CCS: MSP430 + LEA - Issue with DSPLib

Hi Marco,

The function is actually operating correctly but your customer is using as if it were an integer multiply routine. However, the _q15 inputs are integer representations of floating point numbers. You can find more information on this data type in the Fixed-Point Data Type section of the DSPLib User's Guide.

In this particular case the integer values 1,2, 3, 4, etc are representations of 0.00003051758, 0.00006103515625, etc using the equation f_q15(x) = x *2^-15. So when src1[1] *src2[1] is executed the real result is 0.00006103515625 * 0.000152587890625 = 0.0000000093132257. Then when this number is converted back to an integer representation by dividing by 2^-15 the result is 0.00030517578 and the integer value of this number is 0. 

A better test of the function would be the following:

#include "driverlib.h"
#include "dsplib/include/DSPLib.h"

#define VECTOR_LENGTH   8
// Load data into shared Ram to work with LEA
// _Q15(x) macro converts floating point representation to integer representation
DSPLIB_DATA(src1,4)
_q15 src1[VECTOR_LENGTH]={_Q15(0.1),_Q15(0.2),_Q15(0.3),_Q15(0.4),_Q15(0.5),_Q15(0.6),_Q15(0.7),_Q15(0.8)};

DSPLIB_DATA(src2,4)
_q15 src2[VECTOR_LENGTH]={_Q15(0.1),_Q15(0.2),_Q15(0.3),_Q15(0.4),_Q15(0.5),_Q15(0.6),_Q15(0.7),_Q15(0.8)};

DSPLIB_DATA(dst,4)
_q15 dst[VECTOR_LENGTH];

_q15 expecResult[VECTOR_LENGTH] = {_Q15(0.01),_Q15(0.04),_Q15(0.09),_Q15(0.16),_Q15(0.25),_Q15(0.36),_Q15(0.49),_Q15(0.64)};

int main(void) {

    WDTCTL = WDTPW | WDTHOLD;       // Stop watchdog timer

    msp_status status; // Status of command

    P1OUT &= ~BIT0;
    P1DIR |= BIT0;

    //--------------------------------------------------------------------------------------------------------------
    msp_mpy_q15_params mpyParams; // data struct
    mpyParams.length = 8; //length of Vector
    // Run LEA mpy calculation
    status = msp_mpy_q15(&mpyParams, src1, src2, dst);
    __no_operation();
    if(status != MSP_SUCCESS)
    {
        P1OUT |= BIT0; //Signal there was a  calculation error
    }
    //--------------------------------------------------------------------------------------------------------------

    uint8_t i = 0;
    while(i < VECTOR_LENGTH)
    {
        if(dst[i] != expecResult[i])
        {
            P1OUT |= BIT0; //Signal there was a  calculation error
        }
        i++;
    }

    //--------------------------------------------------------------------------------------------------------------
    msp_add_q15_params addParams; // data struct
    addParams.length = 8; //length of Vector
    // Run LEA add calculation
    status = msp_add_q15(&addParams, src1, src2, dst);
    __no_operation();
    if(status != MSP_SUCCESS)
    {
        P1OUT |= BIT0; //Signal there was a  calculation error
    }
    //--------------------------------------------------------------------------------------------------------------
}

Let me know if you have any questions. 

Best regards, 
Caleb Overbay