comparison arm/dsputil_arm_s.S @ 8681:624ef0eb38f1 libavcodec

ARM: reorder some instructions in put_pixels*_arm for speed gains

8680:83453ea78558

        and r10, r6, r14
        and r11, r7, r14
        andeq r14, r14, r14, \rnd #1
        add r8, r8, r10
        add r9, r9, r11
        addeq r8, r8, r14
        addeq r9, r9, r14
        ldr r14, =0xfcfcfcfc >> 2
        and r4, r14, r4, lsr #2
        and r5, r14, r5, lsr #2
        and r6, r14, r6, lsr #2
        and r7, r14, r7, lsr #2
        add r10, r4, r6
        add r11, r5, r7
        subs r3, r3, #1
.endm

6:      stmfd sp!, {r8-r11}
        RND_XY2_IT \align, \rnd
        ldmfd sp!, {r4-r7}
        add r4, r4, r8
        add r5, r5, r9
        add r6, r6, r10
        add r7, r7, r11
        ldr r14, =0x0f0f0f0f
        and r4, r14, r4, lsr #2
        and r5, r14, r5, lsr #2
        add r4, r4, r6
        add r5, r5, r7
        stmia r0, {r4-r5}

8681:624ef0eb38f1

        and r10, r6, r14
        and r11, r7, r14
        andeq r14, r14, r14, \rnd #1
        add r8, r8, r10
        add r9, r9, r11
        ldr r12, =0xfcfcfcfc >> 2
        addeq r8, r8, r14
        addeq r9, r9, r14
        and r4, r12, r4, lsr #2
        and r5, r12, r5, lsr #2
        and r6, r12, r6, lsr #2
        and r7, r12, r7, lsr #2
        add r10, r4, r6
        add r11, r5, r7
        subs r3, r3, #1
.endm

6:      stmfd sp!, {r8-r11}
        RND_XY2_IT \align, \rnd
        ldmfd sp!, {r4-r7}
        add r4, r4, r8
        add r5, r5, r9
        ldr r14, =0x0f0f0f0f
        add r6, r6, r10
        add r7, r7, r11
        and r4, r14, r4, lsr #2
        and r5, r14, r5, lsr #2
        add r4, r4, r6
        add r5, r5, r7
        stmia r0, {r4-r5}