• State
  • Locked Locked
  • Replies 6 replies
  • Subscribers 31 subscribers
  • Views 2481 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

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.

Tiva C 129 Using GPIO with AHB

Dennis Begun
Intellectual 825 points

On the EK-TM4C129XL, when I use this GPIO initialization code:

ROM_SysCtlPeripheralEnable(SYSCTL_PERIPH_GPIOJ);
ROM_GPIOPinTypeGPIOOutput(GPIO_PORTJ_BASE, GPIO_PIN_1);

I can then write to PJ1 with

GPIO_PORTJ_AHB_DATA_R = ...

And everything works. 

Does this use the AHB or the APB? Or do I need to initialize with

SysCtlGPIOAHBEnable(SYSCTL_PERIPH_GPIOJ);
GPIOPinTypeGPIOOutput(GPIO_PORTJ_AHB_BASE, GPIO_PIN_1);

Because the above doesn't work, it gets stuck inside the second call.

I am asking because I would like to control the data register with 

*((uint32_t *)(MY_PORT_BASE + TIVA_GPIO_DATA_REGISTER_OFFSET)) = ...;

And I'm not sure whether to use GPIO_PORTJ_BASE or GPIO_PORTJ_AHB_BASE for MY_PORT_BASE

  • 0
    Intellectual 825 points
    While I'm at it - will there be a significant performance difference between GPIO ports A-J and ports K-Q? I am specifically interested in raising/lowering different IOs on different ports at the same time.
  • 0 in reply to Dennis Begun
    TI__Guru**** 244380 points
    Hello Dennis,

    On the TM4C129 the AHB is now the default, so the Macro definition for previously APB is now valid as AHB. The movement from APB to AHB reduces the clock latency so you would be getting better performance for direct GPIO. The number of simultaneous toggle is still 8-pins.

    Regards
    Amit
  • 0 in reply to Dennis Begun
    Guru 47900 points

    Dennis Begun said:
    I am specifically interested in raising/lowering different IOs on different ports at the same time. 

    Some ARM MCUs (even simpler ones than those here) have (long) supported 16 bit ports.

    As Amit states - this vendor locked in on 8 bits - unless you venture to EPI-class parts - you're stuck.

    Classic means from "olde days" was to employ small FPGA/CPLD (configured as a latch) and then perform writes from multiple ports.   When your input data is correctly assembled - only then do you "clock" the FPGA/CPLD.  (which latches the data and then presents it - after some HW delay)

    In some cases - if you're lucky - you may "get away" with writing to your (unknown) external device from one port (yet not clocking) and then doing the same from further port(s) - only then "clocking" that multi-port data in.   (not all devices will accept such transactions...)

  • 0 in reply to cb1
    Intellectual 825 points

    Thanks, but I'm afraid this answer is a little over my head. Let me ask a simple specific question:

    Suppose I want to toggle pin 1 on port J and pin 6 on port M on the above mentioned processor at the same time, or as closely together as possible. What should my code be? 

    Right now my best performing code is

    ROM_SysCtlPeripheralEnable(SYSCTL_PERIPH_GPIOJ);
ROM_SysCtlPeripheralEnable(SYSCTL_PERIPH_GPIOM);

MAP_GPIOPinTypeGPIOOutput(GPIO_PORTJ_BASE, GPIO_PIN_1);
MAP_GPIOPinTypeGPIOOutput(GPIO_PORTM_BASE, GPIO_PIN_6);

GPIO_PORTJ_AHB_DATA_R |= 0x02;
GPIO_PORTM_DATA_R |= 0x40;

    This raises PM6 about 10 clock ticks after it raises PJ1 (as observed with a scope). 

    Can I do better?

  • 0 in reply to Dennis Begun
    TI__Guru**** 244380 points
    Hello Dennis

    Yes, it can do better. In the code above you are doing a read modify write operation. Instead if you use the bit banded address or PM6 it would be faster. The bit banded address is the value you want to write left shifted by 2

    E.g. for 0x40 it would be 0x40 << 2 = 0x100

    Hence for the address HWREG(0x40063100) = 0x40 it would be faster

    Regards
    Amit
  • 0 in reply to Amit Ashara
    Guru 47900 points

    While it may be "faster" - it can never, "toggle in unison" - which (if I may) is poster's (real) and original desire.  May I quote poster?

    "I am specifically interested in raising/lowering different IOs on different ports at the same time." 

    (while vendor here is skilled - his method can never succeed - his MCU is not built to allow that!)   Real issue is the degree of "toggle precision" your *unknown/undescribed" external IC or device requires.   Yet still - vendor's suggestion is "Band-Aid" - his chip cannot achieve simultaneous toggle across multiple ports!
     
    If (likely when) you require simultaneous control of "beyond 8 pins - or pins from different ports" - unless you employ this vendor's 129 series and operate under "EPI" - you are "out of luck."

    Your whole "hi-speed bus" idea - and vendor's further detailing - I believe to be misguided!   Issue is NOT speed - it is simultaneous data toggle across multiple ports - which vendor's MCUs (extra 129 series) cannot do!.   Staggered data transfer is staggered transfer - no matter how "close" the port to port delay!

    Method I suggested involves your employing separate latch ICs - external to your MCU - which may then be "loaded" by multiple ports w/in this vendor's common MCUs (any of them). There really is no limit to the number of bits you may then load - and finally transfer.

    By choosing a "clocked latch" - all such data - emanating from multiple MCU Ports - can be "simultaneously" clocked via one GPIO. (not among the ones which are to be latched)

    In this method - all signals (even those from different ports) at the output of one or multiple latches - will "toggle together" (via your clock) - just as you seek...