will not use.

    but upp is a great little thing and if I learn how to joing/parallel it with PRU; then it will be a real coprocessor core!

    oh.. sorry for talking too much.

    thanks and please let me know about the new cpus with upp whenever you'll get info.

    ma

     

     

    .

  • Hi

    asad1 said:

    oh.. sorry for talking too much.

    thanks and please let me know about the new cpus with upp whenever you'll get info.

    On the contrary, I appreciate the time you took to provide your feedback on this.It is much appreciated. I have also forwarded your post to other architecture & product teams. Will keep you posted whenever it is worthy sharing on the public forums. Please also stay in touch with your TI local field office on roadmap devices and strategy.

    Regards

    Mukul

     

  • Appreciate it.

    we are a small start up and need all the help we can get.

    In late 1980's studied/programmed on Multi-processors(80286 and 80386)

    but it took this long to decide to build some full fledge versions!

    thanks.