Posted on March 03, 2016 at 19:45Sure, if you compare the running of the two loops, the original code
executed several more instructions per loop, every loop iteration it
would read the same memory address into the same register, which it
didn't need to do. You can also eliminate the adds that increments
the pointer by using the stmia [store and increment after] instruction;
for a single store operation, stm and stmia both take 2 cycles. (On more
powerful (cortex-m3 and up) cores, you usually use postfix addressing,
i.e. str r0, [r1], #4 to do this, but postfix isn't supported on the m0,
stmia is, however).
In the old code, the loop part was
ldr r3, =_sidata
ldr r3, [r3, r1]
str r3, [r0, r1]
adds r1, r1, #4
ldr r0, =_sdata
ldr r3, =_edata
adds r2, r0, r1
cmp r2, r3
bcc CopyDataInit
5 memory operations x 2 cycles each = 10 cycles
+ 3 alu operations x 1 cycle each = 3 cycles
+ 1 branch opreation x 1 cycle (usu) = 1 cycle
For 14 cycles / loop. In the new code, the loop part is
ldmia r2!, {r3}
stmia r0!, {r3}
cmp r0, r1
bcc CopyDataInitializersLoop
2 memory/alu operations x 2 cycles each = 4 cycles
1 alu operation x 1 cycle each = 1 cycle
1 branch operation x 1 cycle (usu) = 1 cycle
For 6 cycles / loop.
Is this clear?
-Pat
Complete Old Copy Data:
movs r1, #0
b LoopCopyDataInit
CopyDataInit:
ldr r3, =_sidata
ldr r3, [r3, r1]
str r3, [r0, r1]
adds r1, r1, #4
LoopCopyDataInit:
ldr r0, =_sdata
ldr r3, =_edata
adds r2, r0, r1
cmp r2, r3
bcc CopyDataInit
Complete New Copy Data:
CopyDataInitializersStart:
ldr r0, =_sdata /* write to this addr */
ldr r1, =_edata /* until you get to this addr */
ldr r2, =_sidata /* reading from this addr */
b CopyDataInitializersEnterLoop
CopyDataInitializersLoop:
ldmia r2!, {r3}
stmia r0!, {r3}
CopyDataInitializersEnterLoop:
cmp r0, r1
bcc CopyDataInitializersLoop
