comparison arm/dsputil_armv6.S @ 11117:ad6d17b36a3a libavcodec

ARMv6 optimised pix_sum

11116:0198e3582544

        bgt             1b

        mov             r0,  lr
        pop             {r4-r6, pc}
.endfunc

11117:ad6d17b36a3a

        bgt             1b

        mov             r0,  lr
        pop             {r4-r6, pc}
.endfunc

function ff_pix_sum_armv6, export=1
        push            {r4-r7, lr}
        mov             r12, #16
        mov             r2,  #0
        mov             r3,  #0
        mov             lr,  #0
        ldr             r4,  [r0]
1:
        subs            r12, r12, #1
        ldr             r5,  [r0, #4]
        usada8          r2,  r4,  lr,  r2
        ldr             r6,  [r0, #8]
        usada8          r3,  r5,  lr,  r3
        ldr             r7,  [r0, #12]
        usada8          r2,  r6,  lr,  r2
        beq             2f
        ldr             r4,  [r0, r1]!
        usada8          r3,  r7,  lr,  r3
        bgt             1b
2:
        usada8          r3,  r7,  lr,  r3
        add             r0,  r2,  r3
        pop             {r4-r7, pc}
.endfunc