|
// Auto-generated file. Do not edit! |
|
// Template: src/f32-gemm/1x8-aarch32-neon-cortex-a53.S.in |
|
// Generator: tools/xngen |
|
// |
|
// Copyright 2023 Google LLC |
|
// |
|
// This source code is licensed under the BSD-style license found in the |
|
// LICENSE file in the root directory of this source tree. |
|
|
|
|
|
|
|
.syntax unified |
|
|
|
// void xnn_f32_gemm_minmax_ukernel_1x8__asm_aarch32_neon_cortex_a53_prfm( |
|
// size_t mr, r0 |
|
// size_t nc, r1 |
|
// size_t kc, r2 -> r0 |
|
// const float* a, r3 |
|
// size_t a_stride, sp + 8 -> (unused) |
|
// const float* w, sp + 12 -> r9 |
|
// float* c, sp + 16 -> r12 |
|
// size_t cm_stride, sp + 20 -> (unused) |
|
// size_t cn_stride, sp + 24 -> r7 |
|
// xnn_f32_minmax_params params) sp + 28 -> (r0) |
|
|
|
// d8-d31, r4-r11,r14(lr) need to be preserved if used. r13(sp),r15(pc) are reserved. |
|
|
|
// Register usage |
|
// A0 r3 d0 |
|
// B r9 d24, d25, d26, d27 |
|
// B d28, d29, d30, d31 |
|
// C0 r12 d16-d17 q8 d18-d19 q9 q10 q11 |
|
// clamp (r0) d4 d5 d6 d7 |
|
|
|
BEGIN_FUNCTION xnn_f32_gemm_minmax_ukernel_1x8__asm_aarch32_neon_cortex_a53_prfm |
|
.arm |
|
|
|
.arch armv7-a |
|
.fpu neon |
|
|
|
|
|
PUSH {r7, r9} // 8 |
|
|
|
LDR r0, [sp, 28] // params |
|
LDR r9, [sp, 12] // w |
|
LDR r12, [sp, 16] // c |
|
|
|
|
|
VLD1.32 {d4[], d5[]}, [r0]! |
|
LDR r7, [sp, 24] // cn_stride |
|
VLD1.32 {d6[], d7[]}, [r0] |
|
|
|
0: |
|
|
|
VLDM r9!, {d16-d19} // Bias |
|
VMOV.I32 q10, 0 // second set of C for pipelining VMLA |
|
SUBS r0, r2, 8 |
|
VMOV.I32 q11, 0 |
|
|
|
PLD [r3, 0] // Prefetch A |
|
PLD [r3, 64] |
|
PLD [r9, 0] // Prefetch B |
|
PLD [r9, 64] |
|
PLD [r9, 128] |
|
PLD [r9, 192] |
|
PLD [r9, 256] |
|
PLD [r9, 320] |
|
PLD [r9, 384] |
|
PLD [r9, 448] |
|
PLD [r9, 512] |
|
PLD [r9, 576] |
|
BLO 3f // less than 2 channels? |
|
|
|
|
|
1: |
|
VLDM r9!, {d24-d27} // B0 |
|
VLD1.32 {d0}, [r3]! // A0 |
|
VLDM r9!, {d28-d31} // B1 |
|
VMLA.F32 q8, q12, d0[0] |
|
VMLA.F32 q9, q13, d0[0] |
|
PLD [r9, 576] // Prefetch B |
|
VMLA.F32 q10, q14, d0[1] |
|
VMLA.F32 q11, q15, d0[1] |
|
SUBS r0, r0, 8 |
|
PLD [r3, 128] // Prefetch A0 |
|
BHS 1b |
|
|
|
|
|
TST r0, 4 |
|
BNE 3f |
|
|
|
2: |
|
VADD.F32 q8, q8, q10 |
|
VADD.F32 q9, q9, q11 |
|
|
|
|
|
VMAX.F32 q8, q8, q2 |
|
SUBS r1, r1, 8 |
|
VMAX.F32 q9, q9, q2 |
|
VMIN.F32 q8, q8, q3 |
|
VMIN.F32 q9, q9, q3 |
|
|
|
|
|
BLO 4f |
|
VST1.32 {d16-d19}, [r12], r7 |
|
SUB r3, r3, r2 |
|
BHI 0b |
|
|
|
POP {r7, r9} |
|
BX lr |
|
|
|
3: |
|
|
|
VLDM r3!, {s0} // A0 |
|
VLDM r9!, {d24-d27} // B0 |
|
VMLA.F32 q8, q12, d0[0] |
|
VMLA.F32 q9, q13, d0[0] |
|
B 2b |
|
|
|
|
|
4: |
|
TST r1, 4 |
|
BEQ 5f |
|
VST1.32 {d16-d17}, [r12]! |
|
VMOV q8, q9 |
|
|
|
5: |
|
TST r1, 2 |
|
BEQ 6f |
|
VST1.32 {d16}, [r12]! |
|
VMOV d16, d17 |
|
|
|
6: |
|
TST r1, 1 |
|
BEQ 7f |
|
VST1.32 {d16[0]}, [r12] |
|
|
|
7: |
|
POP {r7, r9} |
|
BX lr |
|
|
|
END_FUNCTION xnn_f32_gemm_minmax_ukernel_1x8__asm_aarch32_neon_cortex_a53_prfm |
|
|
|
|
|
.section ".note.GNU-stack","",%progbits |
|
|
|
|