| .macro push_v_regs |
| stp X8, X9, [sp, #-16]! |
| stp X10, X11, [sp, #-16]! |
| stp X12, X13, [sp, #-16]! |
| stp X14, X15, [sp, #-16]! |
| stp X16, X17, [sp, #-16]! |
| stp X18, X19, [sp, #-16]! |
| stp X20, X21, [sp, #-16]! |
| stp X22, X24, [sp, #-16]! |
| stp X29, X30, [sp, #-16]! |
| .endm |
| |
| .macro pop_v_regs |
| ldp X29, X30, [sp], #16 |
| ldp X22, X24, [sp], #16 |
| ldp X20, X21, [sp], #16 |
| ldp X18, X19, [sp], #16 |
| ldp X16, X17, [sp], #16 |
| ldp X14, X15, [sp], #16 |
| ldp X12, X13, [sp], #16 |
| ldp X10, X11, [sp], #16 |
| ldp X8, X9, [sp], #16 |
| .endm |
| |
| |
| .text |
| .p2align 2 |
| .global ixheaacd_fft32x32_ld2_armv8 |
| |
| ixheaacd_fft32x32_ld2_armv8: |
| |
| // STMFD sp!, {x4-x12,x14} |
| push_v_regs |
| stp x19, x20, [sp, #-16]! |
| |
| //DIT Radix-4 FFT First Stage |
| //First Butterfly |
| MOV x0, x2 |
| MOV x1, x3 |
| LDR w2, [x0] //x_0 = x[0 ] |
| sxtw x2, w2 |
| LDR w3, [x0, #32] //x_2 = x[8 ] |
| sxtw x3, w3 |
| LDR w4, [x0, #64] //x_4 = x[16] |
| sxtw x4, w4 |
| LDR w5, [x0, #96] //x_6 = x[24] |
| sxtw x5, w5 |
| ADD w6, w2, w4 //xh0_0 = x_0 + x_4 |
| SUB w7, w2, w4 //xl0_0 = x_0 - x_4 |
| ADD w8, w3, w5 //xh0_1 = x_2 + x_6 |
| SUB w9, w3, w5 //xl0_1 = x_2 - x_6 |
| |
| LDR w2, [x0, #4] //x_1 = x[0 +1] |
| sxtw x2, w2 |
| LDR w3, [x0, #36] //x_3 = x[8 +1] |
| sxtw x3, w3 |
| LDR w4, [x0, #68] //x_5 = x[16+1] |
| sxtw x4, w4 |
| LDR w5, [x0, #100] //x_7 = x[24+1] |
| sxtw x5, w5 |
| ADD w10, w2, w4 //xh1_0 = x_1 + x_5 |
| SUB w11, w2, w4 //xl1_0 = x_1 - x_5 |
| ADD w12, w3, w5 //xh1_1 = x_3 + x_7 |
| SUB w14, w3, w5 //xl1_1 = x_3 - x_7 |
| |
| ADD w2, w6, w8 //n00 = xh0_0 + xh0_1 |
| ADD w3, w7, w14 //n10 = xl0_0 + xl1_1 |
| SUB w4, w6, w8 //n20 = xh0_0 - xh0_1 |
| SUB w5, w7, w14 //n30 = xl0_0 - xl1_1 |
| STR w2, [x0] //x[0 ] = n00 |
| STR w3, [x0, #32] //x[8 ] = n10 |
| STR w4, [x0, #64] //x[16] = n20 |
| STR w5, [x0, #96] //x[24] = n30 |
| |
| ADD w2, w10, w12 //n01 = xh1_0 + xh1_1 |
| SUB w3, w11, w9 //n11 = xl1_0 - xl0_1 |
| SUB w4, w10, w12 //n21 = xh1_0 - xh1_1 |
| ADD w5, w11, w9 //n31 = xl1_0 + xl0_1 |
| STR w2, [x0, #4] //x[1 ] = n01 |
| STR w3, [x0, #36] //x[8+1 ] = n11 |
| STR w4, [x0, #68] //x[16+1] = n21 |
| STR w5, [x0, #100] //x[24+1] = n31 |
| |
| //Second Butterfly |
| LDR w2, [x0, #8] //x_0 = x[2 ] |
| sxtw x2, w2 |
| LDR w3, [x0, #40] //x_2 = x[10] |
| sxtw x3, w3 |
| LDR w4, [x0, #72] //x_4 = x[18] |
| sxtw x4, w4 |
| LDR w5, [x0, #104] //x_6 = x[26] |
| sxtw x5, w5 |
| ADD w6, w2, w4 //xh0_0 = x_0 + x_4 |
| SUB w7, w2, w4 //xl0_0 = x_0 - x_4 |
| ADD w8, w3, w5 //xh0_1 = x_2 + x_6 |
| SUB w9, w3, w5 //xl0_1 = x_2 - x_6 |
| |
| LDR w2, [x0, #12] //x_1 = x[2 +1] |
| sxtw x2, w2 |
| LDR w3, [x0, #44] //x_3 = x[10+1] |
| sxtw x3, w3 |
| LDR w4, [x0, #76] //x_5 = x[18+1] |
| sxtw x4, w4 |
| LDR w5, [x0, #108] //x_7 = x[26+1] |
| sxtw x5, w5 |
| ADD w10, w2, w4 //xh1_0 = x_1 + x_5 |
| SUB w11, w2, w4 //xl1_0 = x_1 - x_5 |
| ADD w12, w3, w5 //xh1_1 = x_3 + x_7 |
| SUB w14, w3, w5 //xl1_1 = x_3 - x_7 |
| |
| ADD w2, w6, w8 //n00 = xh0_0 + xh0_1 |
| ADD w3, w7, w14 //n10 = xl0_0 + xl1_1 |
| SUB w4, w6, w8 //n20 = xh0_0 - xh0_1 |
| SUB w5, w7, w14 //n30 = xl0_0 - xl1_1 |
| STR w2, [x0, #8] //x[2 ] = n00 |
| STR w3, [x0, #40] //x[10] = n10 |
| STR w4, [x0, #72] //x[18] = n20 |
| STR w5, [x0, #104] //x[26] = n30 |
| |
| ADD w2, w10, w12 //n01 = xh1_0 + xh1_1 |
| SUB w3, w11, w9 //n11 = xl1_0 - xl0_1 |
| SUB w4, w10, w12 //n21 = xh1_0 - xh1_1 |
| ADD w5, w11, w9 //n31 = xl1_0 + xl0_1 |
| STR w2, [x0, #12] //x[2 +1] = n01 |
| STR w3, [x0, #44] //x[10+1] = n11 |
| STR w4, [x0, #76] //x[18+1] = n21 |
| STR w5, [x0, #108] //x[26+1] = n31 |
| |
| //Third Butterfly |
| LDR w2, [x0, #16] //x_0 = x[4 ] |
| sxtw x2, w2 |
| LDR w3, [x0, #48] //x_2 = x[12] |
| sxtw x3, w3 |
| LDR w4, [x0, #80] //x_4 = x[20] |
| sxtw x4, w4 |
| LDR w5, [x0, #112] //x_6 = x[28] |
| sxtw x5, w5 |
| ADD w6, w2, w4 //xh0_0 = x_0 + x_4 |
| SUB w7, w2, w4 //xl0_0 = x_0 - x_4 |
| ADD w8, w3, w5 //xh0_1 = x_2 + x_6 |
| SUB w9, w3, w5 //xl0_1 = x_2 - x_6 |
| |
| LDR w2, [x0, #20] //x_1 = x[4 +1] |
| sxtw x2, w2 |
| LDR w3, [x0, #52] //x_3 = x[12+1] |
| sxtw x3, w3 |
| LDR w4, [x0, #84] //x_5 = x[20+1] |
| sxtw x4, w4 |
| LDR w5, [x0, #116] //x_7 = x[28+1] |
| sxtw x5, w5 |
| ADD w10, w2, w4 //xh1_0 = x_1 + x_5 |
| SUB w11, w2, w4 //xl1_0 = x_1 - x_5 |
| ADD w12, w3, w5 //xh1_1 = x_3 + x_7 |
| SUB w14, w3, w5 //xl1_1 = x_3 - x_7 |
| |
| ADD w2, w6, w8 //n00 = xh0_0 + xh0_1 |
| ADD w3, w7, w14 //n10 = xl0_0 + xl1_1 |
| SUB w4, w6, w8 //n20 = xh0_0 - xh0_1 |
| SUB w5, w7, w14 //n30 = xl0_0 - xl1_1 |
| STR w2, [x0, #16] //x[4 ] = n00 |
| STR w3, [x0, #48] //x[12] = n10 |
| STR w4, [x0, #80] //x[20] = n20 |
| STR w5, [x0, #112] //x[28] = n30 |
| |
| ADD w2, w10, w12 //n01 = xh1_0 + xh1_1 |
| SUB w3, w11, w9 //n11 = xl1_0 - xl0_1 |
| SUB w4, w10, w12 //n21 = xh1_0 - xh1_1 |
| ADD w5, w11, w9 //n31 = xl1_0 + xl0_1 |
| STR w2, [x0, #20] //x[4 +1] = n01 |
| STR w3, [x0, #52] //x[12+1] = n11 |
| STR w4, [x0, #84] //x[20+1] = n21 |
| STR w5, [x0, #116] //x[28+1] = n31 |
| |
| //Fourth Butterfly |
| LDR w2, [x0, #24] //x_0 = x[6 ] |
| sxtw x2, w2 |
| LDR w3, [x0, #56] //x_2 = x[14] |
| sxtw x3, w3 |
| LDR w4, [x0, #88] //x_4 = x[22] |
| sxtw x4, w4 |
| LDR w5, [x0, #120] //x_6 = x[30] |
| sxtw x5, w5 |
| ADD w6, w2, w4 //xh0_0 = x_0 + x_4 |
| SUB w7, w2, w4 //xl0_0 = x_0 - x_4 |
| ADD w8, w3, w5 //xh0_1 = x_2 + x_6 |
| SUB w9, w3, w5 //xl0_1 = x_2 - x_6 |
| |
| LDR w2, [x0, #28] //x_1 = x[6 +1] |
| sxtw x2, w2 |
| LDR w3, [x0, #60] //x_3 = x[14+1] |
| sxtw x3, w3 |
| LDR w4, [x0, #92] //x_5 = x[22+1] |
| sxtw x4, w4 |
| LDR w5, [x0, #124] //x_7 = x[30+1] |
| sxtw x5, w5 |
| ADD w10, w2, w4 //xh1_0 = x_1 + x_5 |
| SUB w11, w2, w4 //xl1_0 = x_1 - x_5 |
| ADD w12, w3, w5 //xh1_1 = x_3 + x_7 |
| SUB w14, w3, w5 //xl1_1 = x_3 - x_7 |
| |
| ADD w2, w6, w8 //n00 = xh0_0 + xh0_1 |
| ADD w3, w7, w14 //n10 = xl0_0 + xl1_1 |
| SUB w4, w6, w8 //n20 = xh0_0 - xh0_1 |
| SUB w5, w7, w14 //n30 = xl0_0 - xl1_1 |
| STR w2, [x0, #24] //x[6 ] = n00 |
| STR w3, [x0, #56] //x[14] = n10 |
| STR w4, [x0, #88] //x[22] = n20 |
| STR w5, [x0, #120] //x[30] = n30 |
| |
| ADD w2, w10, w12 //n01 = xh1_0 + xh1_1 |
| SUB w3, w11, w9 //n11 = xl1_0 - xl0_1 |
| SUB w4, w10, w12 //n21 = xh1_0 - xh1_1 |
| ADD w5, w11, w9 //n31 = xl1_0 + xl0_1 |
| STR w2, [x0, #28] //x[6 +1] = n01 |
| STR w3, [x0, #60] //x[14+1] = n11 |
| STR w4, [x0, #92] //x[22+1] = n21 |
| STR w5, [x0, #124] //x[30+1] = n31 |
| |
| |
| //DIT Radix-4 FFT Second Stage |
| //First Butterfly |
| LDR w2, [x0] //inp_0qr = x[0] |
| sxtw x2, w2 |
| LDR w3, [x0, #8] //inp_1qr = x[2] |
| sxtw x3, w3 |
| LDR w4, [x0, #16] //inp_2qr = x[4] |
| sxtw x4, w4 |
| LDR w5, [x0, #24] //inp_3qr = x[6] |
| sxtw x5, w5 |
| ADD w6, w2, w4 //sum_0qr = mul_0qr + mul_2qr |
| SUB w7, w2, w4 //sum_1qr = mul_0qr - mul_2qr |
| ADD w8, w3, w5 //sum_2qr = mul_1qr + mul_3qr |
| SUB w9, w3, w5 //sum_3qr = mul_1qr - mul_3qr |
| |
| LDR w2, [x0, #4] //inp_0qi = x[1] |
| sxtw x2, w2 |
| LDR w3, [x0, #12] //inp_1qi = x[3] |
| sxtw x3, w3 |
| LDR w4, [x0, #20] //inp_2qi = x[5] |
| sxtw x4, w4 |
| LDR w5, [x0, #28] //inp_3qi = x[7] |
| sxtw x5, w5 |
| ADD w10, w2, w4 //sum_0qi = mul_0qi + mul_2qi |
| SUB w11, w2, w4 //sum_1qi = mul_0qi - mul_2qi |
| ADD w12, w3, w5 //sum_2qi = mul_1qi + mul_3qi |
| SUB w14, w3, w5 //sum_3qi = mul_1qi - mul_3qi |
| |
| ADD w2, w6, w8 //sum_0qr + sum_2qr |
| ADD w3, w7, w14 //sum_1qr + sum_3qi |
| SUB w4, w6, w8 //sum_0qr - sum_2qr |
| SUB w5, w7, w14 //sum_1qr - sum_3qi |
| STR w2, [x1] //y[0 ] = sum_0qr + sum_2qr |
| STR w3, [x1, #32] //y[8 ] = sum_1qr + sum_3qi |
| STR w4, [x1, #64] //y[16] = sum_0qr - sum_2qr |
| STR w5, [x1, #96] //y[24] = sum_1qr - sum_3qi |
| |
| ADD w2, w10, w12 //sum_0qi + sum_2qi |
| SUB w3, w11, w9 //sum_1qi - sum_3qr |
| SUB w4, w10, w12 //sum_0qi - sum_2qi |
| ADD w5, w11, w9 //sum_1qi + sum_3qr |
| STR w2, [x1, #4] //y[0 +1] = sum_0qi + sum_2qi |
| STR w3, [x1, #36] //y[8 +1] = sum_1qi - sum_3qr |
| STR w4, [x1, #68] //y[16+1] = sum_0qi - sum_2qi |
| STR w5, [x1, #100] //y[24+1] = sum_1qi + sum_3qr |
| |
| |
| //Load twiddle factors |
| // LDR w11, =2310960706 //0x89BE7642 |
| LDR w11, =0x7642 |
| sxth w11, w11 |
| LDR w21, =0x89BE |
| sxth w21, w21 |
| // LDR w12, =3473158396 //0xCF0430FC |
| LDR w12, =0x30FC |
| sxth w12, w12 |
| LDR w22, =0xCF04 |
| sxth w22, w22 |
| // LDR w14, =2776455811 //0xA57D5A83 |
| LDR w14, =0x5A83 |
| sxth w14, w14 |
| LDR w24, =0xA57D |
| sxth w24, w24 |
| |
| //Second Butterfly |
| LDR w2, [x0, #32] //mul_0qr = inp_0qr = x[8] |
| sxtw x2, w2 |
| LDR w3, [x0, #36] //mul_0qi = inp_1qr = x[9] |
| sxtw x3, w3 |
| |
| LDR w5, [x0, #40] //inp_1qr = x[10] |
| sxtw x5, w5 |
| LDR w6, [x0, #44] //inp_1qi = x[11] |
| sxtw x6, w6 |
| |
| SMULL x4, w5, w11 |
| ASR x4, x4, #16 |
| // SMULWB x4, x5, x11 //mul_1qr = mpy_16_32_ns( 0x7642 , inp_1qr) |
| |
| SMULL x20, w6, w12 |
| ASR x20, x20, #16 |
| ADD w4, w4, w20 |
| // SMLAWB x4, x6, x12, x4 //mul_1qr -= mpy_16_32_ns(-0x30FC , inp_1qi) |
| |
| SMULL x5, w5, w22 |
| ASR x5, x5, #16 |
| // SMULWT x5, x5, x12 //mul_1qi = mpy_16_32_ns(-0x30FC , inp_1qr) |
| |
| LDR w7, [x0, #48] //inp_2qr = x[12] |
| sxtw x7, w7 |
| LDR w8, [x0, #52] //inp_2qi = x[13] |
| sxtw x8, w8 |
| |
| //Moved for delay slot |
| SMULL x20, w6, w11 |
| ASR x20, x20, #16 |
| ADD w5, w5, w20 |
| // SMLAWB x5, x6, x11, x5 //mul_1qi += mpy_16_32_ns( 0x7642 , inp_1qi) |
| |
| ADD w6, w7, w8 //(inp_2qr + inp_2qi) |
| |
| SMULL x6, w6, w14 |
| ASR x6, x6, #16 |
| // SMULWB x6, x6, x14 //mul_2qr = mpy_16_32_ns(0x5A83 , (inp_2qr + inp_2qi)) |
| |
| SUB w7, w8, w7 //(-inp_2qr + inp_2qi) |
| |
| SMULL x7, w7, w14 |
| ASR x7, x7, #16 |
| // SMULWB x7, x7, x14 //mul_2qi = mpy_16_32_ns(0x5A83 , (-inp_2qr + inp_2qi)) |
| |
| LDR x9 , [x0, #56] //inp_3qr = x[14] |
| sxtw x9, w9 |
| LDR w10, [x0, #60] //inp_3qi = x[15] |
| sxtw x10, w10 |
| |
| SMULL x8, w9, w12 |
| ASR x8, x8, #16 |
| // SMULWB x8, x9 , x12 //mul_3qr = mpy_16_32_ns( 0x30FC , inp_3qr) |
| |
| SMULL x20, w10, w11 |
| ASR x20, x20, #16 |
| ADD w8, w8, w20 |
| // SMLAWB x8, x10, x11, x8 //mul_3qr -= mpy_16_32_ns(-0x7642 , inp_3qi)// |
| |
| SMULL x9, w9 , w21 |
| ASR x9, x9, #16 |
| // SMULWT x9, x9 , x11 //mul_3qi = mpy_16_32_ns(-0x7642 , inp_3qr) |
| |
| SMULL x20, w10, w12 |
| ASR x20, x20, #16 |
| ADD w9, w9, w20 |
| // SMLAWB x9, x10, x12, x9 //mul_3qi += mpy_16_32_ns( 0x30FC , inp_3qi) |
| |
| ADD w10, w2, w6, lsl #1 //sum_0qr = mul_0qr + (mul_2qr << 1) |
| SUB w2 , w2, w6, lsl #1 //sum_1qr = mul_0qr - (mul_2qr << 1) |
| ADD w6 , w4, w8 //sum_2qr = mul_1qr + mul_3qr |
| SUB w4 , w4, w8 //sum_3qr = mul_1qr - mul_3qr |
| |
| ADD w8 , w3, w7, lsl #1 //sum_0qi = mul_0qi + (mul_2qi << 1) |
| SUB w3 , w3, w7, lsl #1 //sum_1qi = mul_0qi - (mul_2qi << 1) |
| ADD w7 , w5, w9 //sum_2qi = mul_1qi + mul_3qi |
| SUB w5 , w5, w9 //sum_3qi = mul_1qi - mul_3qi |
| |
| ADD w9 , w10, w6, lsl #1 //sum_0qr + (sum_2qr << 1) |
| SUB w10, w10, w6, lsl #1 //sum_0qr - (sum_2qr << 1) |
| ADD w6 , w2 , w5, lsl #1 //sum_1qr + (sum_3qi << 1) |
| SUB w2 , w2 , w5, lsl #1 //sum_1qr - (sum_3qi << 1) |
| STR w9 , [x1, #8] //y[2 ] = sum_0qr + (sum_2qr << 1) |
| STR w10, [x1, #72] //y[18] = sum_0qr - (sum_2qr << 1) |
| STR w6 , [x1, #40] //y[10] = sum_1qr + (sum_3qi << 1) |
| STR w2 , [x1, #104] //y[26] = sum_1qr - (sum_3qi << 1) |
| |
| ADD w5 , w8 , w7, lsl #1 //sum_0qi + (sum_2qi << 1) |
| SUB w8 , w8 , w7, lsl #1 //sum_0qi - (sum_2qi << 1) |
| SUB w7 , w3 , w4, lsl #1 //sum_1qi - (sum_3qr << 1) |
| ADD w3 , w3 , w4, lsl #1 //sum_1qi + (sum_3qr << 1) |
| STR w5 , [x1, #12] //y[2 +1] = sum_0qi + (sum_2qi << 1) |
| STR w8 , [x1, #76] //y[18+1] = sum_0qi - (sum_2qi << 1) |
| STR w7 , [x1, #44] //y[10+1] = sum_1qi - (sum_3qr << 1) |
| STR w3 , [x1, #108] //y[26+1] = sum_1qi + (sum_3qr << 1) |
| |
| //Third Butterfly |
| LDR w2, [x0, #64] //mul_0qr = inp_0qr = x[16] |
| sxtw x2, w2 |
| LDR w5, [x0, #72] //inp_1qr = x[18] |
| sxtw x5, w5 |
| LDR w6, [x0, #76] //inp_1qi = x[19] |
| sxtw x6, w6 |
| //Moved for delay slot |
| LDR w3, [x0, #68] //mul_0qi = inp_1qr = x[17] |
| sxtw x3, w3 |
| |
| ADD w4, w5, w6 //(inp_1qr + inp_1qi) |
| |
| SMULL x4, w4, w14 |
| ASR x4, x4, #16 |
| // SMULWB x4, x4, x14 //mul_1qr = mpy_16_32_ns(0x5A83 , (inp_1qr + inp_1qi)) |
| SUB w5, w6, w5 //(-inp_1qr + inp_1qi) |
| |
| SMULL x5, w5, w14 |
| ASR x5, x5, #16 |
| // SMULWB x5, x5, x14 //mul_1qi = mpy_16_32_ns(0x5A83 , (-inp_1qr + inp_1qi)) |
| |
| LDR w6, [x0, #84] //mul_2qr = inp_2qi = x[21] |
| sxtw x6, w6 |
| LDR x9 , [x0, #88] //inp_3qr = x[22] |
| sxtw x9, w9 |
| LDR w10, [x0, #92] //inp_3qi = x[23] |
| sxtw x10, w10 |
| //Moved for delay slot |
| LDR w7, [x0, #80] //mul_2qi = inp_2qr = x[20] |
| sxtw x7, w7 |
| |
| SUB w8 , w10, w9 //(-inp_3qr + inp_3qi) |
| |
| SMULL x8, w8, w14 |
| ASR x8, x8, #16 |
| // SMULWB x8 , x8 , x14 //mul_3qr = mpy_16_32_ns( 0x5A83 , (-inp_3qr + inp_3qi)) |
| |
| ADD w9 , w9 , w10 //(inp_3qr + inp_3qi) |
| |
| SMULL x9, w9, w24 |
| ASR x9, x9, #16 |
| // SMULWT x9 , x9 , x14 //mul_3qi = mpy_16_32_ns(-0x5A83 , (inp_3qr + inp_3qi)) |
| |
| ADD w10, w2, w6 //sum_0qr = mul_0qr + mul_2qr |
| SUB w2 , w2, w6 //sum_1qr = mul_0qr - mul_2qr |
| ADD w6 , w4, w8 //sum_2qr = mul_1qr + mul_3qr |
| SUB w4 , w4, w8 //sum_3qr = mul_1qr - mul_3qr |
| |
| SUB w8 , w3, w7 //sum_0qi = mul_0qi - mul_2qi |
| ADD w3 , w3, w7 //sum_1qi = mul_0qi + mul_2qi |
| ADD w7 , w5, w9 //sum_2qi = mul_1qi + mul_3qi |
| SUB w5 , w5, w9 //sum_3qi = mul_1qi - mul_3qi |
| |
| ADD w9 , w10, w6, lsl #1 //sum_0qr + (sum_2qr << 1) |
| SUB w10, w10, w6, lsl #1 //sum_0qr - (sum_2qr << 1) |
| ADD w6 , w2 , w5, lsl #1 //sum_1qr + (sum_3qi << 1) |
| SUB w2 , w2 , w5, lsl #1 //sum_1qr - (sum_3qi << 1) |
| STR w9 , [x1, #16] //y[4 ] = sum_0qr + (sum_2qr << 1) |
| STR w10, [x1, #80] //y[20] = sum_0qr - (sum_2qr << 1) |
| STR w6 , [x1, #48] //y[12] = sum_1qr + (sum_3qi << 1) |
| STR w2 , [x1, #112] //y[28] = sum_1qr - (sum_3qi << 1) |
| |
| ADD w5, w8, w7, lsl #1 //sum_0qi + (sum_2qi << 1) |
| SUB w8, w8, w7, lsl #1 //sum_0qi - (sum_2qi << 1) |
| SUB w7, w3, w4, lsl #1 //sum_1qi - (sum_3qr << 1) |
| ADD w3, w3, w4, lsl #1 //sum_1qi + (sum_3qr << 1) |
| STR w5 , [x1, #20] //y[4 +1] = sum_0qi + (sum_2qi << 1) |
| STR w8 , [x1, #84] //y[20+1] = sum_0qi - (sum_2qi << 1) |
| STR w7 , [x1, #52] //y[12+1] = sum_1qi - (sum_3qr << 1) |
| STR w3 , [x1, #116] //y[28+1] = sum_1qi + (sum_3qr << 1) |
| |
| //Fourth Butterfly |
| LDR w2, [x0, #96] //mul_0qr = inp_0qr = x[24] |
| sxtw x2, w2 |
| LDR w3, [x0, #100] //mul_0qi = inp_1qr = x[25] |
| sxtw x3, w3 |
| |
| LDR w5, [x0, #104] //inp_1qr = x[26] |
| sxtw x5, w5 |
| LDR w6, [x0, #108] //inp_1qi = x[27] |
| sxtw x6, w6 |
| |
| SMULL x4, w5, w12 |
| ASR x4, x4, #16 |
| // SMULWB x4, x5, x12 //mul_1qr = mpy_16_32_ns( 0x30FC , inp_1qr) |
| |
| SMULL x20, w6, w11 |
| ASR x20, x20, #16 |
| ADD w4, w4, w20 |
| // SMLAWB x4, x6, x11, x4 //mul_1qr -= mpy_16_32_ns(-0x7642 , inp_1qi) |
| |
| SMULL x5, w5, w21 |
| ASR x5, x5, #16 |
| // SMULWT x5, x5, x11 //mul_1qi = mpy_16_32_ns(-0x7642 , inp_1qr) |
| |
| LDR w7, [x0, #112] //inp_2qr = x[28] |
| sxtw x7, w7 |
| LDR w8, [x0, #116] //inp_2qi = x[29] |
| sxtw x8, w8 |
| |
| //Moved for delay slot |
| SMULL x20, w6, w12 |
| ASR x20, x20, #16 |
| ADD w5, w5, w20 |
| // SMLAWB x5, x6, x12, x5 //mul_1qi += mpy_16_32_ns( 0x30FC , inp_1qi) |
| |
| SUB w6, w8, w7 //(-inp_2qr + inp_2qi) |
| |
| SMULL x6, w6, w14 |
| ASR x6, x6, #16 |
| // SMULWB x6, x6, x14 //mul_2qr = mpy_16_32_ns( 0x5A83 , (-inp_2qr + inp_2qi)) |
| ADD w7, w8, w7 //(inp_2qr + inp_2qi) |
| |
| SMULL x7, w7, w24 |
| ASR x7, x7, #16 |
| // SMULWT x7, x7, x14 //mul_2qi = mpy_16_32_ns(-0x5A83 , (inp_2qr + inp_2qi)) |
| |
| LDR w9 , [x0, #120] //inp_3qr = x[30] |
| sxtw x9, w9 |
| LDR w10, [x0, #124] //inp_3qi = x[31] |
| sxtw x10, w10 |
| |
| SMULL x8, w9, w21 |
| ASR x8, x8, #16 |
| // SMULWT x8, x9 , x11 //mul_3qr = mpy_16_32_ns(-0x7642 , inp_3qr) |
| |
| SMULL x20, w10, w22 |
| ASR x20, x20, #16 |
| ADD w8, w8, w20 |
| // SMLAWT x8, x10, x12, x8 //mul_3qr -= mpy_16_32_ns( 0x30FC , inp_3qi)// |
| |
| SMULL x9, w9, w12 |
| ASR x9, x9, #16 |
| // SMULWB x9, x9 , x12 //mul_3qi = mpy_16_32_ns( 0x30FC , inp_3qr) |
| |
| SMULL x20, w10, w21 |
| ASR x20, x20, #16 |
| ADD w9, w9, w20 |
| // SMLAWT x9, x10, x11, x9 //mul_3qi += mpy_16_32_ns(-0x7642 , inp_3qi) |
| |
| ADD w10, w2, w6, lsl #1 //sum_0qr = mul_0qr + (mul_2qr << 1) |
| SUB w2 , w2, w6, lsl #1 //sum_1qr = mul_0qr - (mul_2qr << 1) |
| ADD w6 , w4, w8 //sum_2qr = mul_1qr + mul_3qr |
| SUB w4 , w4, w8 //sum_3qr = mul_1qr - mul_3qr |
| |
| ADD w8 , w3, w7, lsl #1 //sum_0qi = mul_0qi + (mul_2qi << 1) |
| SUB w3 , w3, w7, lsl #1 //sum_1qi = mul_0qi - (mul_2qi << 1) |
| ADD w7 , w5, w9 //sum_2qi = mul_1qi + mul_3qi |
| SUB w5 , w5, w9 //sum_3qi = mul_1qi - mul_3qi |
| |
| ADD w9 , w10, w6, lsl #1 //sum_0qr + (sum_2qr << 1) |
| SUB w10, w10, w6, lsl #1 //sum_0qr - (sum_2qr << 1) |
| ADD w6 , w2 , w5, lsl #1 //sum_1qr + (sum_3qi << 1) |
| SUB w2 , w2 , w5, lsl #1 //sum_1qr - (sum_3qi << 1) |
| STR w9 , [x1, #24] //y[6 ] = sum_0qr + (sum_2qr << 1) |
| STR w10, [x1, #88] //y[22] = sum_0qr - (sum_2qr << 1) |
| STR w6 , [x1, #56] //y[14] = sum_1qr + (sum_3qi << 1) |
| STR w2 , [x1, #120] //y[30] = sum_1qr - (sum_3qi << 1) |
| |
| ADD w5 , w8 , w7, lsl #1 //sum_0qi + (sum_2qi << 1) |
| SUB w8 , w8 , w7, lsl #1 //sum_0qi - (sum_2qi << 1) |
| SUB w7 , w3 , w4, lsl #1 //sum_1qi - (sum_3qr << 1) |
| ADD w3 , w3 , w4, lsl #1 //sum_1qi + (sum_3qr << 1) |
| STR w5 , [x1, #28] //y[6 +1] = sum_0qi + (sum_2qi << 1) |
| STR w8 , [x1, #92] //y[22+1] = sum_0qi - (sum_2qi << 1) |
| STR w7 , [x1, #60] //y[14+1] = sum_1qi - (sum_3qr << 1) |
| STR w3 , [x1, #124] //y[30+1] = sum_1qi + (sum_3qr << 1) |
| |
| // LDMFD sp!, {x4-x12,x15} |
| ldp x19, x20, [sp], #16 |
| pop_v_regs |
| ret |
| |
