Tool/software: Code Composer Studio
Hi All,
I am trying to run a particular code which needs more than 32,768 bytes.
Is there some way I can use the blank memory (93% of FLASH 262kbytes) as shown in the image below?
Is there a way to partition the memory as per user's need? If yes, which file should I update?
Thanks in advance for your help
Rohan Chawhan said:I have somehow managed to bring down the SRAM usage to 44% using the same arrays for different data storage at different time. (NOT tested yet... will update the result here once done!)
As I had mentioned in this earlier post to update the results
Here they are -
//----include libraries
//----variable declarations
extern int sin_tab[198]; // look up table for sine wave 1 period 198 points
extern float B[257]; // FIR fiter coefficients ;order 256
main(void)
{
//Clock is set at 80MHz
SysCtlClockSet(SYSCTL_SYSDIV_2_5 | SYSCTL_USE_PLL | SYSCTL_XTAL_16MHZ | SYSCTL_OSC_MAIN);
//Initialize ADC,GPIOE,TIMER
SysCtlPeripheralEnable(SYSCTL_PERIPH_GPIOE); //Pin for ADC input
SysCtlPeripheralEnable(SYSCTL_PERIPH_ADC0); //ADC peripheral
SysCtlPeripheralEnable(SYSCTL_PERIPH_TIMER0); //Timer0 Peripheral
//Initialize the data acquisition array with zeroes
for(i=0;i<1500;i++)
{
data0[i]=0;
}
i=0;
//Configure TIMER0
TimerConfigure(TIMER0_BASE, TIMER_CFG_SPLIT_PAIR | TIMER_CFG_A_PERIODIC);
TimerLoadSet(TIMER0_BASE, TIMER_A, 0x1); //sampling @500KHz
TimerControlTrigger(TIMER0_BASE,TIMER_A,true);
//Configure ADC0
GPIOPinTypeADC(GPIO_PORTE_BASE, GPIO_PIN_3); //set GPIO_E3 for channel 1
ADCSequenceConfigure(ADC0_BASE, 3, ADC_TRIGGER_TIMER, 0); //sequencer 3
ADCSequenceStepConfigure(ADC0_BASE, 3, 0, ADC_CTL_CH0 | ADC_CTL_IE | ADC_CTL_END);
ADCSequenceEnable(ADC0_BASE, 3);
while(1)
{
//While c=1 the ADC will continue to store the Photodetector output in
//the data0[] array. The ADC sampling frequency will be defined by the
//Timer triggering and ADC0IntHandler handles the above stated operations.
if(c==1)
{
//Enable interrupts
TimerEnable(TIMER0_BASE, TIMER_A);
ADCIntEnable(ADC0_BASE, 3);
IntEnable(INT_ADC0SS3);
i=0;
c=2;
}
//While c=4 the acquired data is smoothened with a 9 points moving average filter
if(c==4)
{
TimerDisable(TIMER0_BASE, TIMER_A);
ADCIntDisable(ADC0_BASE, 3);
//IntDisable(INT_ADC0SS3);
c=5;
//........weighted 9 points MA filter.......//
for(i=4;i<1497;i++)
{
data0[i]=(data0[i-4]+data0[i-3]+data0[i-2]+data0[i]+data0[i]+data0[i+1]+data0[i+2]+data0[i+3]+data0[i+4])/9;
}
i=0;
c=6;
}
//While c=6 the signal is further processed, i.e. multiplied with
//reference sinusoid and then filtered out from 256 order FIR filter
if(c==6)
{
//........Multiplying by reference sinusoid.fs = 100ksps.......//
for(i=0;i<1500;i++)
{
//"(((float)sin_tab[((harm*6*i)+(int)(phase/1.81818181))%198])-500)*0.002" is a sine
//wave generated with look up table of 198 points within range 0-1000
// harm - controls the harmonic user wants to choose
// phase - controls phase delay in degrees(with 198 points user can jump by 1.818181 step)
mult[i] = data0[i]*(((float)sin_tab[((harm*6*i)+(int)(phase/1.81818181))%198])-500)*0.002;
}
m=0;
i=0;
//FIR Filter 0rder 256, B stores the filter coefficients
for(m = 0; m < 1200; m++)
{
for(i = 0; i < 257; i++)
{
temp = temp + B[256-i]*mult[m + i];
}
//i = 0;
lkin[m] = temp;
temp = 0;
}
}
}
}
ADC ISR
//read the ADC output in the variable "value"
ADCSequenceDataGet(ADC0_BASE, 3, value);
//check whether the "value" is greater than a threshold
//and update the status check variable "c"
if(c<3)
{
if(value[0]<(uint32_t)1600)
{
c=3;
i=0;
}
}
//Start storing the data in data0[]
//Upon completion update the status variable "c"
if(c==3)
{
if(i<1500)
{
data0[i++]=(float)value[0];
}
else
{
i=0;
c=4;
}
}
ADCIntClear(ADC0_BASE, 3);
Can this be further reduced and made effiecient? (I think it should be possible !)
Hey all
I know that below given query isn't related solely to this POST TITLE. But still I am posting it here because in earlier post I have provided the code snippet (almost full) and now starting a new thread for the below query felt unnecessary to me, as it is not much technical based.
So,
The query : - I want to start reading the ADC values at the falling edge as shown in the image below. Relating to the code given in earlier post, can anyone suggest any improvement to do that? Only the ADC data acquisition part, and not the smoothening and later stuffs
The signal is periodic so should look like a ramp-like signal starting from the lowest point and at the end it wouldn't matter if the full ramp is covered (or, may be slightly more/less than 1 period of ramp-like signal)