Push clobbered neon registers on the stack
According to the calling convention, the registers q4-q7 should be preserved by functions. The caller (generated by the compiler) could be using those registers anywhere for any intermediate data. Functions that use more than 12 of the qX registers must push the clobbered registers on the stack in order to be able to restore them afterwards. In functions that don't use all 16 registers, but clobber some of the callee saved registers q4-q7, one or more of them are remapped to reduce the number of registers that have to be saved/restored. This incurs a very small (around 0.5%) slowdown in the decoder and encoder.
This commit is contained in:
Martin Storsjö
parent
811c647c0e
commit
c011890764
@ -277,6 +277,7 @@
|
||||
#endif
|
||||
|
||||
WELS_ASM_FUNC_BEGIN DeblockLumaLt4V_neon
|
||||
vpush {q4-q7}
|
||||
vdup.u8 q11, r2
|
||||
vdup.u8 q9, r3
|
||||
|
||||
@ -290,7 +291,7 @@ WELS_ASM_FUNC_BEGIN DeblockLumaLt4V_neon
|
||||
vld1.u8 {q5}, [r0]
|
||||
sub r2, r2, r1
|
||||
|
||||
ldr r3, [sp, #0]
|
||||
ldr r3, [sp, #64]
|
||||
vld1.s8 {d31}, [r3]
|
||||
vdup.s8 d28, d31[0]
|
||||
vdup.s8 d30, d31[1]
|
||||
@ -332,10 +333,12 @@ WELS_ASM_FUNC_BEGIN DeblockLumaLt4V_neon
|
||||
vst1.u8 {q3}, [r2] , r1
|
||||
vst1.u8 {q7}, [r2]
|
||||
|
||||
vpop {q4-q7}
|
||||
WELS_ASM_FUNC_END
|
||||
|
||||
|
||||
WELS_ASM_FUNC_BEGIN DeblockLumaEq4V_neon
|
||||
vpush {q4-q7}
|
||||
|
||||
vdup.u8 q5, r2
|
||||
vdup.u8 q4, r3
|
||||
@ -393,10 +396,12 @@ WELS_ASM_FUNC_BEGIN DeblockLumaEq4V_neon
|
||||
DIFF_LUMA_EQ4_MASK q3, q14, q0, q4
|
||||
vst1.u8 {q4}, [r3], r1
|
||||
|
||||
vpop {q4-q7}
|
||||
WELS_ASM_FUNC_END
|
||||
|
||||
|
||||
WELS_ASM_FUNC_BEGIN DeblockLumaLt4H_neon
|
||||
vpush {q4-q7}
|
||||
|
||||
vdup.u8 q11, r2
|
||||
vdup.u8 q9, r3
|
||||
@ -429,7 +434,7 @@ WELS_ASM_FUNC_END
|
||||
|
||||
sub r0, r0, r1, lsl #4
|
||||
|
||||
ldr r3, [sp, #0]
|
||||
ldr r3, [sp, #64]
|
||||
vld1.s8 {d31}, [r3]
|
||||
vdup.s8 d28, d31[0]
|
||||
vdup.s8 d30, d31[1]
|
||||
@ -488,10 +493,12 @@ WELS_ASM_FUNC_END
|
||||
STORE_LUMA_DATA_4 d6, d7, d8, d9, 4, 5
|
||||
STORE_LUMA_DATA_4 d6, d7, d8, d9, 6, 7
|
||||
|
||||
vpop {q4-q7}
|
||||
WELS_ASM_FUNC_END
|
||||
|
||||
|
||||
WELS_ASM_FUNC_BEGIN DeblockLumaEq4H_neon
|
||||
vpush {q4-q7}
|
||||
vdup.u8 q5, r2
|
||||
vdup.u8 q4, r3
|
||||
|
||||
@ -591,6 +598,7 @@ WELS_ASM_FUNC_BEGIN DeblockLumaEq4H_neon
|
||||
STORE_LUMA_DATA_3 d7,d8,d9,d13,d14,d15,6
|
||||
STORE_LUMA_DATA_3 d7,d8,d9,d13,d14,d15,7
|
||||
|
||||
vpop {q4-q7}
|
||||
WELS_ASM_FUNC_END
|
||||
|
||||
|
||||
@ -647,9 +655,10 @@ WELS_ASM_FUNC_END
|
||||
|
||||
|
||||
WELS_ASM_FUNC_BEGIN DeblockChromaEq4V_neon
|
||||
vpush {q4-q5}
|
||||
|
||||
vdup.u8 q11, r3
|
||||
ldr r3, [sp, #0]
|
||||
ldr r3, [sp, #32]
|
||||
|
||||
sub r0, r0, r2 , lsl #1
|
||||
sub r1, r1, r2, lsl #1
|
||||
@ -670,10 +679,10 @@ WELS_ASM_FUNC_END
|
||||
|
||||
vmov q11, q10
|
||||
|
||||
DIFF_CHROMA_EQ4_P0Q0 d0, d2, d4, d6, q4, q5, q6, d14, d0 // Cb::p0' q0'
|
||||
DIFF_CHROMA_EQ4_P0Q0 d1, d3, d5, d7, q12, q13, q14, d15, d1 // Cr::p0' q0'
|
||||
DIFF_CHROMA_EQ4_P0Q0 d0, d2, d4, d6, q4, q5, q8, d30, d0 // Cb::p0' q0'
|
||||
DIFF_CHROMA_EQ4_P0Q0 d1, d3, d5, d7, q12, q13, q14, d31, d1 // Cr::p0' q0'
|
||||
|
||||
vbsl.u8 q10, q7, q1
|
||||
vbsl.u8 q10, q15, q1
|
||||
vst1.u8 {d20}, [r0], r2
|
||||
vst1.u8 {d21}, [r1], r2
|
||||
|
||||
@ -681,6 +690,7 @@ WELS_ASM_FUNC_END
|
||||
vst1.u8 {d22}, [r0]
|
||||
vst1.u8 {d23}, [r1]
|
||||
|
||||
vpop {q4-q5}
|
||||
WELS_ASM_FUNC_END
|
||||
|
||||
WELS_ASM_FUNC_BEGIN DeblockChromaLt4H_neon
|
||||
@ -747,8 +757,9 @@ WELS_ASM_FUNC_BEGIN DeblockChromaLt4H_neon
|
||||
WELS_ASM_FUNC_END
|
||||
|
||||
WELS_ASM_FUNC_BEGIN DeblockChromaEq4H_neon
|
||||
vpush {q4-q5}
|
||||
vdup.u8 q11, r3
|
||||
ldr r3, [sp, #0]
|
||||
ldr r3, [sp, #32]
|
||||
|
||||
sub r0, r0, #2
|
||||
sub r1, r1, #2
|
||||
@ -792,6 +803,7 @@ WELS_ASM_FUNC_BEGIN DeblockChromaEq4H_neon
|
||||
STORE_CHROMA_DATA_4 d0, d1, d2, d3, d4, d5, d6, d7, 6
|
||||
STORE_CHROMA_DATA_4 d0, d1, d2, d3, d4, d5, d6, d7, 7
|
||||
|
||||
vpop {q4-q5}
|
||||
WELS_ASM_FUNC_END
|
||||
|
||||
|
||||
|
||||
@ -1140,7 +1140,8 @@ WELS_ASM_FUNC_END
|
||||
|
||||
WELS_ASM_FUNC_BEGIN McHorVer22WidthEq16_neon
|
||||
push {r4}
|
||||
ldr r4, [sp, #4]
|
||||
vpush {q4-q7}
|
||||
ldr r4, [sp, #68]
|
||||
|
||||
sub r0, #2 //src[-2]
|
||||
sub r0, r0, r1, lsl #1 //src[-2*src_stride-2]
|
||||
@ -1246,13 +1247,15 @@ w16_hv_mc_luma_loop:
|
||||
sub r4, #4
|
||||
cmp r4, #0
|
||||
bne w16_hv_mc_luma_loop
|
||||
vpop {q4-q7}
|
||||
pop {r4}
|
||||
WELS_ASM_FUNC_END
|
||||
|
||||
|
||||
WELS_ASM_FUNC_BEGIN McHorVer22WidthEq8_neon
|
||||
push {r4}
|
||||
ldr r4, [sp, #4]
|
||||
vpush {q4}
|
||||
ldr r4, [sp, #20]
|
||||
|
||||
sub r0, #2 //src[-2]
|
||||
sub r0, r0, r1, lsl #1 //src[-2*src_stride-2]
|
||||
@ -1275,66 +1278,68 @@ WELS_ASM_FUNC_BEGIN McHorVer22WidthEq8_neon
|
||||
|
||||
w8_hv_mc_luma_loop:
|
||||
|
||||
vld1.u8 {q5}, [r0], r1 //use 13(8+5), =src[3]
|
||||
vld1.u8 {q8}, [r0], r1 //use 13(8+5), =src[3]
|
||||
//the 1st row
|
||||
pld [r0]
|
||||
// vertical filtered into q6/q7
|
||||
FILTER_6TAG_8BITS_TO_16BITS d0, d2, d4, d6, d8, d10, q6, q14, q15 // 8 avail
|
||||
FILTER_6TAG_8BITS_TO_16BITS d1, d3, d5, d7, d9, d11, q7, q14, q15 // 5 avail
|
||||
// vertical filtered into q9/q10
|
||||
FILTER_6TAG_8BITS_TO_16BITS d0, d2, d4, d6, d8, d16, q9, q14, q15 // 8 avail
|
||||
FILTER_6TAG_8BITS_TO_16BITS d1, d3, d5, d7, d9, d17, q10, q14, q15 // 5 avail
|
||||
// horizon filtered
|
||||
UNPACK_2_16BITS_TO_ABC q6, q7, q11, q12, q13
|
||||
FILTER_3_IN_16BITS_TO_8BITS q11, q12, q13, d12 //output to q6[0]
|
||||
vst1.u8 d12, [r2], r3 //write 8Byte
|
||||
UNPACK_2_16BITS_TO_ABC q9, q10, q11, q12, q13
|
||||
FILTER_3_IN_16BITS_TO_8BITS q11, q12, q13, d18 //output to q9[0]
|
||||
vst1.u8 d18, [r2], r3 //write 8Byte
|
||||
|
||||
vld1.u8 {q0}, [r0], r1 //read 2nd row
|
||||
//the 2nd row
|
||||
pld [r0]
|
||||
// vertical filtered into q6/q7
|
||||
FILTER_6TAG_8BITS_TO_16BITS d2, d4, d6, d8, d10, d0, q6, q14, q15 // 8 avail
|
||||
FILTER_6TAG_8BITS_TO_16BITS d3, d5, d7, d9, d11, d1, q7, q14, q15 // 5 avail
|
||||
// vertical filtered into q9/q10
|
||||
FILTER_6TAG_8BITS_TO_16BITS d2, d4, d6, d8, d16, d0, q9, q14, q15 // 8 avail
|
||||
FILTER_6TAG_8BITS_TO_16BITS d3, d5, d7, d9, d17, d1, q10, q14, q15 // 5 avail
|
||||
// horizon filtered
|
||||
UNPACK_2_16BITS_TO_ABC q6, q7, q11, q12, q13
|
||||
FILTER_3_IN_16BITS_TO_8BITS q11, q12, q13, d12 //output to q6[0]
|
||||
vst1.u8 d12, [r2], r3 //write 8Byte
|
||||
UNPACK_2_16BITS_TO_ABC q9, q10, q11, q12, q13
|
||||
FILTER_3_IN_16BITS_TO_8BITS q11, q12, q13, d18 //output to q9[0]
|
||||
vst1.u8 d18, [r2], r3 //write 8Byte
|
||||
|
||||
vld1.u8 {q1}, [r0], r1 //read 3rd row
|
||||
//the 3rd row
|
||||
pld [r0]
|
||||
// vertical filtered into q6/q7
|
||||
FILTER_6TAG_8BITS_TO_16BITS d4, d6, d8, d10, d0, d2, q6, q14, q15 // 8 avail
|
||||
FILTER_6TAG_8BITS_TO_16BITS d5, d7, d9, d11, d1, d3, q7, q14, q15 // 5 avail
|
||||
// vertical filtered into q9/q10
|
||||
FILTER_6TAG_8BITS_TO_16BITS d4, d6, d8, d16, d0, d2, q9, q14, q15 // 8 avail
|
||||
FILTER_6TAG_8BITS_TO_16BITS d5, d7, d9, d17, d1, d3, q10, q14, q15 // 5 avail
|
||||
// horizon filtered
|
||||
UNPACK_2_16BITS_TO_ABC q6, q7, q11, q12, q13
|
||||
FILTER_3_IN_16BITS_TO_8BITS q11, q12, q13, d12 //output to q6[0]
|
||||
vst1.u8 d12, [r2], r3 //write 8Byte
|
||||
UNPACK_2_16BITS_TO_ABC q9, q10, q11, q12, q13
|
||||
FILTER_3_IN_16BITS_TO_8BITS q11, q12, q13, d18 //output to q9[0]
|
||||
vst1.u8 d18, [r2], r3 //write 8Byte
|
||||
|
||||
vld1.u8 {q2}, [r0], r1 //read 4th row
|
||||
//the 4th row
|
||||
pld [r0]
|
||||
// vertical filtered into q6/q7
|
||||
FILTER_6TAG_8BITS_TO_16BITS d6, d8, d10, d0, d2, d4, q6, q14, q15 // 8 avail
|
||||
FILTER_6TAG_8BITS_TO_16BITS d7, d9, d11, d1, d3, d5, q7, q14, q15 // 5 avail
|
||||
// vertical filtered into q9/q10
|
||||
FILTER_6TAG_8BITS_TO_16BITS d6, d8, d16, d0, d2, d4, q9, q14, q15 // 8 avail
|
||||
FILTER_6TAG_8BITS_TO_16BITS d7, d9, d17, d1, d3, d5, q10, q14, q15 // 5 avail
|
||||
// horizon filtered
|
||||
UNPACK_2_16BITS_TO_ABC q6, q7, q11, q12, q13
|
||||
FILTER_3_IN_16BITS_TO_8BITS q11, q12, q13, d12 //output to q6[0]
|
||||
vst1.u8 d12, [r2], r3 //write 8Byte
|
||||
UNPACK_2_16BITS_TO_ABC q9, q10, q11, q12, q13
|
||||
FILTER_3_IN_16BITS_TO_8BITS q11, q12, q13, d18 //output to q9[0]
|
||||
vst1.u8 d18, [r2], r3 //write 8Byte
|
||||
|
||||
//q4~q5, q0~q2, --> q0~q4
|
||||
vswp q0, q4
|
||||
vswp q2, q4
|
||||
vmov q3, q1
|
||||
vmov q1, q5
|
||||
vmov q1, q8
|
||||
|
||||
sub r4, #4
|
||||
cmp r4, #0
|
||||
bne w8_hv_mc_luma_loop
|
||||
vpop {q4}
|
||||
pop {r4}
|
||||
WELS_ASM_FUNC_END
|
||||
|
||||
|
||||
WELS_ASM_FUNC_BEGIN McHorVer22WidthEq4_neon
|
||||
push {r4 ,r5, r6}
|
||||
ldr r6, [sp, #12]
|
||||
vpush {q4-q7}
|
||||
ldr r6, [sp, #76]
|
||||
|
||||
sub r0, #2 //src[-2]
|
||||
sub r0, r0, r1, lsl #1 //src[-2*src_stride-2]
|
||||
@ -1417,6 +1422,7 @@ w4_hv_mc_luma_loop:
|
||||
cmp r6, #0
|
||||
bne w4_hv_mc_luma_loop
|
||||
|
||||
vpop {q4-q7}
|
||||
pop {r4, r5, r6}
|
||||
WELS_ASM_FUNC_END
|
||||
|
||||
@ -1889,7 +1895,8 @@ WELS_ASM_FUNC_END
|
||||
|
||||
WELS_ASM_FUNC_BEGIN McHorVer22Width17_neon
|
||||
push {r4}
|
||||
ldr r4, [sp, #4]
|
||||
vpush {q4-q7}
|
||||
ldr r4, [sp, #68]
|
||||
|
||||
sub r0, #2 //src[-2]
|
||||
sub r0, r0, r1, lsl #1 //src[-2*src_stride-2]
|
||||
@ -2016,13 +2023,15 @@ w17_hv_mc_luma_loop:
|
||||
UNPACK_1_IN_8x16BITS_TO_8BITS d2, d22, d23, q11 //output to d2[0]
|
||||
vst1.u8 {d2[0]}, [r2], r3 //write 16th Byte
|
||||
|
||||
vpop {q4-q7}
|
||||
pop {r4}
|
||||
WELS_ASM_FUNC_END
|
||||
|
||||
|
||||
WELS_ASM_FUNC_BEGIN McHorVer22Width9_neon
|
||||
push {r4}
|
||||
ldr r4, [sp, #4]
|
||||
vpush {q4}
|
||||
ldr r4, [sp, #20]
|
||||
|
||||
sub r0, #2 //src[-2]
|
||||
sub r0, r0, r1, lsl #1 //src[-2*src_stride-2]
|
||||
@ -2046,78 +2055,79 @@ WELS_ASM_FUNC_BEGIN McHorVer22Width9_neon
|
||||
|
||||
w9_hv_mc_luma_loop:
|
||||
|
||||
vld1.u8 {q5}, [r0], r1 //use 14(9+5), =src[3]
|
||||
vld1.u8 {q8}, [r0], r1 //use 14(9+5), =src[3]
|
||||
//the 1st row
|
||||
pld [r0]
|
||||
// vertical filtered into q6/q7
|
||||
FILTER_6TAG_8BITS_TO_16BITS d0, d2, d4, d6, d8, d10, q6, q14, q15 // 8 avail
|
||||
FILTER_6TAG_8BITS_TO_16BITS d1, d3, d5, d7, d9, d11, q7, q14, q15 // 6 avail
|
||||
// vertical filtered into q9/q10
|
||||
FILTER_6TAG_8BITS_TO_16BITS d0, d2, d4, d6, d8, d16, q9, q14, q15 // 8 avail
|
||||
FILTER_6TAG_8BITS_TO_16BITS d1, d3, d5, d7, d9, d17, q10, q14, q15 // 6 avail
|
||||
// horizon filtered
|
||||
UNPACK_2_16BITS_TO_ABC q6, q7, q11, q12, q13
|
||||
FILTER_3_IN_16BITS_TO_8BITS q11, q12, q13, d12 //output to q6[0]
|
||||
vst1.u8 d12, [r2]! //write 8Byte
|
||||
UNPACK_1_IN_8x16BITS_TO_8BITS d13, d14, d15, q7 //output to d13[0]
|
||||
vst1.u8 {d13[0]}, [r2], r3 //write 8th Byte
|
||||
UNPACK_2_16BITS_TO_ABC q9, q10, q11, q12, q13
|
||||
FILTER_3_IN_16BITS_TO_8BITS q11, q12, q13, d18 //output to q9[0]
|
||||
vst1.u8 d18, [r2]! //write 8Byte
|
||||
UNPACK_1_IN_8x16BITS_TO_8BITS d19, d20, d21, q10 //output to d19[0]
|
||||
vst1.u8 {d19[0]}, [r2], r3 //write 8th Byte
|
||||
|
||||
vld1.u8 {q0}, [r0], r1 //read 2nd row
|
||||
//the 2nd row
|
||||
pld [r0]
|
||||
// vertical filtered into q6/q7
|
||||
FILTER_6TAG_8BITS_TO_16BITS d2, d4, d6, d8, d10, d0, q6, q14, q15 // 8 avail
|
||||
FILTER_6TAG_8BITS_TO_16BITS d3, d5, d7, d9, d11, d1, q7, q14, q15 // 6 avail
|
||||
// vertical filtered into q9/q10
|
||||
FILTER_6TAG_8BITS_TO_16BITS d2, d4, d6, d8, d16, d0, q9, q14, q15 // 8 avail
|
||||
FILTER_6TAG_8BITS_TO_16BITS d3, d5, d7, d9, d17, d1, q10, q14, q15 // 6 avail
|
||||
// horizon filtered
|
||||
UNPACK_2_16BITS_TO_ABC q6, q7, q11, q12, q13
|
||||
FILTER_3_IN_16BITS_TO_8BITS q11, q12, q13, d12 //output to q6[0]
|
||||
vst1.u8 d12, [r2]! //write 8Byte
|
||||
UNPACK_1_IN_8x16BITS_TO_8BITS d13, d14, d15, q7 //output to d13[0]
|
||||
vst1.u8 {d13[0]}, [r2], r3 //write 8th Byte
|
||||
UNPACK_2_16BITS_TO_ABC q9, q10, q11, q12, q13
|
||||
FILTER_3_IN_16BITS_TO_8BITS q11, q12, q13, d18 //output to q9[0]
|
||||
vst1.u8 d18, [r2]! //write 8Byte
|
||||
UNPACK_1_IN_8x16BITS_TO_8BITS d19, d20, d21, q10 //output to d19[0]
|
||||
vst1.u8 {d19[0]}, [r2], r3 //write 8th Byte
|
||||
|
||||
vld1.u8 {q1}, [r0], r1 //read 3rd row
|
||||
//the 3rd row
|
||||
pld [r0]
|
||||
// vertical filtered into q6/q7
|
||||
FILTER_6TAG_8BITS_TO_16BITS d4, d6, d8, d10, d0, d2, q6, q14, q15 // 8 avail
|
||||
FILTER_6TAG_8BITS_TO_16BITS d5, d7, d9, d11, d1, d3, q7, q14, q15 // 6 avail
|
||||
// vertical filtered into q9/q10
|
||||
FILTER_6TAG_8BITS_TO_16BITS d4, d6, d8, d16, d0, d2, q9, q14, q15 // 8 avail
|
||||
FILTER_6TAG_8BITS_TO_16BITS d5, d7, d9, d17, d1, d3, q10, q14, q15 // 6 avail
|
||||
// horizon filtered
|
||||
UNPACK_2_16BITS_TO_ABC q6, q7, q11, q12, q13
|
||||
FILTER_3_IN_16BITS_TO_8BITS q11, q12, q13, d12 //output to q6[0]
|
||||
vst1.u8 d12, [r2]! //write 8Byte
|
||||
UNPACK_1_IN_8x16BITS_TO_8BITS d13, d14, d15, q7 //output to d13[0]
|
||||
vst1.u8 {d13[0]}, [r2], r3 //write 8th Byte
|
||||
UNPACK_2_16BITS_TO_ABC q9, q10, q11, q12, q13
|
||||
FILTER_3_IN_16BITS_TO_8BITS q11, q12, q13, d18 //output to q9[0]
|
||||
vst1.u8 d18, [r2]! //write 8Byte
|
||||
UNPACK_1_IN_8x16BITS_TO_8BITS d19, d20, d21, q10 //output to d19[0]
|
||||
vst1.u8 {d19[0]}, [r2], r3 //write 8th Byte
|
||||
|
||||
vld1.u8 {q2}, [r0], r1 //read 4th row
|
||||
//the 4th row
|
||||
pld [r0]
|
||||
// vertical filtered into q6/q7
|
||||
FILTER_6TAG_8BITS_TO_16BITS d6, d8, d10, d0, d2, d4, q6, q14, q15 // 8 avail
|
||||
FILTER_6TAG_8BITS_TO_16BITS d7, d9, d11, d1, d3, d5, q7, q14, q15 // 6 avail
|
||||
// vertical filtered into q9/q10
|
||||
FILTER_6TAG_8BITS_TO_16BITS d6, d8, d16, d0, d2, d4, q9, q14, q15 // 8 avail
|
||||
FILTER_6TAG_8BITS_TO_16BITS d7, d9, d17, d1, d3, d5, q10, q14, q15 // 6 avail
|
||||
// horizon filtered
|
||||
UNPACK_2_16BITS_TO_ABC q6, q7, q11, q12, q13
|
||||
FILTER_3_IN_16BITS_TO_8BITS q11, q12, q13, d12 //output to q6[0]
|
||||
vst1.u8 d12, [r2]! //write 8Byte
|
||||
UNPACK_1_IN_8x16BITS_TO_8BITS d13, d14, d15, q7 //output to d13[0]
|
||||
vst1.u8 {d13[0]}, [r2], r3 //write 8th Byte
|
||||
UNPACK_2_16BITS_TO_ABC q9, q10, q11, q12, q13
|
||||
FILTER_3_IN_16BITS_TO_8BITS q11, q12, q13, d18 //output to q9[0]
|
||||
vst1.u8 d18, [r2]! //write 8Byte
|
||||
UNPACK_1_IN_8x16BITS_TO_8BITS d19, d20, d21, q10 //output to d19[0]
|
||||
vst1.u8 {d19[0]}, [r2], r3 //write 8th Byte
|
||||
|
||||
//q4~q5, q0~q2, --> q0~q4
|
||||
//q4~q8, q0~q2, --> q0~q4
|
||||
vswp q0, q4
|
||||
vswp q2, q4
|
||||
vmov q3, q1
|
||||
vmov q1, q5
|
||||
vmov q1, q8
|
||||
|
||||
sub r4, #4
|
||||
cmp r4, #1
|
||||
bne w9_hv_mc_luma_loop
|
||||
//the last row
|
||||
vld1.u8 {q5}, [r0], r1 //use 14(9+5), =src[3]
|
||||
// vertical filtered into q6/q7
|
||||
FILTER_6TAG_8BITS_TO_16BITS d0, d2, d4, d6, d8, d10, q6, q14, q15 // 8 avail
|
||||
FILTER_6TAG_8BITS_TO_16BITS d1, d3, d5, d7, d9, d11, q7, q14, q15 // 6 avail
|
||||
vld1.u8 {q8}, [r0], r1 //use 14(9+5), =src[3]
|
||||
// vertical filtered into q9/q10
|
||||
FILTER_6TAG_8BITS_TO_16BITS d0, d2, d4, d6, d8, d16, q9, q14, q15 // 8 avail
|
||||
FILTER_6TAG_8BITS_TO_16BITS d1, d3, d5, d7, d9, d17, q10, q14, q15 // 6 avail
|
||||
// horizon filtered
|
||||
UNPACK_2_16BITS_TO_ABC q6, q7, q11, q12, q13
|
||||
FILTER_3_IN_16BITS_TO_8BITS q11, q12, q13, d12 //output to q6[0]
|
||||
vst1.u8 d12, [r2]! //write 8Byte
|
||||
UNPACK_1_IN_8x16BITS_TO_8BITS d13, d14, d15, q7 //output to d13[0]
|
||||
vst1.u8 {d13[0]}, [r2], r3 //write 8th Byte
|
||||
UNPACK_2_16BITS_TO_ABC q9, q10, q11, q12, q13
|
||||
FILTER_3_IN_16BITS_TO_8BITS q11, q12, q13, d18 //output to q9[0]
|
||||
vst1.u8 d18, [r2]! //write 8Byte
|
||||
UNPACK_1_IN_8x16BITS_TO_8BITS d19, d20, d21, q10 //output to d19[0]
|
||||
vst1.u8 {d19[0]}, [r2], r3 //write 8th Byte
|
||||
vpop {q4}
|
||||
pop {r4}
|
||||
WELS_ASM_FUNC_END
|
||||
|
||||
|
||||
@ -535,10 +535,11 @@ WELS_ASM_FUNC_END
|
||||
|
||||
WELS_ASM_FUNC_BEGIN WelsIntra8x8Combined3Satd_neon
|
||||
stmdb sp!, {r4-r7, lr}
|
||||
vpush {q4-q7}
|
||||
|
||||
//Get the data from stack
|
||||
ldr r4, [sp, #32] //p_dec_cr
|
||||
ldr r5, [sp, #36] //p_enc_cr
|
||||
ldr r4, [sp, #96] //p_dec_cr
|
||||
ldr r5, [sp, #100] //p_enc_cr
|
||||
|
||||
//Get the top line data to 'd29(cb), d31(cr)'(16 bytes)
|
||||
sub r6, r0, r1
|
||||
@ -637,8 +638,8 @@ WELS_ASM_FUNC_BEGIN WelsIntra8x8Combined3Satd_neon
|
||||
HDM_TRANSFORM_4X4_L0 d7, d9, d25, d21, d13, d11, d10, d28, d30
|
||||
|
||||
//Get the data from stack
|
||||
ldr r5, [sp, #20] //the addr of Best_mode
|
||||
ldr r6, [sp, #24] //the value of i_lambda
|
||||
ldr r5, [sp, #84] //the addr of Best_mode
|
||||
ldr r6, [sp, #88] //the value of i_lambda
|
||||
|
||||
vrshr.u16 d11, #1
|
||||
vpaddl.u16 d11, d11
|
||||
@ -668,6 +669,7 @@ WELS_ASM_FUNC_BEGIN WelsIntra8x8Combined3Satd_neon
|
||||
str r6, [r5]
|
||||
mov r0, lr
|
||||
|
||||
vpop {q4-q7}
|
||||
ldmia sp!, {r4-r7, lr}
|
||||
WELS_ASM_FUNC_END
|
||||
|
||||
|
||||
@ -38,85 +38,85 @@
|
||||
vld1.64 {q0}, [r0,:128], r1
|
||||
vld1.64 {q1}, [r2], r3
|
||||
|
||||
vsubl.u8 q4, d0, d2
|
||||
vld1.64 {q2}, [r0,:128], r1
|
||||
|
||||
vsubl.u8 q6, d1, d3
|
||||
vld1.64 {q3}, [r2], r3
|
||||
|
||||
vsubl.u8 q5, d4, d6
|
||||
vld1.64 {q0}, [r0,:128], r1
|
||||
|
||||
vsubl.u8 q7, d5, d7
|
||||
vld1.64 {q1}, [r2], r3
|
||||
|
||||
vsubl.u8 q8, d0, d2
|
||||
vsubl.u8 q8, d0, d2
|
||||
vld1.64 {q2}, [r0,:128], r1
|
||||
|
||||
vsubl.u8 q10, d1, d3
|
||||
vadd.s16 q0, q4, q5
|
||||
vld1.64 {q3}, [r2], r3
|
||||
|
||||
vsubl.u8 q9, d4, d6
|
||||
vld1.64 {q0}, [r0,:128], r1
|
||||
|
||||
vsubl.u8 q11, d5, d7
|
||||
vld1.64 {q1}, [r2], r3
|
||||
|
||||
vsubl.u8 q12, d0, d2
|
||||
vld1.64 {q2}, [r0,:128], r1
|
||||
|
||||
vsubl.u8 q14, d1, d3
|
||||
vadd.s16 q0, q8, q9
|
||||
|
||||
vld1.64 {q3}, [r2], r3
|
||||
vsub.s16 q1, q4, q5
|
||||
vsub.s16 q1, q8, q9
|
||||
|
||||
vsubl.u8 q9, d4, d6
|
||||
vsubl.u8 q11, d5, d7
|
||||
vsubl.u8 q13, d4, d6
|
||||
vsubl.u8 q15, d5, d7
|
||||
|
||||
vadd.s16 q2, q8, q9
|
||||
vsub.s16 q3, q8, q9
|
||||
vadd.s16 q2, q12, q13
|
||||
vsub.s16 q3, q12, q13
|
||||
|
||||
vadd.s16 q4, q6, q7
|
||||
vsub.s16 q5, q6, q7
|
||||
vadd.s16 q8, q10, q11
|
||||
vsub.s16 q9, q10, q11
|
||||
|
||||
vadd.s16 q6, q10, q11
|
||||
vsub.s16 q7, q10, q11
|
||||
vadd.s16 q10, q14, q15
|
||||
vsub.s16 q11, q14, q15
|
||||
|
||||
vadd.s16 q8, q0, q2
|
||||
vsub.s16 q10, q0, q2
|
||||
vadd.s16 q12, q0, q2
|
||||
vsub.s16 q14, q0, q2
|
||||
|
||||
vadd.s16 q9, q4, q6
|
||||
vsub.s16 q11, q4, q6
|
||||
vadd.s16 q13, q8, q10
|
||||
vsub.s16 q15, q8, q10
|
||||
|
||||
vsub.s16 q0, q1, q3
|
||||
vadd.s16 q2, q1, q3
|
||||
|
||||
vsub.s16 q1, q5, q7
|
||||
vadd.s16 q3, q5, q7
|
||||
vsub.s16 q1, q9, q11
|
||||
vadd.s16 q3, q9, q11
|
||||
|
||||
vtrn.16 q8, q10
|
||||
vtrn.16 q9, q11
|
||||
vtrn.16 q12, q14
|
||||
vtrn.16 q13, q15
|
||||
|
||||
vadd.s16 q4, q8, q10
|
||||
vabd.s16 q6, q8, q10
|
||||
vadd.s16 q8, q12, q14
|
||||
vabd.s16 q10, q12, q14
|
||||
|
||||
vadd.s16 q5, q9, q11
|
||||
vabd.s16 q7, q9, q11
|
||||
|
||||
vabs.s16 q4, q4
|
||||
vabs.s16 q5, q5
|
||||
|
||||
vtrn.16 q0, q2
|
||||
vtrn.16 q1, q3
|
||||
|
||||
vadd.s16 q8, q0, q2
|
||||
vabd.s16 q10, q0, q2
|
||||
|
||||
vadd.s16 q9, q1, q3
|
||||
vabd.s16 q11, q1, q3
|
||||
vadd.s16 q9, q13, q15
|
||||
vabd.s16 q11, q13, q15
|
||||
|
||||
vabs.s16 q8, q8
|
||||
vabs.s16 q9, q9
|
||||
|
||||
vtrn.32 q4, q6
|
||||
vtrn.32 q5, q7
|
||||
vtrn.16 q0, q2
|
||||
vtrn.16 q1, q3
|
||||
|
||||
vadd.s16 q12, q0, q2
|
||||
vabd.s16 q14, q0, q2
|
||||
|
||||
vadd.s16 q13, q1, q3
|
||||
vabd.s16 q15, q1, q3
|
||||
|
||||
vabs.s16 q12, q12
|
||||
vabs.s16 q13, q13
|
||||
|
||||
vtrn.32 q8, q10
|
||||
vtrn.32 q9, q11
|
||||
|
||||
vmax.s16 q0, q4, q6
|
||||
vmax.s16 q1, q5, q7
|
||||
vmax.s16 q2, q8, q10
|
||||
vmax.s16 q3, q9, q11
|
||||
vtrn.32 q12, q14
|
||||
vtrn.32 q13, q15
|
||||
|
||||
vmax.s16 q0, q8, q10
|
||||
vmax.s16 q1, q9, q11
|
||||
vmax.s16 q2, q12, q14
|
||||
vmax.s16 q3, q13, q15
|
||||
|
||||
vadd.u16 q0, q0, q1
|
||||
vadd.u16 q2, q2, q3
|
||||
@ -128,49 +128,49 @@
|
||||
vld1.64 {d1}, [r2], r3
|
||||
|
||||
vld1.64 {d2}, [r0,:64], r1
|
||||
vsubl.u8 q4, d0, d1
|
||||
vsubl.u8 q8, d0, d1
|
||||
|
||||
vld1.64 {d3}, [r2], r3
|
||||
vsubl.u8 q5, d2, d3
|
||||
vsubl.u8 q9, d2, d3
|
||||
|
||||
vld1.64 {d4}, [r0,:64], r1
|
||||
vld1.64 {d5}, [r2], r3
|
||||
|
||||
vadd.s16 q8, q4, q5
|
||||
vsubl.u8 q6, d4, d5
|
||||
vadd.s16 q12, q8, q9
|
||||
vsubl.u8 q10, d4, d5
|
||||
|
||||
vld1.64 {d6}, [r0,:64], r1
|
||||
vld1.64 {d7}, [r2], r3
|
||||
|
||||
vsubl.u8 q7, d6, d7
|
||||
vsub.s16 q9, q4, q5
|
||||
vsubl.u8 q11, d6, d7
|
||||
vsub.s16 q13, q8, q9
|
||||
|
||||
vadd.s16 q10, q6, q7
|
||||
vsub.s16 q11, q6, q7
|
||||
vadd.s16 q14, q10, q11
|
||||
vsub.s16 q15, q10, q11
|
||||
|
||||
vadd.s16 q0, q8, q10
|
||||
vsub.s16 q1, q8, q10
|
||||
vadd.s16 q0, q12, q14
|
||||
vsub.s16 q1, q12, q14
|
||||
|
||||
vsub.s16 q2, q9, q11
|
||||
vadd.s16 q3, q9, q11
|
||||
vsub.s16 q2, q13, q15
|
||||
vadd.s16 q3, q13, q15
|
||||
|
||||
vtrn.16 q0, q1
|
||||
vtrn.16 q2, q3
|
||||
|
||||
vadd.s16 q4, q0, q1
|
||||
vabd.s16 q5, q0, q1
|
||||
vadd.s16 q8, q0, q1
|
||||
vabd.s16 q9, q0, q1
|
||||
|
||||
vabs.s16 q4, q4
|
||||
vadd.s16 q6, q2, q3
|
||||
vabs.s16 q8, q8
|
||||
vadd.s16 q10, q2, q3
|
||||
|
||||
vabd.s16 q7, q2, q3
|
||||
vabs.s16 q6, q6
|
||||
vabd.s16 q11, q2, q3
|
||||
vabs.s16 q10, q10
|
||||
|
||||
vtrn.32 q4, q5
|
||||
vtrn.32 q6, q7
|
||||
vtrn.32 q8, q9
|
||||
vtrn.32 q10, q11
|
||||
|
||||
vmax.s16 q0, q4, q5
|
||||
vmax.s16 q1, q6, q7
|
||||
vmax.s16 q0, q8, q9
|
||||
vmax.s16 q1, q10, q11
|
||||
.endm
|
||||
|
||||
.macro SAD_16x4
|
||||
@ -221,6 +221,7 @@
|
||||
|
||||
|
||||
WELS_ASM_FUNC_BEGIN WelsSampleSad16x16_neon
|
||||
vpush {q4-q7}
|
||||
|
||||
vld1.64 {q0}, [r0, :128], r1
|
||||
vld1.64 {q1}, [r2], r3
|
||||
@ -257,10 +258,13 @@ WELS_ASM_FUNC_BEGIN WelsSampleSad16x16_neon
|
||||
vpaddl.u16 d0, d0
|
||||
vpaddl.u32 d0, d0
|
||||
vmov.u32 r0, d0[0]
|
||||
|
||||
vpop {q4-q7}
|
||||
WELS_ASM_FUNC_END
|
||||
|
||||
|
||||
WELS_ASM_FUNC_BEGIN WelsSampleSad16x8_neon
|
||||
vpush {q4-q7}
|
||||
|
||||
vld1.64 {q0}, [r0, :128], r1
|
||||
vld1.64 {q1}, [r2], r3
|
||||
@ -295,6 +299,7 @@ WELS_ASM_FUNC_BEGIN WelsSampleSad16x8_neon
|
||||
vpaddl.u16 d0, d0
|
||||
vpaddl.u32 d0, d0
|
||||
vmov.u32 r0, d0[0]
|
||||
vpop {q4-q7}
|
||||
WELS_ASM_FUNC_END
|
||||
|
||||
|
||||
@ -735,86 +740,94 @@ WELS_ASM_FUNC_END
|
||||
|
||||
|
||||
WELS_ASM_FUNC_BEGIN WelsSampleSatd16x16_neon
|
||||
vpush {q7}
|
||||
|
||||
SATD_16x4
|
||||
vadd.u16 q15, q0, q2
|
||||
vadd.u16 q7, q0, q2
|
||||
|
||||
SATD_16x4
|
||||
vadd.u16 q15, q15, q0
|
||||
vadd.u16 q15, q15, q2
|
||||
vadd.u16 q7, q7, q0
|
||||
vadd.u16 q7, q7, q2
|
||||
|
||||
SATD_16x4
|
||||
vadd.u16 q15, q15, q0
|
||||
vadd.u16 q15, q15, q2
|
||||
vadd.u16 q7, q7, q0
|
||||
vadd.u16 q7, q7, q2
|
||||
|
||||
SATD_16x4
|
||||
vadd.u16 q15, q15, q0
|
||||
vadd.u16 q15, q15, q2
|
||||
vadd.u16 q7, q7, q0
|
||||
vadd.u16 q7, q7, q2
|
||||
|
||||
vadd.u16 d0, d30, d31
|
||||
vadd.u16 d0, d14, d15
|
||||
vpaddl.u16 d0, d0
|
||||
vpaddl.u32 d0, d0
|
||||
|
||||
vmov.32 r0, d0[0]
|
||||
vpop {q7}
|
||||
WELS_ASM_FUNC_END
|
||||
|
||||
|
||||
WELS_ASM_FUNC_BEGIN WelsSampleSatd16x8_neon
|
||||
vpush {q7}
|
||||
|
||||
SATD_16x4
|
||||
vadd.u16 q15, q0, q2
|
||||
vadd.u16 q7, q0, q2
|
||||
|
||||
SATD_16x4
|
||||
vadd.u16 q15, q15, q0
|
||||
vadd.u16 q15, q15, q2
|
||||
vadd.u16 q7, q7, q0
|
||||
vadd.u16 q7, q7, q2
|
||||
|
||||
vadd.u16 d0, d30, d31
|
||||
vadd.u16 d0, d14, d15
|
||||
vpaddl.u16 d0, d0
|
||||
vpaddl.u32 d0, d0
|
||||
|
||||
vmov.32 r0, d0[0]
|
||||
vpop {q7}
|
||||
WELS_ASM_FUNC_END
|
||||
|
||||
|
||||
WELS_ASM_FUNC_BEGIN WelsSampleSatd8x16_neon
|
||||
vpush {q7}
|
||||
|
||||
SATD_8x4
|
||||
vadd.u16 q15, q0, q1
|
||||
vadd.u16 q7, q0, q1
|
||||
|
||||
SATD_8x4
|
||||
vadd.u16 q15, q15, q0
|
||||
vadd.u16 q15, q15, q1
|
||||
vadd.u16 q7, q7, q0
|
||||
vadd.u16 q7, q7, q1
|
||||
|
||||
SATD_8x4
|
||||
vadd.u16 q15, q15, q0
|
||||
vadd.u16 q15, q15, q1
|
||||
vadd.u16 q7, q7, q0
|
||||
vadd.u16 q7, q7, q1
|
||||
|
||||
SATD_8x4
|
||||
vadd.u16 q15, q15, q0
|
||||
vadd.u16 q15, q15, q1
|
||||
vadd.u16 q7, q7, q0
|
||||
vadd.u16 q7, q7, q1
|
||||
|
||||
vadd.u16 d0, d30, d31
|
||||
vadd.u16 d0, d14, d15
|
||||
vpaddl.u16 d0, d0
|
||||
vpaddl.u32 d0, d0
|
||||
|
||||
vmov.32 r0, d0[0]
|
||||
vpop {q7}
|
||||
WELS_ASM_FUNC_END
|
||||
|
||||
|
||||
WELS_ASM_FUNC_BEGIN WelsSampleSatd8x8_neon
|
||||
vpush {q7}
|
||||
|
||||
SATD_8x4
|
||||
vadd.u16 q15, q0, q1
|
||||
vadd.u16 q7, q0, q1
|
||||
|
||||
SATD_8x4
|
||||
vadd.u16 q15, q15, q0
|
||||
vadd.u16 q15, q15, q1
|
||||
vadd.u16 q7, q7, q0
|
||||
vadd.u16 q7, q7, q1
|
||||
|
||||
vadd.u16 d0, d30, d31
|
||||
vadd.u16 d0, d14, d15
|
||||
vpaddl.u16 d0, d0
|
||||
vpaddl.u32 d0, d0
|
||||
|
||||
vmov.32 r0, d0[0]
|
||||
vpop {q7}
|
||||
WELS_ASM_FUNC_END
|
||||
|
||||
|
||||
|
||||
@ -635,8 +635,9 @@ WELS_ASM_FUNC_END
|
||||
|
||||
WELS_ASM_FUNC_BEGIN VAACalcSadSsdBgd_neon
|
||||
stmdb sp!, {r0-r12, r14}
|
||||
vpush {q4-q7}
|
||||
|
||||
ldr r4, [sp, #56] //r4 keeps the pic_stride
|
||||
ldr r4, [sp, #120] //r4 keeps the pic_stride
|
||||
|
||||
sub r5, r4, #1
|
||||
lsl r5, r5, #4 //r5 keeps the little step
|
||||
@ -645,12 +646,12 @@ WELS_ASM_FUNC_BEGIN VAACalcSadSsdBgd_neon
|
||||
sub r6, r2, r6 //r6 keeps the big step
|
||||
|
||||
|
||||
ldr r8, [sp, #64]//psad8x8
|
||||
ldr r9, [sp, #68]//psum16x16
|
||||
ldr r10, [sp, #72]//psqsum16x16
|
||||
ldr r11, [sp, #76]//psqdiff16x16
|
||||
ldr r12, [sp, #80]//p_sd8x8
|
||||
ldr r14, [sp, #84]//p_mad8x8
|
||||
ldr r8, [sp, #128]//psad8x8
|
||||
ldr r9, [sp, #132]//psum16x16
|
||||
ldr r10, [sp, #136]//psqsum16x16
|
||||
ldr r11, [sp, #140]//psqdiff16x16
|
||||
ldr r12, [sp, #144]//p_sd8x8
|
||||
ldr r14, [sp, #148]//p_mad8x8
|
||||
|
||||
vmov.i8 q12, #0
|
||||
|
||||
@ -700,11 +701,12 @@ vaa_calc_sad_ssd_bgd_width_loop:
|
||||
bne vaa_calc_sad_ssd_bgd_height_loop
|
||||
|
||||
//psadframe
|
||||
ldr r7, [sp, #60]//psadframe
|
||||
ldr r7, [sp, #124]//psadframe
|
||||
|
||||
vadd.i32 d24, d24, d25
|
||||
vst1.32 {d24[0]}, [r7]
|
||||
|
||||
vpop {q4-q7}
|
||||
ldmia sp!, {r0-r12, r14}
|
||||
|
||||
WELS_ASM_FUNC_END
|
||||
@ -912,8 +914,10 @@ WELS_ASM_FUNC_END
|
||||
|
||||
WELS_ASM_FUNC_BEGIN VAACalcSadVar_neon
|
||||
stmdb sp!, {r4-r11}
|
||||
vpush {q4}
|
||||
vpush {q6-q7}
|
||||
|
||||
ldr r4, [sp, #32] //r4 keeps the pic_stride
|
||||
ldr r4, [sp, #80] //r4 keeps the pic_stride
|
||||
|
||||
sub r5, r4, #1
|
||||
lsl r5, r5, #4 //r5 keeps the little step
|
||||
@ -921,10 +925,10 @@ WELS_ASM_FUNC_BEGIN VAACalcSadVar_neon
|
||||
lsl r6, r4, #4
|
||||
sub r6, r2, r6 //r6 keeps the big step
|
||||
|
||||
ldr r7, [sp, #36] //psadframe
|
||||
ldr r8, [sp, #40] //psad8x8
|
||||
ldr r9, [sp, #44] //psum16x16
|
||||
ldr r10, [sp, #48] //psqsum16x16
|
||||
ldr r7, [sp, #84] //psadframe
|
||||
ldr r8, [sp, #88] //psad8x8
|
||||
ldr r9, [sp, #92] //psum16x16
|
||||
ldr r10, [sp, #96] //psqsum16x16
|
||||
|
||||
vmov.i8 q12, #0
|
||||
vaa_calc_sad_var_height_loop:
|
||||
@ -961,6 +965,8 @@ bne vaa_calc_sad_var_height_loop
|
||||
vadd.i32 d24, d24, d25
|
||||
vst1.32 {d24[0]}, [r7]
|
||||
|
||||
vpop {q6-q7}
|
||||
vpop {q4}
|
||||
ldmia sp!, {r4-r11}
|
||||
WELS_ASM_FUNC_END
|
||||
|
||||
@ -1080,8 +1086,10 @@ WELS_ASM_FUNC_END
|
||||
|
||||
WELS_ASM_FUNC_BEGIN VAACalcSadSsd_neon
|
||||
stmdb sp!, {r4-r12}
|
||||
vpush {q4}
|
||||
vpush {q6-q7}
|
||||
|
||||
ldr r4, [sp, #36] //r4 keeps the pic_stride
|
||||
ldr r4, [sp, #84] //r4 keeps the pic_stride
|
||||
|
||||
sub r5, r4, #1
|
||||
lsl r5, r5, #4 //r5 keeps the little step
|
||||
@ -1089,11 +1097,11 @@ WELS_ASM_FUNC_BEGIN VAACalcSadSsd_neon
|
||||
lsl r6, r4, #4
|
||||
sub r6, r2, r6 //r6 keeps the big step
|
||||
|
||||
ldr r7, [sp, #40] //psadframe
|
||||
ldr r8, [sp, #44] //psad8x8
|
||||
ldr r9, [sp, #48] //psum16x16
|
||||
ldr r10, [sp, #52] //psqsum16x16
|
||||
ldr r11, [sp, #56] //psqdiff16x16
|
||||
ldr r7, [sp, #88] //psadframe
|
||||
ldr r8, [sp, #92] //psad8x8
|
||||
ldr r9, [sp, #96] //psum16x16
|
||||
ldr r10, [sp, #100] //psqsum16x16
|
||||
ldr r11, [sp, #104] //psqdiff16x16
|
||||
|
||||
vmov.i8 q12, #0
|
||||
vaa_calc_sad_ssd_height_loop:
|
||||
@ -1137,6 +1145,8 @@ vaa_calc_sad_ssd_width_loop:
|
||||
vadd.i32 d24, d24, d25
|
||||
vst1.32 {d24[0]}, [r7]
|
||||
|
||||
vpop {q6-q7}
|
||||
vpop {q4}
|
||||
ldmia sp!, {r4-r12}
|
||||
WELS_ASM_FUNC_END
|
||||
|
||||
|
||||
Loading…
Reference in New Issue
Block a user