Optimization of fixed number of cycles in embedded C language

1、 Cycle optimization

（1） Loops are the focus of code optimization

（2）C The code mainly uses for Loop or while loop

2、 A fixed number of cycles

problem ： What is? ARM Write on for The most efficient way to cycle ？

example ： A fixed number of cycles

int checksum_v5(int *data)
{
    unsigned int i;
    int sum = 0;
    for(i = 0; i < 64; i++)
    {
            sum += *(data++);
    }
    return sum;
}

The assembly code of the above program is ：

checksum_v5_s

        MOV r2,#0                ;r2 = data
        
        MOV r0,#0                ; sum = 0

        MOV r1,#0                ;  i = 0, Counter

checksum_v5_loop

       LDR r3,[r2] #4               ;r3 = *(data++)

        ADD r1,r1,#1                        ;r1++

        CMP r1,#0x40                        ;compare i,64

        ADD r0,r3,r0                               ;sum +=  r3       

        BCC checksum_v5_loop                ;if(i < 64) loop

        MOV pc,r14                                      ;return sum

3、 Discuss ：
ARM In compilation , Only two instructions are needed to implement the loop
（1） The subtraction instruction is used as a loop count , At the same time, set the result condition flag
（2） A conditional branch instruction

4、 The cycle code is modified to ：

int checksum_v6(int *data)
{
    unsigned int i;
    int sum = 0;
    for(i = 64; i != 0; i--)
    {
            sum += *(data++);
    }
    return sum;
}

checksum_v6_s

        MOV r2,#0                ;r2 = data
        
        MOV r0,#0                ; sum = 0

        MOV r1,#0X40                ;  i = 64, Counter

checksum_v6_loop

       LDR r3,[r2] #4               ;r3 = *(data++)

       SUBS r1,r1,#1                ;i-- and set flags

        ADD r0,r3,r0                               ;sum +=  r3       

        BNE checksum_v6_loop                ;if(i != 0) loop

        MOV pc,r14                                      ;return sum

5、 summary

The cyclic variable in the cyclic body adopts the way of self decreasing rather than the way of self increasing , Assembly code should be optimized .