AVR Freaks

Some people don't like typing I guess.

I'm wondering if the compiler would be clever enough to reduce compile time constants down to zero RAM and a single instruction. Like say you had something like

typedef struct = {
    REAL_PORT *real_port,
    VIRT_PORT *virt_port,
    uint8_t mask
    bool virtual;
} PORT_PIN_t;

const PORT_PIN_t led_pin = {NULL, &VPORT0, (1<<5), true);

static void set_pin_high(PORT_PIN_t *pin)
{
    if (!virtual)
        pin->real_port->OUTSET = pin->mask;
    else
        pin->virt_port->OUT |= pin->mask;
}

You could probably create a load of macros to generate everything you wanted and have a nice easy "Arduino" style interface if you wanted to.

But yes, I agree with clawson, just type faster :-)

It's also worth noting that virtual ports can be slower in some circumstances. cbi/sbi can only act on one bit, where as the OUTSET/OUTCLR registers can operate on multiple bits.

Hay, the compiler did a good job :-) With optimization at -O1 and a real port:

static inline void pin_high(const PORT_PIN_t *pin)
{
	if (!pin->is_virtual)
		pin->real_port->OUTSET = pin->mask;
 22c:	81 e0       	ldi	r24, 0x01	; 1
 22e:	e0 e0       	ldi	r30, 0x00	; 0
 230:	f6 e0       	ldi	r31, 0x06	; 6
 232:	85 83       	std	Z+5, r24	; 0x05

And with a virtual port:

static inline void pin_high(const PORT_PIN_t *pin)
{
	if (!pin->is_virtual)
		pin->real_port->OUTSET = pin->mask;
	else
		pin->virtual_port->OUT |= pin->mask;
 22c:	88 9a       	sbi	0x11, 0	; 17

What I don't get is why the compiler bother going through Z in the first place (when it is used as a constant that is).

This would do the same, in less cycles and in less codespace:

ldi  r24, 0x01
sts  0x0605, r24

I never bothered with virtual ports. I find outclr/outset of greater value to me, avoiding the problems with non-atomic read-modify-writes.

What I like about XMega and bit-addressing, is the gpio registers at address 0-15. They are great for boolean flags. Setting and testing bits there is fast and atomic.

Since the Atmega328 need 2 Clocks to change PIN-Output  like PINA=1 , I was dissapointet to see

ATXMEGA 16A4 need 4 Clocks like PORTA.OUTTGL=1 .

My question is what is the 2 Clockcode  for  ATXMEGA to toggle a OUTPUT-PIN ?

I thought Virtual-Ports with direkt addressing may be faster therefore I place my question here ?

With Atxmega16A4 and Atxmega128a1 i measure 8Mhz with this code

PORTA.OUTTGL=1;PORTA.OUTTGL=1;PORTA.OUTTGL=1;................100xtimes

With my UNO (16Mhz Quarz)  :  PINA=1;PINA=1;PINA=1;........100xtimes

i measure 8Mhz !

How can I get the XMEGA-clock ?. It shoult be 32 Mhz with this

// Use 2Mhz internal RC with 16x PLL

  OSC.PLLCTRL = OSC_PLLSRC_RC2M_gc | 16 ;
  OSC.CTRL |= OSC_PLLEN_bm ; // enable the PLL...
  while( (OSC.STATUS & OSC_PLLRDY_bm) == 0 ){} // wait until it's stable

  // And now we can (finally) switch to the PLL as a clocksource.
  CCP = CCP_IOREG_gc;       // protected write follows   
  CLK.CTRL = CLK_SCLKSEL_PLL_gc;  // The System clock is now the PLL output (2MhzRC * 16)

You can do everything in one cycle with VPORTs.  SET, CLR or TGL.

Xmega does SBI in one cycle.   Don't ask me how.   Mega takes 2 cycles.

David.

No, you don't say PORTA.OUTSET = (1<<bit)

You say VPORT1.OUT |= (1<<bit) after assigning PORTA to VPORT1

The Compiler produces an SBI instruction because VPORTs are all in IN/OUT memory.

Xmegas have got lots of PORTs.  So you have to choose the "best" ones to access with the 4 possible VPORTs.

The XTinys have only got a few PORTs so you get VPORTs automagically.

David.

VPORT1.IN = (1<<bit) compiles as an OUT instruction.   Which is one cycle even on a Mega.

Go on.   Xmega is brilliant for GPIO.   One cycle for changing a single bit.    You can toggle 1-8 bits in one cycle with any AVR.   Since Xmega is clocked faster than a regular Mega you are always faster per cycle.

Yes,   you have to choose the best 4 PORTs to use with VPORT0-VPORT3 on an Xmega.

XTinys have VPORTA-VPORTD always available.

Xmega beats Cortex-M0 hands down.   And compares with Cortex-M3 and Cortex-M4 at faster clocks at least for external 8-bit bus e.g.8080.8 or 6800-8.

David.

The XMEGA has tremendous advantages over the MEGA. I'd say the most important are DMA, multi-level interrupts and the event system. Combining DMA with SPI you can get arbitrary signals with a resolution of 62.5ns; That's how I drive WS2812 LEDs.

As I suggested before, start a new thread describing what you're trying to achieve.

Thx David,

now I can toggle 16Mhz with Virtual-Ports .

I dont understand your VPORT2.IN=.. , I think a read-Command is use as  x=VPORT2.IN .

I test  virtual Read with your sketch and it works

#define readpin (1<<5)        //PC5 pin  connect with MASK

while(1)  //toggle Mask-Pin till Virtual readError    (Scope-Test)
    {
        VPORT2.OUT|= MASK;  asm("NOP"); asm("NOP");      //2 NOPS needed ! ! !
        if(VPORT2.IN & readpin) {VPORT2.OUT &=~MASK;  asm("NOP");  asm("NOP");  if(VPORT2.IN & readpin) break;} else break;
    }

/*
    while(1) //toggle Mask-Pin till readError    (Scope-Test)
    {
        PORTC.OUT|= MASK;asm("NOP");   //1NOP needed
        if(PORTC.IN & readpin) {PORTC.OUT &=~MASK;asm("NOP");if(PORTC.IN & readpin) break;} else break;
    }
*/

Do you know reduce the NOPS with Config the  readpin :PINnCTRL – Pin n Configuration register ?

(The readpin is only connect with the MASK-Pin )

John