iOS Neon assembler sample questions - ios

Question

Just trying http://api.madewithmarmalade.com/ExampleArmASM.html and using iOS; the program run if I comment out the loop and the res is printed as 28. But if not comment it out, it will abend without printing the res.
Any hint why and how to fix it.
Thanks in advance.
My code is as follows:
#include <stdio.h>
#include <stdlib.h>
#define ARRAY_SIZE 512
#if defined __arm__ && defined __ARM_NEON__
static int computeSumNeon(const int a[])
{
// Computes the sum of all elements in the input array
int res = 0;
asm(".align 4 \n\t" //dennis warning avoiding
"vmov.i32 q8, #0 \n\t" //clear our accumulator register
"mov r3, #512 \n\t" //Loop condition n = ARRAY_SIZE
// ".loop1: \n\t" // No loop add 0-7 works as 28
"vld1.32 {d0, d1, d2, d3}, [%[input]]! \n\t" //load 8 elements into d0, d1, d2, d3 = q0, q1
"pld [%[input]] \n\t" // preload next set of elements
"vadd.i32 q8, q0, q8 \n\t" // q8 += q0
"vadd.i32 q8, q1, q8 \n\t" // q8 += q1
"subs r3, r3, #8 \n\t" // n -= 8
// "bne .loop1 \n\t" // n == 0?
"vpadd.i32 d0, d16, d17 \n\t" // d0[0] = d16[0] + d16[1], d0[1] = d17[0] + d17[1]
"vpaddl.u32 d0, d0 \n\t" // d0[0] = d0[0] + d0[1]
"vmov.32 %[result], d0[0] \n\t"
: [result] "=r" (res) , [input] "+r" (a)
:
: "q0", "q1", "q8", "r3");
return res;
}
#else
static int computeSumNeon(const int a[])
{
int i, res = 0;
for (i = 0; i < ARRAY_SIZE; i++)
res += a[i];
}
#endif
...
#implementation AppDelegate
- (BOOL)application:(UIApplication *)application didFinishLaunchingWithOptions:(NSDictionary *)launchOptions {
// Override point for customization after application launch.
//int* inp;
int inp[ARRAY_SIZE];
//posix_memalign((void**)&inp, 64, ARRAY_SIZE*sizeof(int)); // Align to cache line size (64bytes on a cortex A8)
// Initialise the array with consecutive integers.
int i;
for (i = 0; i < ARRAY_SIZE; i++)
{
inp[i] = i;
}
for (i = 0; i < ARRAY_SIZE; i++)
{
printf("%i,", inp[i]);
}
printf("\n\n sum 0-7:%i\n", 0+1+2+3+4+5+6+7);
int res = 0;
res = computeSumNeon(inp);
printf("res NEO :%i\n", res);
// free(inp); // error pointer being free was not allocated !!!
UISplitViewController *splitViewController = (UISplitViewController *)self.window.rootViewController;
UINavigationController *navigationController = [splitViewController.viewControllers lastObject];
navigationController.topViewController.navigationItem.leftBarButtonItem = splitViewController.displayModeButtonItem;
splitViewController.delegate = self;
return YES;
}
- (void)applicationWillResignActive:(UIApplication *)application {
...
==== assembly code generated
.align 1
.code 16 # #computeSumNeon
.thumb_func _computeSumNeon
_computeSumNeon:
Lfunc_begin3:
.loc 18 133 0 is_stmt 1 # ...
.cfi_startproc
# BB#0:
sub sp, #8
movs r1, #0
str r0, [sp, #4]
.loc 18 135 9 prologue_end # ...
Ltmp18:
str r1, [sp]
.loc 18 136 5 # ...
ldr r0, [sp, #4]
# InlineAsm Start
.align 4
vmov.i32 q8, #0x0
movw r3, #504
.loop1:
vld1.32 {d0, d1, d2, d3}, [r0]!
vadd.i32 q8, q0, q8
vadd.i32 q8, q1, q8
subs r3, #8
bne .loop1
vpadd.i32 d0, d16, d17
vpaddl.u32 d0, d0
vmov.32 r1, d0[0]
# InlineAsm End
str r1, [sp]
str r0, [sp, #4]
.loc 18 155 12 # ...
ldr r0, [sp]
.loc 18 155 5 is_stmt 0 # ...
add sp, #8
bx lr
Ltmp19:
Lfunc_end3:
.cfi_endproc