TMS320F28379D: How can I increase ram memory (.ebss memory error)

Hi Omer,

As you said, I added the section from LS0 to LS5 and other RAM sections, but as seen in the figure I added below,after this procedure, the values assigned to the variables on the expression screen are incorrect (All values are zero). For example, I define float w1=0.4, w2=0.3, w3=0.2, w4=0.1; but their values are very different in the expression window. This is also valid for other variables.

In addition, if I need, how can I sue flash like RAM?

Hello Baris,

Are you defining their values outside the scope of a function or within some part of the main program? Are you using COFF or EABI output format in your project properties?

bariscvs said:

In addition, if I need, how can I sue flash like RAM?

To set up with Flash, simply right click on the project and go to Build Configurations > Set Active > and select one of the Flash options appropriate for you. Next, right-click on the RAM linker command file and select "Exclude from build" and then right-click on the Flash linker command file and unselect "Exclude form build". This should set your project to run from Flash.

Best regards,

Omer Amir

Hella Omer, 

My section to define values is below. 

#include "Lab.h" // Main include file
#include "math.h"
#include "IQmathLib.h"
//#define _IQdiv(A,B) ((float)(A) / (float)(B))
#define _IQ24div(A,B) ((float)(A) / (float)(B))

int Vbat1, Vsc1, Ibat1, Isc1, Iyuk1; // Sensörlerden okunacak dijital veriler
float Vbat2, Vsc2, Ibat2, Isc2, Iyuk2; // İşlenen sensör verileri
int N=20, sayac, j, jj, Pbatref; // N: Tahmin bölgesi boyutu
float gamae[60][1], gama[10][1], zref[80][1],h=0.000005, Energy_sc_f=10, Energy_bat_f; // h adım aralığı
float A1, A2, oran;
int donusturucu1=1, k, l, m, n, o, p, iii, uye1, uye2, uye3, uye3, uye_son;
float T_anah, pulse, Pyuk, Preq, soc_bat, soc_sc;
float Pbat, Psc;
unsigned long int f_anah=10000, n1, N_dongu1;
float Pbatmak=1100, SOCbatmak=0.9, SOCscmak=0.9, SOCscth=0.4;
// MPC ön hazırlık parametre hesabı
int a=1;
float w1=0.4, w2=0.3, w3=0.2, w4=0.1;
float E[80][80], F[80][1], GG[80][1], NM[60][80], zmin[60][1], zmak[60][1];

In addition to this, I have some defining part in main kod as in below.

while(1) // endless loop - wait for an interrupt
{
GpioDataRegs.GPATOGGLE.bit.GPIO26 = 1;
// Ölçüm işlemi;
Vbat1= (AdcaResultRegs.ADCRESULT0);
Vsc1= (AdcaResultRegs.ADCRESULT2);
Ibat1= (AdcaResultRegs.ADCRESULT3);
Isc1= (AdcaResultRegs.ADCRESULT4);
Iyuk1= (AdcaResultRegs.ADCRESULT5);

// Ölçüm kalibrasyonu
Vbat2= (Vbat1)*0.002102245*0.4;
Vsc2= (Vsc1)*0.002102245*0.4;
Ibat2= (Ibat1)*0.002102245*0.4;
Isc2= (Isc1)*0.002102245*0.4;
Iyuk2= (Iyuk1)*0.002102245*0.4;

Pyuk=295;//Vbat2*Iyuk2;
soc_bat=0.8;//Ibat2;
soc_sc=1;//Isc2;

Definitions both outside the main code and inside the values do not appear in the express window.

Finally, I'm using COFF format.

Hello Baris,

If you define the variables individually instead of inline, is there any change? I know I had trouble initializing global variables in a single line, but I don't remember if I used the same syntax/variable type.

Try this:

float w1=0.4;
float w2=0.3;
float w3=0.2;
float w4=0.1;

Best regards,

Omer Amir

Dear Omer,

Sorry it took me a while to reply. In fact, my RAM memory problem was partially solved with your previous response.

My code has some big size matrix e.g. F[60][80], but with using this size, I receive the memory error message I mentioned before.

I can solve this error with decreasing the matrix size. For example F[60][60]. In fact, this version of matrix is satisfactory. But in future studies, if I need big size matrix over than [60][60], how can I solve.

Thanks...

Hello Baris,

bariscvs said:

My code has some big size matrix e.g. F[60][80], but with using this size, I receive the memory error message I mentioned before.

I assume you're talking about the program not fitting into the available memory. You can merge sections (there should be examples of this within the linker command file), but I don't think it's proper to split a 2D array across multiple memory units with how it gets accessed (I would need to double-check this if you need the array to be stored in RAM). Is there a reason you can't store the array on Flash instead?

bariscvs said:

I can solve this error with decreasing the matrix size. For example F[60][60]. In fact, this version of matrix is satisfactory. But in future studies, if I need big size matrix over than [60][60], how can I solve.

This is a limitation of the memory units allocated for the specific sections (with the same details as above).

Best regards,

Omer Amir