STM32F407 Matrix Computation

[tjes50220]

Posted on April 30, 2015 at 08:07

Hello, 

I use STM32F407  discovery board and keil uvision4.        I want to do matrix communication but I meet the following weird problems.        My code is as below.

float Num[2]= {0.000144962712488309, -0.000602215526678967};

    float Den[2]= {1, -5.62630205892449};          

  float ACCd[2];

        printf(''ACC 3 \n'');        for (int i=0; i<2; i++)        {        ACCd[i]=Num[i]*Den[i];        }      printf(''ACC 4 \n'');      It can compile but I can't see any result even the print of  

ACC 3

.    When I modified the code to      

float Num[2]= {0.000144962712488309, -0.000602215526678967};

     

float Den[2]= {1, -5.62630205892449};   

      

float ACCd[2];

        printf(''ACC 3 \n'');       ACCd[0]=Num[0]*Den[0];       ACCd[1]=Num[1]*Den[1];      printf(''ACC 4 \n'');       And it can work fine. Do I miss some constraints about the ST?Thanks #stm32 #matrix

[AvaTar]

Posted on April 30, 2015 at 12:15

I don't see a grave issue in the provided code.

Have turned off any optimisation, and did you try to single-step in a debugger ? However, there are two issues you migth stumble upon later:    float Num[2]= {0.000144962712488309, -0.000602215526678967}; First, these are not floating point, but double constants. Make it 'float' with an 'f' at the end (like '0.000144

f

'), otherwise this will impact performance. Second, 'float' is not capable of holding the precision you specified. Check if/how this affects your algorithm.

[Tesla DeLorean]

Posted on April 30, 2015 at 13:37

Not sure I like the C++ syntax, and the lack of real context of the code, but it looks reasonable enough.

I'd make sure I pulled in math.h, check the MicroLib, FPU and optimization options.

[tjes50220]

Posted on May 03, 2015 at 17:47I found that the original sample code didn't enable the fpu.I just need to enable.