| /external/XNNPACK/src/f32-argmaxpool/ |
| D | 4x-wasmsimd-c4.c | 63 const v128_t vm1 = wasm_f32x4_gt(vi1, vmax); in xnn_f32_argmaxpool_ukernel_4x__wasmsimd_c4() local
64 vmax = wasm_v128_bitselect(vi1, vmax, vm1); in xnn_f32_argmaxpool_ukernel_4x__wasmsimd_c4()
65 vidx = wasm_v128_bitselect(wasm_i32x4_const_splat(1), vidx, vm1); in xnn_f32_argmaxpool_ukernel_4x__wasmsimd_c4()
89 const v128_t vm1 = wasm_f32x4_gt(vi1, vmax); in xnn_f32_argmaxpool_ukernel_4x__wasmsimd_c4() local
90 vmax = wasm_v128_bitselect(vi1, vmax, vm1); in xnn_f32_argmaxpool_ukernel_4x__wasmsimd_c4()
91 vidx = wasm_v128_bitselect(wasm_i32x4_const_splat(1), vidx, vm1); in xnn_f32_argmaxpool_ukernel_4x__wasmsimd_c4()
|
| D | 4x-neon-c4.c | 58 const uint32x4_t vm1 = vcgtq_f32(vi1, vmax); in xnn_f32_argmaxpool_ukernel_4x__neon_c4() local
59 vmax = vbslq_f32(vm1, vi1, vmax); in xnn_f32_argmaxpool_ukernel_4x__neon_c4()
60 vidx = vbslq_u32(vm1, vmovq_n_u32(1), vidx); in xnn_f32_argmaxpool_ukernel_4x__neon_c4()
82 const uint32x4_t vm1 = vcgtq_f32(vi1, vmax); in xnn_f32_argmaxpool_ukernel_4x__neon_c4() local
83 vmax = vbslq_f32(vm1, vi1, vmax); in xnn_f32_argmaxpool_ukernel_4x__neon_c4()
84 vidx = vbslq_u32(vm1, vmovq_n_u32(1), vidx); in xnn_f32_argmaxpool_ukernel_4x__neon_c4()
|
| D | 4x-sse2-c4.c | 62 const __m128i vm1 = _mm_castps_si128(_mm_cmpgt_ps(vi1, vmax)); in xnn_f32_argmaxpool_ukernel_4x__sse2_c4() local
64 vidx = _mm_or_si128(_mm_andnot_si128(vm1, vidx), _mm_and_si128(vm1, _mm_set1_epi32(1))); in xnn_f32_argmaxpool_ukernel_4x__sse2_c4()
88 const __m128i vm1 = _mm_castps_si128(_mm_cmpgt_ps(vi1, vmax)); in xnn_f32_argmaxpool_ukernel_4x__sse2_c4() local
90 vidx = _mm_or_si128(_mm_andnot_si128(vm1, vidx), _mm_and_si128(vm1, _mm_set1_epi32(1))); in xnn_f32_argmaxpool_ukernel_4x__sse2_c4()
|
| D | 9p8x-wasmsimd-c4.c | 78 const v128_t vm1 = wasm_f32x4_gt(vi1, vmax); in xnn_f32_argmaxpool_ukernel_9p8x__wasmsimd_c4() local
79 vmax = wasm_v128_bitselect(vi1, vmax, vm1); in xnn_f32_argmaxpool_ukernel_9p8x__wasmsimd_c4()
80 vidx = wasm_v128_bitselect(wasm_i32x4_const_splat(1), vidx, vm1); in xnn_f32_argmaxpool_ukernel_9p8x__wasmsimd_c4()
167 const v128_t vm1 = wasm_f32x4_gt(vi1, vmax); in xnn_f32_argmaxpool_ukernel_9p8x__wasmsimd_c4() local
169 vmax = wasm_v128_bitselect(vi1, vmax, vm1); in xnn_f32_argmaxpool_ukernel_9p8x__wasmsimd_c4()
170 vidx = wasm_v128_bitselect(vidx1, vidx, vm1); in xnn_f32_argmaxpool_ukernel_9p8x__wasmsimd_c4()
282 const v128_t vm1 = wasm_f32x4_gt(vi1, vmax); in xnn_f32_argmaxpool_ukernel_9p8x__wasmsimd_c4() local
284 vmax = wasm_v128_bitselect(vi1, vmax, vm1); in xnn_f32_argmaxpool_ukernel_9p8x__wasmsimd_c4()
285 vidx = wasm_v128_bitselect(vidx1, vidx, vm1); in xnn_f32_argmaxpool_ukernel_9p8x__wasmsimd_c4()
339 const v128_t vm1 = wasm_f32x4_gt(vi1, vmax); in xnn_f32_argmaxpool_ukernel_9p8x__wasmsimd_c4() local
[all …]
|
| D | 9p8x-neon-c4.c | 69 const uint32x4_t vm1 = vcgtq_f32(vi1, vmax); in xnn_f32_argmaxpool_ukernel_9p8x__neon_c4() local
70 vmax = vbslq_f32(vm1, vi1, vmax); in xnn_f32_argmaxpool_ukernel_9p8x__neon_c4()
71 vidx = vbslq_u32(vm1, vmovq_n_u32(1), vidx); in xnn_f32_argmaxpool_ukernel_9p8x__neon_c4()
148 const uint32x4_t vm1 = vcgtq_f32(vi1, vmax); in xnn_f32_argmaxpool_ukernel_9p8x__neon_c4() local
150 vmax = vbslq_f32(vm1, vi1, vmax); in xnn_f32_argmaxpool_ukernel_9p8x__neon_c4()
151 vidx = vbslq_u32(vm1, vidx1, vidx); in xnn_f32_argmaxpool_ukernel_9p8x__neon_c4()
251 const uint32x4_t vm1 = vcgtq_f32(vi1, vmax); in xnn_f32_argmaxpool_ukernel_9p8x__neon_c4() local
253 vmax = vbslq_f32(vm1, vi1, vmax); in xnn_f32_argmaxpool_ukernel_9p8x__neon_c4()
254 vidx = vbslq_u32(vm1, vidx1, vidx); in xnn_f32_argmaxpool_ukernel_9p8x__neon_c4()
306 const uint32x4_t vm1 = vcgtq_f32(vi1, vmax); in xnn_f32_argmaxpool_ukernel_9p8x__neon_c4() local
[all …]
|
| D | 9p8x-sse2-c4.c | 78 const __m128i vm1 = _mm_castps_si128(_mm_cmpgt_ps(vi1, vmax)); in xnn_f32_argmaxpool_ukernel_9p8x__sse2_c4() local
80 vidx = _mm_or_si128(_mm_andnot_si128(vm1, vidx), _mm_and_si128(vm1, _mm_set1_epi32(1))); in xnn_f32_argmaxpool_ukernel_9p8x__sse2_c4()
167 const __m128i vm1 = _mm_castps_si128(_mm_cmpgt_ps(vi1, vmax)); in xnn_f32_argmaxpool_ukernel_9p8x__sse2_c4() local
170 vidx = _mm_or_si128(_mm_andnot_si128(vm1, vidx), _mm_and_si128(vm1, vidx1)); in xnn_f32_argmaxpool_ukernel_9p8x__sse2_c4()
282 const __m128i vm1 = _mm_castps_si128(_mm_cmpgt_ps(vi1, vmax)); in xnn_f32_argmaxpool_ukernel_9p8x__sse2_c4() local
285 vidx = _mm_or_si128(_mm_andnot_si128(vm1, vidx), _mm_and_si128(vm1, vidx1)); in xnn_f32_argmaxpool_ukernel_9p8x__sse2_c4()
339 const __m128i vm1 = _mm_castps_si128(_mm_cmpgt_ps(vi1, vmax)); in xnn_f32_argmaxpool_ukernel_9p8x__sse2_c4() local
342 vidx = _mm_or_si128(_mm_andnot_si128(vm1, vidx), _mm_and_si128(vm1, vidx1)); in xnn_f32_argmaxpool_ukernel_9p8x__sse2_c4()
|
| D | 9x-neon-c4.c | 88 const uint32x4_t vm1 = vcgtq_f32(vi1, vmax); in xnn_f32_argmaxpool_ukernel_9x__neon_c4() local
89 vmax = vbslq_f32(vm1, vi1, vmax); in xnn_f32_argmaxpool_ukernel_9x__neon_c4()
90 vidx = vbslq_u32(vm1, vmovq_n_u32(1), vidx); in xnn_f32_argmaxpool_ukernel_9x__neon_c4()
137 const uint32x4_t vm1 = vcgtq_f32(vi1, vmax); in xnn_f32_argmaxpool_ukernel_9x__neon_c4() local
138 vmax = vbslq_f32(vm1, vi1, vmax); in xnn_f32_argmaxpool_ukernel_9x__neon_c4()
139 vidx = vbslq_u32(vm1, vmovq_n_u32(1), vidx); in xnn_f32_argmaxpool_ukernel_9x__neon_c4()
|
| D | 9x-wasmsimd-c4.c | 98 const v128_t vm1 = wasm_f32x4_gt(vi1, vmax); in xnn_f32_argmaxpool_ukernel_9x__wasmsimd_c4() local
99 vmax = wasm_v128_bitselect(vi1, vmax, vm1); in xnn_f32_argmaxpool_ukernel_9x__wasmsimd_c4()
100 vidx = wasm_v128_bitselect(wasm_i32x4_const_splat(1), vidx, vm1); in xnn_f32_argmaxpool_ukernel_9x__wasmsimd_c4()
149 const v128_t vm1 = wasm_f32x4_gt(vi1, vmax); in xnn_f32_argmaxpool_ukernel_9x__wasmsimd_c4() local
150 vmax = wasm_v128_bitselect(vi1, vmax, vm1); in xnn_f32_argmaxpool_ukernel_9x__wasmsimd_c4()
151 vidx = wasm_v128_bitselect(wasm_i32x4_const_splat(1), vidx, vm1); in xnn_f32_argmaxpool_ukernel_9x__wasmsimd_c4()
|
| D | 9x-sse2-c4.c | 97 const __m128i vm1 = _mm_castps_si128(_mm_cmpgt_ps(vi1, vmax)); in xnn_f32_argmaxpool_ukernel_9x__sse2_c4() local
99 vidx = _mm_or_si128(_mm_andnot_si128(vm1, vidx), _mm_and_si128(vm1, _mm_set1_epi32(1))); in xnn_f32_argmaxpool_ukernel_9x__sse2_c4()
148 const __m128i vm1 = _mm_castps_si128(_mm_cmpgt_ps(vi1, vmax)); in xnn_f32_argmaxpool_ukernel_9x__sse2_c4() local
150 vidx = _mm_or_si128(_mm_andnot_si128(vm1, vidx), _mm_and_si128(vm1, _mm_set1_epi32(1))); in xnn_f32_argmaxpool_ukernel_9x__sse2_c4()
|
| /external/XNNPACK/src/qs8-vcvt/gen/ |
| D | vcvt-ssse3-x32.c | 41 const __m128i vm1 = _mm_cmpgt_epi8(_mm_setzero_si128(), vx1); in xnn_qs8_vcvt_ukernel__ssse3_x32() local
42 __m128i vacc2 = _mm_unpacklo_epi8(vx1, vm1); in xnn_qs8_vcvt_ukernel__ssse3_x32()
43 __m128i vacc3 = _mm_unpackhi_epi8(vx1, vm1); in xnn_qs8_vcvt_ukernel__ssse3_x32()
|
| D | vcvt-sse2-x32.c | 40 const __m128i vm1 = _mm_cmpgt_epi8(_mm_setzero_si128(), vx1); in xnn_qs8_vcvt_ukernel__sse2_x32() local
41 const __m128i vextx2 = _mm_unpacklo_epi8(vx1, vm1); in xnn_qs8_vcvt_ukernel__sse2_x32()
42 const __m128i vextx3 = _mm_unpackhi_epi8(vx1, vm1); in xnn_qs8_vcvt_ukernel__sse2_x32()
|
| /external/clang/test/CXX/special/class.copy/ |
| D | p11.0x.copy.cpp | 77 } vm1, vm2(vm1); // expected-error {{deleted}} variable
|
| /external/XNNPACK/src/bf16-gemm/gen/ |
| D | 5x4c8-minmax-neonbf16-bfdot.c | 136 const uint16x8_t vm1 = vceqq_u16(vreinterpretq_u16_bf16(vb1), vmovq_n_u16(0)); in xnn_bf16_gemm_minmax_ukernel_5x4c8__neonbf16_bfdot() local
150 … const bfloat16x8_t va0x1 = vreinterpretq_bf16_u16(vbicq_u16(vreinterpretq_u16_bf16(va0), vm1)); in xnn_bf16_gemm_minmax_ukernel_5x4c8__neonbf16_bfdot()
152 … const bfloat16x8_t va1x1 = vreinterpretq_bf16_u16(vbicq_u16(vreinterpretq_u16_bf16(va1), vm1)); in xnn_bf16_gemm_minmax_ukernel_5x4c8__neonbf16_bfdot()
154 … const bfloat16x8_t va2x1 = vreinterpretq_bf16_u16(vbicq_u16(vreinterpretq_u16_bf16(va2), vm1)); in xnn_bf16_gemm_minmax_ukernel_5x4c8__neonbf16_bfdot()
156 … const bfloat16x8_t va3x1 = vreinterpretq_bf16_u16(vbicq_u16(vreinterpretq_u16_bf16(va3), vm1)); in xnn_bf16_gemm_minmax_ukernel_5x4c8__neonbf16_bfdot()
158 … const bfloat16x8_t va4x1 = vreinterpretq_bf16_u16(vbicq_u16(vreinterpretq_u16_bf16(va4), vm1)); in xnn_bf16_gemm_minmax_ukernel_5x4c8__neonbf16_bfdot()
|
| D | 4x4c8-minmax-neonbf16-bfdot.c | 120 const uint16x8_t vm1 = vceqq_u16(vreinterpretq_u16_bf16(vb1), vmovq_n_u16(0)); in xnn_bf16_gemm_minmax_ukernel_4x4c8__neonbf16_bfdot() local
132 … const bfloat16x8_t va0x1 = vreinterpretq_bf16_u16(vbicq_u16(vreinterpretq_u16_bf16(va0), vm1)); in xnn_bf16_gemm_minmax_ukernel_4x4c8__neonbf16_bfdot()
134 … const bfloat16x8_t va1x1 = vreinterpretq_bf16_u16(vbicq_u16(vreinterpretq_u16_bf16(va1), vm1)); in xnn_bf16_gemm_minmax_ukernel_4x4c8__neonbf16_bfdot()
136 … const bfloat16x8_t va2x1 = vreinterpretq_bf16_u16(vbicq_u16(vreinterpretq_u16_bf16(va2), vm1)); in xnn_bf16_gemm_minmax_ukernel_4x4c8__neonbf16_bfdot()
138 … const bfloat16x8_t va3x1 = vreinterpretq_bf16_u16(vbicq_u16(vreinterpretq_u16_bf16(va3), vm1)); in xnn_bf16_gemm_minmax_ukernel_4x4c8__neonbf16_bfdot()
|
| D | 3x4c8-minmax-neonbf16-bfdot.c | 104 const uint16x8_t vm1 = vceqq_u16(vreinterpretq_u16_bf16(vb1), vmovq_n_u16(0)); in xnn_bf16_gemm_minmax_ukernel_3x4c8__neonbf16_bfdot() local
114 … const bfloat16x8_t va0x1 = vreinterpretq_bf16_u16(vbicq_u16(vreinterpretq_u16_bf16(va0), vm1)); in xnn_bf16_gemm_minmax_ukernel_3x4c8__neonbf16_bfdot()
116 … const bfloat16x8_t va1x1 = vreinterpretq_bf16_u16(vbicq_u16(vreinterpretq_u16_bf16(va1), vm1)); in xnn_bf16_gemm_minmax_ukernel_3x4c8__neonbf16_bfdot()
118 … const bfloat16x8_t va2x1 = vreinterpretq_bf16_u16(vbicq_u16(vreinterpretq_u16_bf16(va2), vm1)); in xnn_bf16_gemm_minmax_ukernel_3x4c8__neonbf16_bfdot()
|
| D | 5x4c8-minmax-neonbf16-bfmlal.c | 157 const uint16x8_t vm1 = vceqq_u16(vreinterpretq_u16_bf16(vb1), vmovq_n_u16(0)); in xnn_bf16_gemm_minmax_ukernel_5x4c8__neonbf16_bfmlal() local
176 … const bfloat16x8_t va0x1 = vreinterpretq_bf16_u16(vbicq_u16(vreinterpretq_u16_bf16(va0), vm1)); in xnn_bf16_gemm_minmax_ukernel_5x4c8__neonbf16_bfmlal()
179 … const bfloat16x8_t va1x1 = vreinterpretq_bf16_u16(vbicq_u16(vreinterpretq_u16_bf16(va1), vm1)); in xnn_bf16_gemm_minmax_ukernel_5x4c8__neonbf16_bfmlal()
182 … const bfloat16x8_t va2x1 = vreinterpretq_bf16_u16(vbicq_u16(vreinterpretq_u16_bf16(va2), vm1)); in xnn_bf16_gemm_minmax_ukernel_5x4c8__neonbf16_bfmlal()
185 … const bfloat16x8_t va3x1 = vreinterpretq_bf16_u16(vbicq_u16(vreinterpretq_u16_bf16(va3), vm1)); in xnn_bf16_gemm_minmax_ukernel_5x4c8__neonbf16_bfmlal()
188 … const bfloat16x8_t va4x1 = vreinterpretq_bf16_u16(vbicq_u16(vreinterpretq_u16_bf16(va4), vm1)); in xnn_bf16_gemm_minmax_ukernel_5x4c8__neonbf16_bfmlal()
|
| D | 4x4c8-minmax-neonbf16-bfmlal.c | 137 const uint16x8_t vm1 = vceqq_u16(vreinterpretq_u16_bf16(vb1), vmovq_n_u16(0)); in xnn_bf16_gemm_minmax_ukernel_4x4c8__neonbf16_bfmlal() local
153 … const bfloat16x8_t va0x1 = vreinterpretq_bf16_u16(vbicq_u16(vreinterpretq_u16_bf16(va0), vm1)); in xnn_bf16_gemm_minmax_ukernel_4x4c8__neonbf16_bfmlal()
156 … const bfloat16x8_t va1x1 = vreinterpretq_bf16_u16(vbicq_u16(vreinterpretq_u16_bf16(va1), vm1)); in xnn_bf16_gemm_minmax_ukernel_4x4c8__neonbf16_bfmlal()
159 … const bfloat16x8_t va2x1 = vreinterpretq_bf16_u16(vbicq_u16(vreinterpretq_u16_bf16(va2), vm1)); in xnn_bf16_gemm_minmax_ukernel_4x4c8__neonbf16_bfmlal()
162 … const bfloat16x8_t va3x1 = vreinterpretq_bf16_u16(vbicq_u16(vreinterpretq_u16_bf16(va3), vm1)); in xnn_bf16_gemm_minmax_ukernel_4x4c8__neonbf16_bfmlal()
|
| D | 2x4c8-minmax-neonbf16-bfdot.c | 88 const uint16x8_t vm1 = vceqq_u16(vreinterpretq_u16_bf16(vb1), vmovq_n_u16(0)); in xnn_bf16_gemm_minmax_ukernel_2x4c8__neonbf16_bfdot() local
96 … const bfloat16x8_t va0x1 = vreinterpretq_bf16_u16(vbicq_u16(vreinterpretq_u16_bf16(va0), vm1)); in xnn_bf16_gemm_minmax_ukernel_2x4c8__neonbf16_bfdot()
98 … const bfloat16x8_t va1x1 = vreinterpretq_bf16_u16(vbicq_u16(vreinterpretq_u16_bf16(va1), vm1)); in xnn_bf16_gemm_minmax_ukernel_2x4c8__neonbf16_bfdot()
|
| D | 3x4c8-minmax-neonbf16-bfmlal.c | 117 const uint16x8_t vm1 = vceqq_u16(vreinterpretq_u16_bf16(vb1), vmovq_n_u16(0)); in xnn_bf16_gemm_minmax_ukernel_3x4c8__neonbf16_bfmlal() local
130 … const bfloat16x8_t va0x1 = vreinterpretq_bf16_u16(vbicq_u16(vreinterpretq_u16_bf16(va0), vm1)); in xnn_bf16_gemm_minmax_ukernel_3x4c8__neonbf16_bfmlal()
133 … const bfloat16x8_t va1x1 = vreinterpretq_bf16_u16(vbicq_u16(vreinterpretq_u16_bf16(va1), vm1)); in xnn_bf16_gemm_minmax_ukernel_3x4c8__neonbf16_bfmlal()
136 … const bfloat16x8_t va2x1 = vreinterpretq_bf16_u16(vbicq_u16(vreinterpretq_u16_bf16(va2), vm1)); in xnn_bf16_gemm_minmax_ukernel_3x4c8__neonbf16_bfmlal()
|
| D | 2x4c8-minmax-neonbf16-bfmlal.c | 97 const uint16x8_t vm1 = vceqq_u16(vreinterpretq_u16_bf16(vb1), vmovq_n_u16(0)); in xnn_bf16_gemm_minmax_ukernel_2x4c8__neonbf16_bfmlal() local
107 … const bfloat16x8_t va0x1 = vreinterpretq_bf16_u16(vbicq_u16(vreinterpretq_u16_bf16(va0), vm1)); in xnn_bf16_gemm_minmax_ukernel_2x4c8__neonbf16_bfmlal()
110 … const bfloat16x8_t va1x1 = vreinterpretq_bf16_u16(vbicq_u16(vreinterpretq_u16_bf16(va1), vm1)); in xnn_bf16_gemm_minmax_ukernel_2x4c8__neonbf16_bfmlal()
|
| D | 1x4c8-minmax-neonbf16-bfdot.c | 72 const uint16x8_t vm1 = vceqq_u16(vreinterpretq_u16_bf16(vb1), vmovq_n_u16(0)); in xnn_bf16_gemm_minmax_ukernel_1x4c8__neonbf16_bfdot() local
78 … const bfloat16x8_t va0x1 = vreinterpretq_bf16_u16(vbicq_u16(vreinterpretq_u16_bf16(va0), vm1)); in xnn_bf16_gemm_minmax_ukernel_1x4c8__neonbf16_bfdot()
|
| D | 4x4c8-minmax-neonfma-zip.c | 158 const uint16x8_t vm1 = vceqq_u16(vb1, vmovq_n_u16(0)); in xnn_bf16_gemm_minmax_ukernel_4x4c8__neonfma_zip() local
171 const uint16x8_t va0x1 = vbicq_u16(va0, vm1); in xnn_bf16_gemm_minmax_ukernel_4x4c8__neonfma_zip()
172 const uint16x8_t va1x1 = vbicq_u16(va1, vm1); in xnn_bf16_gemm_minmax_ukernel_4x4c8__neonfma_zip()
173 const uint16x8_t va2x1 = vbicq_u16(va2, vm1); in xnn_bf16_gemm_minmax_ukernel_4x4c8__neonfma_zip()
174 const uint16x8_t va3x1 = vbicq_u16(va3, vm1); in xnn_bf16_gemm_minmax_ukernel_4x4c8__neonfma_zip()
|
| D | 4x4c8-minmax-neonfma-shland.c | 158 const uint16x8_t vm1 = vceqq_u16(vb1, vmovq_n_u16(0)); in xnn_bf16_gemm_minmax_ukernel_4x4c8__neonfma_shland() local
171 const uint16x8_t va0x1 = vbicq_u16(va0, vm1); in xnn_bf16_gemm_minmax_ukernel_4x4c8__neonfma_shland()
172 const uint16x8_t va1x1 = vbicq_u16(va1, vm1); in xnn_bf16_gemm_minmax_ukernel_4x4c8__neonfma_shland()
173 const uint16x8_t va2x1 = vbicq_u16(va2, vm1); in xnn_bf16_gemm_minmax_ukernel_4x4c8__neonfma_shland()
174 const uint16x8_t va3x1 = vbicq_u16(va3, vm1); in xnn_bf16_gemm_minmax_ukernel_4x4c8__neonfma_shland()
|
| /external/XNNPACK/src/qs8-vlrelu/gen/ |
| D | vlrelu-ssse3-x32.c | 42 const __m128i vm1 = _mm_cmpgt_epi8(_mm_setzero_si128(), vx1); in xnn_qs8_vlrelu_ukernel__ssse3_x32() local
43 __m128i vacc2 = _mm_unpacklo_epi8(vx1, vm1); in xnn_qs8_vlrelu_ukernel__ssse3_x32()
44 __m128i vacc3 = _mm_unpackhi_epi8(vx1, vm1); in xnn_qs8_vlrelu_ukernel__ssse3_x32()
|
| D | vlrelu-sse2-x32.c | 43 const __m128i vm1 = _mm_cmpgt_epi8(_mm_setzero_si128(), vx1); in xnn_qs8_vlrelu_ukernel__sse2_x32() local
44 __m128i vextx2 = _mm_unpacklo_epi8(vx1, vm1); in xnn_qs8_vlrelu_ukernel__sse2_x32()
45 __m128i vextx3 = _mm_unpackhi_epi8(vx1, vm1); in xnn_qs8_vlrelu_ukernel__sse2_x32()
|