Compiler Error

A number of the matrix programs I have written work correctly, computing the correct answers for more than half of a matrix, but then I get an error message than says:

"Exiting due to signal SIGNOFP
Coprocessor not available at eip=0001edf ...."

and so on.

What does this error message mean?

I attached a screen shot of the error message if that helps...

It means you did something wrong in your program. Probably something undefined.

Apart from figuring out that your compiler is DJGPP, there's not a lot I could suggest.

Since you have a cmd.exe, it means you're at least on NT, so I'm guessing that your processor is at least a pentium.

Plus the fact that it works for a while and then throws this error. If this error appeared at all, it was usually on the first attempted use of a co-processor.

Which brings us back to bugs in your code.

Do you think you could produce a short example which you could post complete, and which throws this error?

I know I had it working before, but I don't recall changing anything that would affect the output. This is just a small segment of the code of a much larger program. Please let me know if you find any errors, however obvious they may be, that I cannot seem to find (or possibly don't know about).

Code:

#include<stdio.h>
#include<math.h>

float scenarioB(void);

int size_PDP, size_PDP2, size_PPA, size_PPA2;

float infoexploitation(float PDP_dB[size_PDP][size_PDP2], float 

Power_per_angle_dB[size_PPA][size_PPA2]);


int main(void)
{
	scenarioB();
}


float scenarioB(void)
{

size_PDP=2, size_PDP2=9, size_PPA=2, size_PPA2=9;

float PDP_dB[2][9]= 

{{0,-5.4287,-2.5162,-5.8905,-9.1603,-12.5105,-15.6126,-18.7147,-21.8168},
		     

{0,10e-9,20e-9,30e-9,40e-9,50e-9,60e-9,70e-9,80e-9}};


float Power_per_angle_dB[2][9]= {{0, -5.4287, -10.8574, -16.2860, 

-21.7147, -1e300, -1e300, -1e300, -1e300}, 
{-1e300, -1e300, -3.2042, -6.3063, -9.4084, -12.5105,  -15.6126,  

-18.7147,  -21.8168}};

infoexploitation(PDP_dB, Power_per_angle_dB);

	
}



float infoexploitation(float PDP_dB[size_PDP][size_PDP2], float 

Power_per_angle_dB[size_PPA][size_PPA2])
{

	int i, j, k;

	printf("\n\n******Information*********\n");

	printf("\n\nPDP:\n");
	for(i=0;i<size_PDP;i++)
{	
	for(j=0;j<size_PDP2;j++)
{		printf("%e\t", PDP_dB[i][j]);
	}
	printf("\n");
}


 /*----------PDP in linear values---------*/

/*THIS IS THE PART OF THE PROGRAM THAT WORKS AND THEN CRASHES.  ACCORDING TO A SOURCE I FOUND...

SIGNOFP
    The No Co-processor signal. Generated if a co-processor (floating-point) instruction is encountered when no co-processor is installed (Int 07h). */

    float PPA_linear[size_PDP][size_PDP2];
 	printf("\n\nPPA Linear:\n");
 	for(i=0; i<size_PDP; i++)
 	{
 		
 		for(j=0; j<size_PDP2; j++)
 		{	
 					
 		PPA_linear[i+1][j+1] = pow(10, 

.1*Power_per_angle_dB[i][j]);
 		printf("%e\t", PPA_linear[i+1][j+1]);
 		}printf("\n");
 	} 


return 0;
}

Thanks.

Well, in the last double-for-loop at the bottom, you walk off the ends (both ends, i and j) of your array.

djgpp assumes that it's "alone" in the system, and it may well get confused by DOS not really being DOS and that Windows NT/2K/XP/Vista is actually managing the FP unit.

Why not get the gcc-mingw compiler and use that to produce a native Windows application rather than a 32-bit DOS extension executable. I don't see anything in the code that should not work in gcc.

Although I think the real problem is that you are using i+1 and j+1 in the infoexploitation function, and that goes outside the valid range of the variable.

Code:

float PPA_linear[size_PDP][size_PDP2];

should be:

Code:

float PPA_linear[size_PDP+1][size_PDP2+1];

--
Mats

Some layout and indentation wouldn't hurt either.

Originally Posted by Salem

Some layout and indentation wouldn't hurt either.

Sure, before I even tried to fix any other problem I did M->, M-<, C-x C-x, M-x indent-region in Emacs.

And there are several warnings about not returning a value from functions in the original code too, if -Wall is used. Those should also be fixed. But I'll leave that to the OP to do.

--
Mats

Thank you. The problem was that I looped past the values. For some reason, I thought that if I made the dimensions of the storing matrix [i+1][j+1], it would still work, the only difference being that instead of the first element being indexed as 0, it would be 1. I now see how that wouldn't work, but I'm still fairly new to programming. Sorry.

As for the lack of layout and indentation, I apologize. I quickly just cut and paste the parts of the larger program. I didn't think to go back and fix it up. Sorry.