+
+/* void scale2x_tiles8(void *dst, const void *src, int w8, int h) */
+function(scale2x_tiles8)
+ push { r4, r14 }
+
+ mov r4, r1
+ add r12, r0, #1024*2
+ mov r14, r2
+
+0:
+ vld1.u16 { q0 }, [ r1, :128 ]!
+ vld1.u16 { q2 }, [ r1, :128 ]!
+ vmov q1, q0
+ vmov q3, q2
+ vzip.16 q0, q1
+ vzip.16 q2, q3
+ subs r14, #2
+ vst1.u16 { q0, q1 }, [ r0, :128 ]!
+ vst1.u16 { q0, q1 }, [ r12, :128 ]!
+ blt 1f
+ vst1.u16 { q2, q3 }, [ r0, :128 ]!
+ vst1.u16 { q2, q3 }, [ r12, :128 ]!
+ bgt 0b
+1:
+ subs r3, #1
+ mov r14, r2
+ add r0, #1024*2*2
+ add r4, #1024*2
+ sub r0, r2, lsl #4+1
+ mov r1, r4
+ add r12, r0, #1024*2
+ bgt 0b
+ nop
+
+ pop { r4, pc }