The difference between float and double accuracy in Linux

When writing code on demand today, it is occasionally found that when you set the variable type float and double calculations under Linux,

The results are not the same.

Requirement: Set value = Incoming Value * 10 * 122.88/1000;

Case: Set value = 1666*10*122.88/1000

= 2047.1808

When set to float, the code:

#include <stdio.h> #include <math.h>unsigned int fun (unsigned int sfn_threshold) {        float f_sfn_ threshold;        printf ("Input parameter=%d\n", sfn_threshold);        F_sfn_threshold = (float) sfn_threshold*1.2288;        printf ("F_sfn_threshold =%8lf\n", f_sfn_threshold);        printf ("%.8lf\n", Fabs (F_sfn_threshold-(unsigned int) f_sfn_threshold));        return 0;} int main (int argc, char **argv) {        unsigned int a=1666;        unsigned int b=10000;        unsigned int c=12888;        unsigned int d=65535;        Fun (a);        return 0;}

Execution results (inconsistent with the results to be obtained):

[Email protected] test_float_compare]# gcc test_float_double_diff.c-o test_float_double_diff_1666[[email protected] Test_float_compare]#[[email protected] test_float_compare]#./test_float_double_diff_1666input PARAMETER=1666F_SFN _threshold = 2047.1807860.18078613[[email protected] test_float_compare]#

When set to double, the code:

#include <stdio.h> #include <math.h>unsigned int fun (unsigned int sfn_threshold) {        double f_sfn_ threshold;        printf ("Input parameter=%d\n", sfn_threshold);        F_sfn_threshold = (double) sfn_threshold*1.2288;        printf ("F_sfn_threshold =%8lf\n", f_sfn_threshold);        printf ("%.8lf\n", Fabs (F_sfn_threshold-(unsigned int) f_sfn_threshold));        return 0;} int main (int argc, char **argv) {        unsigned int a=1666;        unsigned int b=10000;        unsigned int c=12888;        unsigned int d=65535;        Fun (a);        return 0;}

Execution Result:

[[Email protected] test_float_compare]#[[email protected] test_float_compare]#./test_float_double_diff_1666input Parameter=1666f_sfn_threshold = 2047.1808000.18080000[[email protected] test_float_compare]#

The difference between float and double accuracy in Linux