.endif
vmov.i16 q15, #0
-.ifc \txfm1,idct
-.ifc \txfm2,idct
+.ifc \txfm1\()_\txfm2,idct_idct
cmp r3, #1
bne 1f
@ DC-only for idct/idct
vmov q3, q2
b 2f
.endif
-.endif
1:
vld1.16 {d4-d7}, [r2,:128]
@ if only idct is involved.
@ The iadst also uses a few coefficients from
@ idct, so those always need to be loaded.
-.ifc \txfm1,iadst
- movrel r12, iadst8_coeffs
- vld1.16 {q1}, [r12,:128]!
- vpush {q4-q7}
+.ifc \txfm1\()_\txfm2,idct_idct
+ movrel r12, idct_coeffs
+ vpush {q4-q5}
vld1.16 {q0}, [r12,:128]
.else
-.ifc \txfm2,iadst
movrel r12, iadst8_coeffs
vld1.16 {q1}, [r12,:128]!
vpush {q4-q7}
vld1.16 {q0}, [r12,:128]
-.else
- movrel r12, idct_coeffs
- vpush {q4-q5}
- vld1.16 {q0}, [r12,:128]
-.endif
.endif
vmov.i16 q2, #0
vmov.i16 q3, #0
-.ifc \txfm1,idct
-.ifc \txfm2,idct
+.ifc \txfm1\()_\txfm2,idct_idct
cmp r3, #1
bne 1f
@ DC-only for idct/idct
vst1.16 {d4[0]}, [r2,:16]
b 2f
.endif
-.endif
1:
vld1.16 {q8-q9}, [r2,:128]!
vld1.16 {q10-q11}, [r2,:128]!
vst1.8 {d10}, [r3,:64], r1
vst1.8 {d11}, [r3,:64], r1
-.ifc \txfm1,iadst
- vpop {q4-q7}
+.ifc \txfm1\()_\txfm2,idct_idct
+ vpop {q4-q5}
.else
-.ifc \txfm2,iadst
vpop {q4-q7}
-.else
- vpop {q4-q5}
-.endif
.endif
bx lr
endfunc
.macro itxfm_func16x16 txfm1, txfm2
function ff_vp9_\txfm1\()_\txfm2\()_16x16_add_neon, export=1
-.ifc \txfm1,idct
-.ifc \txfm2,idct
+.ifc \txfm1\()_\txfm2,idct_idct
cmp r3, #1
beq idct16x16_dc_add_neon
-.endif
.endif
push {r4-r7,lr}
-.ifc \txfm1,iadst
- vpush {q4-q7}
-.else
-.ifc \txfm2,iadst
+.ifnc \txfm1\()_\txfm2,idct_idct
vpush {q4-q7}
-.endif
.endif
mov r7, sp
.endr
mov sp, r7
-.ifc \txfm1,iadst
- vpop {q4-q7}
-.else
-.ifc \txfm2,iadst
+.ifnc \txfm1\()_\txfm2,idct_idct
vpop {q4-q7}
-.endif
.endif
pop {r4-r7,pc}
endfunc