// Auto-generated file. Do not edit! // Template: src/f32-dwconv2d-chw/3x3s2p1-neon.c.in // Generator: tools/xngen // // Copyright 2020 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. #include #include #include #include void xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__neonfma_4x4( size_t input_height, size_t input_width, const float* input, const float* weights, const float* zero, float* output, uint32_t padding_top, const union xnn_f32_chw_params params[restrict XNN_MIN_ELEMENTS(1)]) { assert(input_height != 0); assert(input_width != 0); assert(input_width % sizeof(float) == 0); assert(padding_top >= 0); assert(padding_top <= 1); const uint32x4_t vmask_even = vld1q_u32(params->neon.mask_even); const uint32x4_t vmask_odd = vld1q_u32(params->neon.mask_odd); const float32x4_t vmax = vld1q_dup_f32(¶ms->neon.max); const float32x4_t vmin = vld1q_dup_f32(¶ms->neon.min); const float32x4_t vw0123 = vld1q_f32(weights); const float32x4_t vw4567 = vld1q_f32(weights + 4); const float32x2_t vw89 = vld1_f32(weights + 8); const size_t input_decrement = round_down_po2(input_width, 4 /* SIMD output width */ * 2 /* subsampling */ * sizeof(float)); const size_t output_width = round_down_po2((input_width + (2 /* padding */ - 3 /* kernel size */ + 2 /* subsampling */) * sizeof(float)) / 2, sizeof(float)); const float* i0 = (const float*) ((uintptr_t) input - ((-padding_top) & input_width)); const float* i1 = (const float*) ((uintptr_t) i0 + input_width); if XNN_UNPREDICTABLE(padding_top != 0) { i0 = zero; } const float* i2 = (const float*) ((uintptr_t) i1 + input_width); const float* i3 = (const float*) ((uintptr_t) i2 + input_width); const float* i4 = (const float*) ((uintptr_t) i3 + input_width); const float* i5 = (const float*) ((uintptr_t) i4 + input_width); const float* i6 = (const float*) ((uintptr_t) i5 + input_width); const float* i7 = (const float*) ((uintptr_t) i6 + input_width); const float* i8 = (const float*) ((uintptr_t) i7 + input_width); float* o0 = output; float* o1 = (float*) ((uintptr_t) o0 + output_width); float* o2 = (float*) ((uintptr_t) o1 + output_width); float* o3 = (float*) ((uintptr_t) o2 + output_width); size_t padded_input_height = input_height + padding_top + 1 /* padding bottom */; size_t output_height = (padded_input_height - 3 /* kernel size */ + 2 /* subsampling */) / 2; do { if XNN_UNPREDICTABLE(padded_input_height < 4) { i2 = zero; } if XNN_UNPREDICTABLE(padded_input_height < 5) { i3 = zero; o1 = o0; } if XNN_UNPREDICTABLE(padded_input_height < 6) { i4 = zero; } if XNN_UNPREDICTABLE(padded_input_height < 7) { i5 = zero; o2 = o1; } if XNN_UNPREDICTABLE(padded_input_height < 8) { i6 = zero; } if XNN_UNPREDICTABLE(padded_input_height < 9) { i7 = zero; o3 = o2; } if XNN_UNPREDICTABLE(padded_input_height < 10) { i8 = zero; } float32x4_t vi0x1357 = vmovq_n_f32(0.0f); float32x4_t vi1x1357 = vmovq_n_f32(0.0f); float32x4_t vi2x1357 = vmovq_n_f32(0.0f); float32x4_t vi3x1357 = vmovq_n_f32(0.0f); float32x4_t vi4x1357 = vmovq_n_f32(0.0f); float32x4_t vi5x1357 = vmovq_n_f32(0.0f); float32x4_t vi6x1357 = vmovq_n_f32(0.0f); float32x4_t vi7x1357 = vmovq_n_f32(0.0f); float32x4_t vi8x1357 = vmovq_n_f32(0.0f); size_t w = input_width; for (; w >= 8 * sizeof(float); w -= 8 * sizeof(float)) { float32x4_t vo0p0 = vdupq_lane_f32(vget_low_f32(vw0123), 0); float32x4_t vo1p0 = vdupq_lane_f32(vget_low_f32(vw0123), 0); float32x4_t vo2p0 = vdupq_lane_f32(vget_low_f32(vw0123), 0); float32x4_t vo3p0 = vdupq_lane_f32(vget_low_f32(vw0123), 0); const float32x4x2_t vi0x8ACE9BDF = vld2q_f32(i0); i0 += 8; const float32x4x2_t vi1x8ACE9BDF = vld2q_f32(i1); i1 += 8; const float32x4x2_t vi2x8ACE9BDF = vld2q_f32(i2); i2 += 8; const float32x4x2_t vi3x8ACE9BDF = vld2q_f32(i3); i3 += 8; const float32x4x2_t vi4x8ACE9BDF = vld2q_f32(i4); i4 += 8; const float32x4x2_t vi5x8ACE9BDF = vld2q_f32(i5); i5 += 8; const float32x4x2_t vi6x8ACE9BDF = vld2q_f32(i6); i6 += 8; const float32x4x2_t vi7x8ACE9BDF = vld2q_f32(i7); i7 += 8; const float32x4x2_t vi8x8ACE9BDF = vld2q_f32(i8); i8 += 8; vo0p0 = vfmaq_lane_f32(vo0p0, vi0x8ACE9BDF.val[0], vget_high_f32(vw0123), 0); vo1p0 = vfmaq_lane_f32(vo1p0, vi2x8ACE9BDF.val[0], vget_high_f32(vw0123), 0); vo2p0 = vfmaq_lane_f32(vo2p0, vi4x8ACE9BDF.val[0], vget_high_f32(vw0123), 0); vo3p0 = vfmaq_lane_f32(vo3p0, vi6x8ACE9BDF.val[0], vget_high_f32(vw0123), 0); vo0p0 = vfmaq_lane_f32(vo0p0, vi1x8ACE9BDF.val[0], vget_low_f32(vw4567), 1); vo1p0 = vfmaq_lane_f32(vo1p0, vi3x8ACE9BDF.val[0], vget_low_f32(vw4567), 1); vo2p0 = vfmaq_lane_f32(vo2p0, vi5x8ACE9BDF.val[0], vget_low_f32(vw4567), 1); vo3p0 = vfmaq_lane_f32(vo3p0, vi7x8ACE9BDF.val[0], vget_low_f32(vw4567), 1); vo0p0 = vfmaq_lane_f32(vo0p0, vi2x8ACE9BDF.val[0], vw89, 0); vo1p0 = vfmaq_lane_f32(vo1p0, vi4x8ACE9BDF.val[0], vw89, 0); vo2p0 = vfmaq_lane_f32(vo2p0, vi6x8ACE9BDF.val[0], vw89, 0); vo3p0 = vfmaq_lane_f32(vo3p0, vi8x8ACE9BDF.val[0], vw89, 0); const float32x4_t vi0x7BDF = vextq_f32(vi0x1357, vi0x8ACE9BDF.val[1], 3); vi0x1357 = vi0x8ACE9BDF.val[1]; const float32x4_t vi1x7BDF = vextq_f32(vi1x1357, vi1x8ACE9BDF.val[1], 3); vi1x1357 = vi1x8ACE9BDF.val[1]; const float32x4_t vi2x7BDF = vextq_f32(vi2x1357, vi2x8ACE9BDF.val[1], 3); vi2x1357 = vi2x8ACE9BDF.val[1]; const float32x4_t vi3x7BDF = vextq_f32(vi3x1357, vi3x8ACE9BDF.val[1], 3); vi3x1357 = vi3x8ACE9BDF.val[1]; const float32x4_t vi4x7BDF = vextq_f32(vi4x1357, vi4x8ACE9BDF.val[1], 3); vi4x1357 = vi4x8ACE9BDF.val[1]; const float32x4_t vi5x7BDF = vextq_f32(vi5x1357, vi5x8ACE9BDF.val[1], 3); vi5x1357 = vi5x8ACE9BDF.val[1]; const float32x4_t vi6x7BDF = vextq_f32(vi6x1357, vi6x8ACE9BDF.val[1], 3); vi6x1357 = vi6x8ACE9BDF.val[1]; const float32x4_t vi7x7BDF = vextq_f32(vi7x1357, vi7x8ACE9BDF.val[1], 3); vi7x1357 = vi7x8ACE9BDF.val[1]; const float32x4_t vi8x7BDF = vextq_f32(vi8x1357, vi8x8ACE9BDF.val[1], 3); vi8x1357 = vi8x8ACE9BDF.val[1]; vo0p0 = vfmaq_lane_f32(vo0p0, vi0x7BDF, vget_low_f32(vw0123), 1); vo1p0 = vfmaq_lane_f32(vo1p0, vi2x7BDF, vget_low_f32(vw0123), 1); vo2p0 = vfmaq_lane_f32(vo2p0, vi4x7BDF, vget_low_f32(vw0123), 1); vo3p0 = vfmaq_lane_f32(vo3p0, vi6x7BDF, vget_low_f32(vw0123), 1); vo0p0 = vfmaq_lane_f32(vo0p0, vi1x7BDF, vget_low_f32(vw4567), 0); vo1p0 = vfmaq_lane_f32(vo1p0, vi3x7BDF, vget_low_f32(vw4567), 0); vo2p0 = vfmaq_lane_f32(vo2p0, vi5x7BDF, vget_low_f32(vw4567), 0); vo3p0 = vfmaq_lane_f32(vo3p0, vi7x7BDF, vget_low_f32(vw4567), 0); vo0p0 = vfmaq_lane_f32(vo0p0, vi2x7BDF, vget_high_f32(vw4567), 1); vo1p0 = vfmaq_lane_f32(vo1p0, vi4x7BDF, vget_high_f32(vw4567), 1); vo2p0 = vfmaq_lane_f32(vo2p0, vi6x7BDF, vget_high_f32(vw4567), 1); vo3p0 = vfmaq_lane_f32(vo3p0, vi8x7BDF, vget_high_f32(vw4567), 1); vo0p0 = vfmaq_lane_f32(vo0p0, vi0x8ACE9BDF.val[1], vget_high_f32(vw0123), 1); vo1p0 = vfmaq_lane_f32(vo1p0, vi2x8ACE9BDF.val[1], vget_high_f32(vw0123), 1); vo2p0 = vfmaq_lane_f32(vo2p0, vi4x8ACE9BDF.val[1], vget_high_f32(vw0123), 1); vo3p0 = vfmaq_lane_f32(vo3p0, vi6x8ACE9BDF.val[1], vget_high_f32(vw0123), 1); vo0p0 = vfmaq_lane_f32(vo0p0, vi1x8ACE9BDF.val[1], vget_high_f32(vw4567), 0); vo1p0 = vfmaq_lane_f32(vo1p0, vi3x8ACE9BDF.val[1], vget_high_f32(vw4567), 0); vo2p0 = vfmaq_lane_f32(vo2p0, vi5x8ACE9BDF.val[1], vget_high_f32(vw4567), 0); vo3p0 = vfmaq_lane_f32(vo3p0, vi7x8ACE9BDF.val[1], vget_high_f32(vw4567), 0); vo0p0 = vfmaq_lane_f32(vo0p0, vi2x8ACE9BDF.val[1], vw89, 1); vo1p0 = vfmaq_lane_f32(vo1p0, vi4x8ACE9BDF.val[1], vw89, 1); vo2p0 = vfmaq_lane_f32(vo2p0, vi6x8ACE9BDF.val[1], vw89, 1); vo3p0 = vfmaq_lane_f32(vo3p0, vi8x8ACE9BDF.val[1], vw89, 1); float32x4_t vo0 = vmaxq_f32(vo0p0, vmin); float32x4_t vo1 = vmaxq_f32(vo1p0, vmin); float32x4_t vo2 = vmaxq_f32(vo2p0, vmin); float32x4_t vo3 = vmaxq_f32(vo3p0, vmin); vo0 = vminq_f32(vo0, vmax); vo1 = vminq_f32(vo1, vmax); vo2 = vminq_f32(vo2, vmax); vo3 = vminq_f32(vo3, vmax); vst1q_f32(o3, vo3); o3 += 4; vst1q_f32(o2, vo2); o2 += 4; vst1q_f32(o1, vo1); o1 += 4; vst1q_f32(o0, vo0); o0 += 4; } // Last block has 0-7 pixels to process. assert(w < 8 * sizeof(float)); if XNN_LIKELY(w != 0) { float32x4_t vo0p0 = vdupq_lane_f32(vget_low_f32(vw0123), 0); float32x4_t vo1p0 = vdupq_lane_f32(vget_low_f32(vw0123), 0); float32x4_t vo2p0 = vdupq_lane_f32(vget_low_f32(vw0123), 0); float32x4_t vo3p0 = vdupq_lane_f32(vget_low_f32(vw0123), 0); const float32x4x2_t vi0x8ACE9BDF = vld2q_f32(i0); const float32x4x2_t vi1x8ACE9BDF = vld2q_f32(i1); const float32x4x2_t vi2x8ACE9BDF = vld2q_f32(i2); const float32x4x2_t vi3x8ACE9BDF = vld2q_f32(i3); const float32x4x2_t vi4x8ACE9BDF = vld2q_f32(i4); const float32x4x2_t vi5x8ACE9BDF = vld2q_f32(i5); const float32x4x2_t vi6x8ACE9BDF = vld2q_f32(i6); const float32x4x2_t vi7x8ACE9BDF = vld2q_f32(i7); const float32x4x2_t vi8x8ACE9BDF = vld2q_f32(i8); const float32x4_t vi0x8ACE = vreinterpretq_f32_u32(vandq_u32(vmask_even, vreinterpretq_u32_f32(vi0x8ACE9BDF.val[0]))); const float32x4_t vi0x9BDF = vreinterpretq_f32_u32(vandq_u32(vmask_odd, vreinterpretq_u32_f32(vi0x8ACE9BDF.val[1]))); const float32x4_t vi1x8ACE = vreinterpretq_f32_u32(vandq_u32(vmask_even, vreinterpretq_u32_f32(vi1x8ACE9BDF.val[0]))); const float32x4_t vi1x9BDF = vreinterpretq_f32_u32(vandq_u32(vmask_odd, vreinterpretq_u32_f32(vi1x8ACE9BDF.val[1]))); const float32x4_t vi2x8ACE = vreinterpretq_f32_u32(vandq_u32(vmask_even, vreinterpretq_u32_f32(vi2x8ACE9BDF.val[0]))); const float32x4_t vi2x9BDF = vreinterpretq_f32_u32(vandq_u32(vmask_odd, vreinterpretq_u32_f32(vi2x8ACE9BDF.val[1]))); const float32x4_t vi3x8ACE = vreinterpretq_f32_u32(vandq_u32(vmask_even, vreinterpretq_u32_f32(vi3x8ACE9BDF.val[0]))); const float32x4_t vi3x9BDF = vreinterpretq_f32_u32(vandq_u32(vmask_odd, vreinterpretq_u32_f32(vi3x8ACE9BDF.val[1]))); const float32x4_t vi4x8ACE = vreinterpretq_f32_u32(vandq_u32(vmask_even, vreinterpretq_u32_f32(vi4x8ACE9BDF.val[0]))); const float32x4_t vi4x9BDF = vreinterpretq_f32_u32(vandq_u32(vmask_odd, vreinterpretq_u32_f32(vi4x8ACE9BDF.val[1]))); const float32x4_t vi5x8ACE = vreinterpretq_f32_u32(vandq_u32(vmask_even, vreinterpretq_u32_f32(vi5x8ACE9BDF.val[0]))); const float32x4_t vi5x9BDF = vreinterpretq_f32_u32(vandq_u32(vmask_odd, vreinterpretq_u32_f32(vi5x8ACE9BDF.val[1]))); const float32x4_t vi6x8ACE = vreinterpretq_f32_u32(vandq_u32(vmask_even, vreinterpretq_u32_f32(vi6x8ACE9BDF.val[0]))); const float32x4_t vi6x9BDF = vreinterpretq_f32_u32(vandq_u32(vmask_odd, vreinterpretq_u32_f32(vi6x8ACE9BDF.val[1]))); const float32x4_t vi7x8ACE = vreinterpretq_f32_u32(vandq_u32(vmask_even, vreinterpretq_u32_f32(vi7x8ACE9BDF.val[0]))); const float32x4_t vi7x9BDF = vreinterpretq_f32_u32(vandq_u32(vmask_odd, vreinterpretq_u32_f32(vi7x8ACE9BDF.val[1]))); const float32x4_t vi8x8ACE = vreinterpretq_f32_u32(vandq_u32(vmask_even, vreinterpretq_u32_f32(vi8x8ACE9BDF.val[0]))); const float32x4_t vi8x9BDF = vreinterpretq_f32_u32(vandq_u32(vmask_odd, vreinterpretq_u32_f32(vi8x8ACE9BDF.val[1]))); vo0p0 = vfmaq_lane_f32(vo0p0, vi0x8ACE, vget_high_f32(vw0123), 0); vo1p0 = vfmaq_lane_f32(vo1p0, vi2x8ACE, vget_high_f32(vw0123), 0); vo2p0 = vfmaq_lane_f32(vo2p0, vi4x8ACE, vget_high_f32(vw0123), 0); vo3p0 = vfmaq_lane_f32(vo3p0, vi6x8ACE, vget_high_f32(vw0123), 0); vo0p0 = vfmaq_lane_f32(vo0p0, vi1x8ACE, vget_low_f32(vw4567), 1); vo1p0 = vfmaq_lane_f32(vo1p0, vi3x8ACE, vget_low_f32(vw4567), 1); vo2p0 = vfmaq_lane_f32(vo2p0, vi5x8ACE, vget_low_f32(vw4567), 1); vo3p0 = vfmaq_lane_f32(vo3p0, vi7x8ACE, vget_low_f32(vw4567), 1); vo0p0 = vfmaq_lane_f32(vo0p0, vi2x8ACE, vw89, 0); vo1p0 = vfmaq_lane_f32(vo1p0, vi4x8ACE, vw89, 0); vo2p0 = vfmaq_lane_f32(vo2p0, vi6x8ACE, vw89, 0); vo3p0 = vfmaq_lane_f32(vo3p0, vi8x8ACE, vw89, 0); const float32x4_t vi0x7BDF = vextq_f32(vi0x1357, vi0x9BDF, 3); const float32x4_t vi1x7BDF = vextq_f32(vi1x1357, vi1x9BDF, 3); const float32x4_t vi2x7BDF = vextq_f32(vi2x1357, vi2x9BDF, 3); const float32x4_t vi3x7BDF = vextq_f32(vi3x1357, vi3x9BDF, 3); const float32x4_t vi4x7BDF = vextq_f32(vi4x1357, vi4x9BDF, 3); const float32x4_t vi5x7BDF = vextq_f32(vi5x1357, vi5x9BDF, 3); const float32x4_t vi6x7BDF = vextq_f32(vi6x1357, vi6x9BDF, 3); const float32x4_t vi7x7BDF = vextq_f32(vi7x1357, vi7x9BDF, 3); const float32x4_t vi8x7BDF = vextq_f32(vi8x1357, vi8x9BDF, 3); vo0p0 = vfmaq_lane_f32(vo0p0, vi0x7BDF, vget_low_f32(vw0123), 1); vo1p0 = vfmaq_lane_f32(vo1p0, vi2x7BDF, vget_low_f32(vw0123), 1); vo2p0 = vfmaq_lane_f32(vo2p0, vi4x7BDF, vget_low_f32(vw0123), 1); vo3p0 = vfmaq_lane_f32(vo3p0, vi6x7BDF, vget_low_f32(vw0123), 1); vo0p0 = vfmaq_lane_f32(vo0p0, vi1x7BDF, vget_low_f32(vw4567), 0); vo1p0 = vfmaq_lane_f32(vo1p0, vi3x7BDF, vget_low_f32(vw4567), 0); vo2p0 = vfmaq_lane_f32(vo2p0, vi5x7BDF, vget_low_f32(vw4567), 0); vo3p0 = vfmaq_lane_f32(vo3p0, vi7x7BDF, vget_low_f32(vw4567), 0); vo0p0 = vfmaq_lane_f32(vo0p0, vi2x7BDF, vget_high_f32(vw4567), 1); vo1p0 = vfmaq_lane_f32(vo1p0, vi4x7BDF, vget_high_f32(vw4567), 1); vo2p0 = vfmaq_lane_f32(vo2p0, vi6x7BDF, vget_high_f32(vw4567), 1); vo3p0 = vfmaq_lane_f32(vo3p0, vi8x7BDF, vget_high_f32(vw4567), 1); vo0p0 = vfmaq_lane_f32(vo0p0, vi0x9BDF, vget_high_f32(vw0123), 1); vo1p0 = vfmaq_lane_f32(vo1p0, vi2x9BDF, vget_high_f32(vw0123), 1); vo2p0 = vfmaq_lane_f32(vo2p0, vi4x9BDF, vget_high_f32(vw0123), 1); vo3p0 = vfmaq_lane_f32(vo3p0, vi6x9BDF, vget_high_f32(vw0123), 1); vo0p0 = vfmaq_lane_f32(vo0p0, vi1x9BDF, vget_high_f32(vw4567), 0); vo1p0 = vfmaq_lane_f32(vo1p0, vi3x9BDF, vget_high_f32(vw4567), 0); vo2p0 = vfmaq_lane_f32(vo2p0, vi5x9BDF, vget_high_f32(vw4567), 0); vo3p0 = vfmaq_lane_f32(vo3p0, vi7x9BDF, vget_high_f32(vw4567), 0); vo0p0 = vfmaq_lane_f32(vo0p0, vi2x9BDF, vw89, 1); vo1p0 = vfmaq_lane_f32(vo1p0, vi4x9BDF, vw89, 1); vo2p0 = vfmaq_lane_f32(vo2p0, vi6x9BDF, vw89, 1); vo3p0 = vfmaq_lane_f32(vo3p0, vi8x9BDF, vw89, 1); float32x4_t vo0 = vmaxq_f32(vo0p0, vmin); float32x4_t vo1 = vmaxq_f32(vo1p0, vmin); float32x4_t vo2 = vmaxq_f32(vo2p0, vmin); float32x4_t vo3 = vmaxq_f32(vo3p0, vmin); vo0 = vminq_f32(vo0, vmax); vo1 = vminq_f32(vo1, vmax); vo2 = vminq_f32(vo2, vmax); vo3 = vminq_f32(vo3, vmax); w += 1 * sizeof(float); if (w & (8 * sizeof(float))) { vst1q_f32(o3, vo3); o3 += 4; vst1q_f32(o2, vo2); o2 += 4; vst1q_f32(o1, vo1); o1 += 4; vst1q_f32(o0, vo0); o0 += 4; } else { float32x2_t vo0_lo = vget_low_f32(vo0); float32x2_t vo1_lo = vget_low_f32(vo1); float32x2_t vo2_lo = vget_low_f32(vo2); float32x2_t vo3_lo = vget_low_f32(vo3); if (w & (4 * sizeof(float))) { vst1_f32(o3, vo3_lo); o3 += 2; vst1_f32(o2, vo2_lo); o2 += 2; vst1_f32(o1, vo1_lo); o1 += 2; vst1_f32(o0, vo0_lo); o0 += 2; vo0_lo = vget_high_f32(vo0); vo1_lo = vget_high_f32(vo1); vo2_lo = vget_high_f32(vo2); vo3_lo = vget_high_f32(vo3); } if (w & (2 * sizeof(float))) { vst1_lane_f32(o3, vo3_lo, 0); o3 += 1; vst1_lane_f32(o2, vo2_lo, 0); o2 += 1; vst1_lane_f32(o1, vo1_lo, 0); o1 += 1; vst1_lane_f32(o0, vo0_lo, 0); o0 += 1; } } } i0 = (const float*) ((uintptr_t) i8 - input_decrement); i1 = (const float*) ((uintptr_t) i0 + input_width); i2 = (const float*) ((uintptr_t) i1 + input_width); i3 = (const float*) ((uintptr_t) i2 + input_width); i4 = (const float*) ((uintptr_t) i3 + input_width); i5 = (const float*) ((uintptr_t) i4 + input_width); i6 = (const float*) ((uintptr_t) i5 + input_width); i7 = (const float*) ((uintptr_t) i6 + input_width); i8 = (const float*) ((uintptr_t) i7 + input_width); o0 = o3; o1 = (float*) ((uintptr_t) o0 + output_width); o2 = (float*) ((uintptr_t) o1 + output_width); o3 = (float*) ((uintptr_t) o2 + output_width); output_height = doz(output_height, 4); padded_input_height = doz(padded_input_height, 8); } while (output_height != 0); }