Tool/software: TI-RTOS
Hello,
I measured two GPIO outputs executing following code(CPU frequency is 600MHz).
GPIO_write(7, GPIO_PIN_VAL_HIGH);
GPIO_write(8, GPIO_PIN_VAL_HIGH);
Measurement result is as below.
I think the delay time is seemed to be too long between two GPIO outputs.
What are the possible causes?
Is it overhead of GPIO driver ?
I also attached CCS Project.
Regards,
U-SK
U-SK,
Please refer to GPIO toggle experiment that is provided in the document below:
http://www.ti.com/lit/ug/tidu461/tidu461.pdf
Screenshot for easy reference:
The PRU is likely better at toggling a simple GPIO than ARM core.
Having said that the number reported are a little on the high side so it could be possible that your build options or the driver code is adding to the absolute best numbers that can be obtained from the ARM. To reduce this the GPIO toggle can be done using lower level register write or using functional CSL API. To do this directly from the register you can use GPIO_DATAOUT or (GPIO_CLRDATAOUT/GPIO_SETDATAOUT ) as described in code above.
The GPIO_write API from the GPIO LLD driver uses indexing so the value passed from the API converts index to port number and pin number and also seems to have interaction to check state of interrupts. YOu can locate the source for the API at GPIO_v1.c in the folder
pdk_am437x_1_0_10\packages\ti\drv\gpio\src\v1
Hope this helps.
Regards,
Rahul
Hi Rahul,
Thank you for your reply.
I added following function and measured gpio toggle time.
BitToggle(SOC_GPIO0_REG + GPIO_SETDATAOUT,SOC_GPIO0_REG + GPIO_CLEARDATAOUT);
The contents of the function are as follows:
unsigned long BitToggle(unsigned long val1, unsigned long val2){
asm(
" mov r2, #0x00004000" "\n\t"
"str r2,[r0]" "\n\t"
"str r2,[r1]" "\n\t"
);
return val1;
}
The measured result is as below.(TMDSIDK437x is running at 600MHz)
I think it seems to be long compered to your bench data.
What kind of device is used for the following data?
How many operating frequencies does the device run?
Regards,
U-SK
