Home
last modified time | relevance | path

Searched refs:vk0x0123 (Results 1 – 25 of 66) sorted by relevance

123

/external/XNNPACK/src/f32-dwconv/gen/
Dup8x4-wasmsimd.c66 const v128_t vk0x0123 = wasm_v128_load(w + 8); in xnn_f32_dwconv_ukernel_up8x4__wasmsimd() local
68 vacc0123p0 = wasm_f32x4_add(vacc0123p0, wasm_f32x4_mul(vi0x0123, vk0x0123)); in xnn_f32_dwconv_ukernel_up8x4__wasmsimd()
114 const v128_t vk0x0123 = wasm_v128_load(w + 8); in xnn_f32_dwconv_ukernel_up8x4__wasmsimd() local
115 vacc0123p0 = wasm_f32x4_add(vacc0123p0, wasm_f32x4_mul(vi0x0123, vk0x0123)); in xnn_f32_dwconv_ukernel_up8x4__wasmsimd()
147 const v128_t vk0x0123 = wasm_v128_load(w + 8); in xnn_f32_dwconv_ukernel_up8x4__wasmsimd() local
148 vacc0123p0 = wasm_f32x4_add(vacc0123p0, wasm_f32x4_mul(vi0x0123, vk0x0123)); in xnn_f32_dwconv_ukernel_up8x4__wasmsimd()
Dup8x4-minmax-sse-acc2.c68 const __m128 vk0x0123 = _mm_load_ps(w + 8); in xnn_f32_dwconv_minmax_ukernel_up8x4__sse_acc2() local
70 vacc0123p0 = _mm_add_ps(vacc0123p0, _mm_mul_ps(vi0x0123, vk0x0123)); in xnn_f32_dwconv_minmax_ukernel_up8x4__sse_acc2()
121 const __m128 vk0x0123 = _mm_load_ps(w + 8); in xnn_f32_dwconv_minmax_ukernel_up8x4__sse_acc2() local
122 vacc0123p0 = _mm_add_ps(vacc0123p0, _mm_mul_ps(vi0x0123, vk0x0123)); in xnn_f32_dwconv_minmax_ukernel_up8x4__sse_acc2()
157 const __m128 vk0x0123 = _mm_load_ps(w + 8); in xnn_f32_dwconv_minmax_ukernel_up8x4__sse_acc2() local
158 vacc0123p0 = _mm_add_ps(vacc0123p0, _mm_mul_ps(vi0x0123, vk0x0123)); in xnn_f32_dwconv_minmax_ukernel_up8x4__sse_acc2()
Dup8x4-minmax-neon.c67 const float32x4_t vk0x0123 = vld1q_f32(w); w += 4; in xnn_f32_dwconv_minmax_ukernel_up8x4__neon() local
69 vacc0123p0 = vmlaq_f32(vacc0123p0, vi0x0123, vk0x0123); in xnn_f32_dwconv_minmax_ukernel_up8x4__neon()
107 const float32x4_t vk0x0123 = vld1q_f32(w + 4); in xnn_f32_dwconv_minmax_ukernel_up8x4__neon() local
108 vacc0123p0 = vmlaq_f32(vacc0123p0, vi0x0123, vk0x0123); in xnn_f32_dwconv_minmax_ukernel_up8x4__neon()
133 const float32x4_t vk0x0123 = vld1q_f32(w + 8); in xnn_f32_dwconv_minmax_ukernel_up8x4__neon() local
134 vacc0123p0 = vmlaq_f32(vacc0123p0, vi0x0123, vk0x0123); in xnn_f32_dwconv_minmax_ukernel_up8x4__neon()
Dup8x4-minmax-wasmsimd-arm-acc2.c68 const v128_t vk0x0123 = wasm_v128_load(w + 8); in xnn_f32_dwconv_minmax_ukernel_up8x4__wasmsimd_arm_acc2() local
70 vacc0123p0 = wasm_f32x4_add(vacc0123p0, wasm_f32x4_mul(vi0x0123, vk0x0123)); in xnn_f32_dwconv_minmax_ukernel_up8x4__wasmsimd_arm_acc2()
122 const v128_t vk0x0123 = wasm_v128_load(w + 8); in xnn_f32_dwconv_minmax_ukernel_up8x4__wasmsimd_arm_acc2() local
123 vacc0123p0 = wasm_f32x4_add(vacc0123p0, wasm_f32x4_mul(vi0x0123, vk0x0123)); in xnn_f32_dwconv_minmax_ukernel_up8x4__wasmsimd_arm_acc2()
158 const v128_t vk0x0123 = wasm_v128_load(w + 8); in xnn_f32_dwconv_minmax_ukernel_up8x4__wasmsimd_arm_acc2() local
159 vacc0123p0 = wasm_f32x4_add(vacc0123p0, wasm_f32x4_mul(vi0x0123, vk0x0123)); in xnn_f32_dwconv_minmax_ukernel_up8x4__wasmsimd_arm_acc2()
Dup8x4-minmax-neonfma.c67 const float32x4_t vk0x0123 = vld1q_f32(w); w += 4; in xnn_f32_dwconv_minmax_ukernel_up8x4__neonfma() local
69 vacc0123p0 = vfmaq_f32(vacc0123p0, vi0x0123, vk0x0123); in xnn_f32_dwconv_minmax_ukernel_up8x4__neonfma()
107 const float32x4_t vk0x0123 = vld1q_f32(w + 4); in xnn_f32_dwconv_minmax_ukernel_up8x4__neonfma() local
108 vacc0123p0 = vfmaq_f32(vacc0123p0, vi0x0123, vk0x0123); in xnn_f32_dwconv_minmax_ukernel_up8x4__neonfma()
133 const float32x4_t vk0x0123 = vld1q_f32(w + 8); in xnn_f32_dwconv_minmax_ukernel_up8x4__neonfma() local
134 vacc0123p0 = vfmaq_f32(vacc0123p0, vi0x0123, vk0x0123); in xnn_f32_dwconv_minmax_ukernel_up8x4__neonfma()
Dup8x4-minmax-neon-acc2.c67 const float32x4_t vk0x0123 = vld1q_f32(w); w += 4; in xnn_f32_dwconv_minmax_ukernel_up8x4__neon_acc2() local
69 vacc0123p0 = vmlaq_f32(vacc0123p0, vi0x0123, vk0x0123); in xnn_f32_dwconv_minmax_ukernel_up8x4__neon_acc2()
110 const float32x4_t vk0x0123 = vld1q_f32(w + 4); in xnn_f32_dwconv_minmax_ukernel_up8x4__neon_acc2() local
111 vacc0123p0 = vmlaq_f32(vacc0123p0, vi0x0123, vk0x0123); in xnn_f32_dwconv_minmax_ukernel_up8x4__neon_acc2()
138 const float32x4_t vk0x0123 = vld1q_f32(w + 8); in xnn_f32_dwconv_minmax_ukernel_up8x4__neon_acc2() local
139 vacc0123p0 = vmlaq_f32(vacc0123p0, vi0x0123, vk0x0123); in xnn_f32_dwconv_minmax_ukernel_up8x4__neon_acc2()
Dup8x4-minmax-neonfma-acc2.c67 const float32x4_t vk0x0123 = vld1q_f32(w); w += 4; in xnn_f32_dwconv_minmax_ukernel_up8x4__neonfma_acc2() local
69 vacc0123p0 = vfmaq_f32(vacc0123p0, vi0x0123, vk0x0123); in xnn_f32_dwconv_minmax_ukernel_up8x4__neonfma_acc2()
110 const float32x4_t vk0x0123 = vld1q_f32(w + 4); in xnn_f32_dwconv_minmax_ukernel_up8x4__neonfma_acc2() local
111 vacc0123p0 = vfmaq_f32(vacc0123p0, vi0x0123, vk0x0123); in xnn_f32_dwconv_minmax_ukernel_up8x4__neonfma_acc2()
138 const float32x4_t vk0x0123 = vld1q_f32(w + 8); in xnn_f32_dwconv_minmax_ukernel_up8x4__neonfma_acc2() local
139 vacc0123p0 = vfmaq_f32(vacc0123p0, vi0x0123, vk0x0123); in xnn_f32_dwconv_minmax_ukernel_up8x4__neonfma_acc2()
Dup8x4-minmax-wasmsimd-arm.c68 const v128_t vk0x0123 = wasm_v128_load(w + 8); in xnn_f32_dwconv_minmax_ukernel_up8x4__wasmsimd_arm() local
70 vacc0123p0 = wasm_f32x4_add(vacc0123p0, wasm_f32x4_mul(vi0x0123, vk0x0123)); in xnn_f32_dwconv_minmax_ukernel_up8x4__wasmsimd_arm()
119 const v128_t vk0x0123 = wasm_v128_load(w + 8); in xnn_f32_dwconv_minmax_ukernel_up8x4__wasmsimd_arm() local
120 vacc0123p0 = wasm_f32x4_add(vacc0123p0, wasm_f32x4_mul(vi0x0123, vk0x0123)); in xnn_f32_dwconv_minmax_ukernel_up8x4__wasmsimd_arm()
153 const v128_t vk0x0123 = wasm_v128_load(w + 8); in xnn_f32_dwconv_minmax_ukernel_up8x4__wasmsimd_arm() local
154 vacc0123p0 = wasm_f32x4_add(vacc0123p0, wasm_f32x4_mul(vi0x0123, vk0x0123)); in xnn_f32_dwconv_minmax_ukernel_up8x4__wasmsimd_arm()
Dup8x4-minmax-wasmsimd-x86.c68 const v128_t vk0x0123 = wasm_v128_load(w + 8); in xnn_f32_dwconv_minmax_ukernel_up8x4__wasmsimd_x86() local
70 vacc0123p0 = wasm_f32x4_add(vacc0123p0, wasm_f32x4_mul(vi0x0123, vk0x0123)); in xnn_f32_dwconv_minmax_ukernel_up8x4__wasmsimd_x86()
119 const v128_t vk0x0123 = wasm_v128_load(w + 8); in xnn_f32_dwconv_minmax_ukernel_up8x4__wasmsimd_x86() local
120 vacc0123p0 = wasm_f32x4_add(vacc0123p0, wasm_f32x4_mul(vi0x0123, vk0x0123)); in xnn_f32_dwconv_minmax_ukernel_up8x4__wasmsimd_x86()
153 const v128_t vk0x0123 = wasm_v128_load(w + 8); in xnn_f32_dwconv_minmax_ukernel_up8x4__wasmsimd_x86() local
154 vacc0123p0 = wasm_f32x4_add(vacc0123p0, wasm_f32x4_mul(vi0x0123, vk0x0123)); in xnn_f32_dwconv_minmax_ukernel_up8x4__wasmsimd_x86()
Dup8x4-minmax-sse.c68 const __m128 vk0x0123 = _mm_load_ps(w + 8); in xnn_f32_dwconv_minmax_ukernel_up8x4__sse() local
70 vacc0123p0 = _mm_add_ps(vacc0123p0, _mm_mul_ps(vi0x0123, vk0x0123)); in xnn_f32_dwconv_minmax_ukernel_up8x4__sse()
118 const __m128 vk0x0123 = _mm_load_ps(w + 8); in xnn_f32_dwconv_minmax_ukernel_up8x4__sse() local
119 vacc0123p0 = _mm_add_ps(vacc0123p0, _mm_mul_ps(vi0x0123, vk0x0123)); in xnn_f32_dwconv_minmax_ukernel_up8x4__sse()
152 const __m128 vk0x0123 = _mm_load_ps(w + 8); in xnn_f32_dwconv_minmax_ukernel_up8x4__sse() local
153 vacc0123p0 = _mm_add_ps(vacc0123p0, _mm_mul_ps(vi0x0123, vk0x0123)); in xnn_f32_dwconv_minmax_ukernel_up8x4__sse()
Dup8x4-minmax-wasmsimd-x86-acc2.c68 const v128_t vk0x0123 = wasm_v128_load(w + 8); in xnn_f32_dwconv_minmax_ukernel_up8x4__wasmsimd_x86_acc2() local
70 vacc0123p0 = wasm_f32x4_add(vacc0123p0, wasm_f32x4_mul(vi0x0123, vk0x0123)); in xnn_f32_dwconv_minmax_ukernel_up8x4__wasmsimd_x86_acc2()
122 const v128_t vk0x0123 = wasm_v128_load(w + 8); in xnn_f32_dwconv_minmax_ukernel_up8x4__wasmsimd_x86_acc2() local
123 vacc0123p0 = wasm_f32x4_add(vacc0123p0, wasm_f32x4_mul(vi0x0123, vk0x0123)); in xnn_f32_dwconv_minmax_ukernel_up8x4__wasmsimd_x86_acc2()
158 const v128_t vk0x0123 = wasm_v128_load(w + 8); in xnn_f32_dwconv_minmax_ukernel_up8x4__wasmsimd_x86_acc2() local
159 vacc0123p0 = wasm_f32x4_add(vacc0123p0, wasm_f32x4_mul(vi0x0123, vk0x0123)); in xnn_f32_dwconv_minmax_ukernel_up8x4__wasmsimd_x86_acc2()
Dup4x4-wasmsimd.c64 const v128_t vk0x0123 = wasm_v128_load(w + 4); in xnn_f32_dwconv_ukernel_up4x4__wasmsimd() local
65 vacc0123p0 = wasm_f32x4_add(vacc0123p0, wasm_f32x4_mul(vi0x0123, vk0x0123)); in xnn_f32_dwconv_ukernel_up4x4__wasmsimd()
97 const v128_t vk0x0123 = wasm_v128_load(w + 4); in xnn_f32_dwconv_ukernel_up4x4__wasmsimd() local
98 vacc0123p0 = wasm_f32x4_add(vacc0123p0, wasm_f32x4_mul(vi0x0123, vk0x0123)); in xnn_f32_dwconv_ukernel_up4x4__wasmsimd()
Dup4x4-minmax-sse.c66 const __m128 vk0x0123 = _mm_load_ps(w + 4); in xnn_f32_dwconv_minmax_ukernel_up4x4__sse() local
67 vacc0123p0 = _mm_add_ps(vacc0123p0, _mm_mul_ps(vi0x0123, vk0x0123)); in xnn_f32_dwconv_minmax_ukernel_up4x4__sse()
100 const __m128 vk0x0123 = _mm_load_ps(w + 4); in xnn_f32_dwconv_minmax_ukernel_up4x4__sse() local
101 vacc0123p0 = _mm_add_ps(vacc0123p0, _mm_mul_ps(vi0x0123, vk0x0123)); in xnn_f32_dwconv_minmax_ukernel_up4x4__sse()
Dup4x4-minmax-wasmsimd-arm.c66 const v128_t vk0x0123 = wasm_v128_load(w + 4); in xnn_f32_dwconv_minmax_ukernel_up4x4__wasmsimd_arm() local
67 vacc0123p0 = wasm_f32x4_add(vacc0123p0, wasm_f32x4_mul(vi0x0123, vk0x0123)); in xnn_f32_dwconv_minmax_ukernel_up4x4__wasmsimd_arm()
101 const v128_t vk0x0123 = wasm_v128_load(w + 4); in xnn_f32_dwconv_minmax_ukernel_up4x4__wasmsimd_arm() local
102 vacc0123p0 = wasm_f32x4_add(vacc0123p0, wasm_f32x4_mul(vi0x0123, vk0x0123)); in xnn_f32_dwconv_minmax_ukernel_up4x4__wasmsimd_arm()
Dup4x4-minmax-wasmsimd-arm-acc2.c66 const v128_t vk0x0123 = wasm_v128_load(w + 4); in xnn_f32_dwconv_minmax_ukernel_up4x4__wasmsimd_arm_acc2() local
67 vacc0123p0 = wasm_f32x4_add(vacc0123p0, wasm_f32x4_mul(vi0x0123, vk0x0123)); in xnn_f32_dwconv_minmax_ukernel_up4x4__wasmsimd_arm_acc2()
103 const v128_t vk0x0123 = wasm_v128_load(w + 4); in xnn_f32_dwconv_minmax_ukernel_up4x4__wasmsimd_arm_acc2() local
104 vacc0123p0 = wasm_f32x4_add(vacc0123p0, wasm_f32x4_mul(vi0x0123, vk0x0123)); in xnn_f32_dwconv_minmax_ukernel_up4x4__wasmsimd_arm_acc2()
Dup4x4-minmax-sse-acc2.c66 const __m128 vk0x0123 = _mm_load_ps(w + 4); in xnn_f32_dwconv_minmax_ukernel_up4x4__sse_acc2() local
67 vacc0123p0 = _mm_add_ps(vacc0123p0, _mm_mul_ps(vi0x0123, vk0x0123)); in xnn_f32_dwconv_minmax_ukernel_up4x4__sse_acc2()
102 const __m128 vk0x0123 = _mm_load_ps(w + 4); in xnn_f32_dwconv_minmax_ukernel_up4x4__sse_acc2() local
103 vacc0123p0 = _mm_add_ps(vacc0123p0, _mm_mul_ps(vi0x0123, vk0x0123)); in xnn_f32_dwconv_minmax_ukernel_up4x4__sse_acc2()
Dup4x4-minmax-neon.c65 const float32x4_t vk0x0123 = vld1q_f32(w); w += 4; in xnn_f32_dwconv_minmax_ukernel_up4x4__neon() local
66 vacc0123p0 = vmlaq_f32(vacc0123p0, vi0x0123, vk0x0123); in xnn_f32_dwconv_minmax_ukernel_up4x4__neon()
91 const float32x4_t vk0x0123 = vld1q_f32(w); w += 4; in xnn_f32_dwconv_minmax_ukernel_up4x4__neon() local
92 vacc0123p0 = vmlaq_f32(vacc0123p0, vi0x0123, vk0x0123); in xnn_f32_dwconv_minmax_ukernel_up4x4__neon()
Dup4x4-minmax-neonfma-acc2.c65 const float32x4_t vk0x0123 = vld1q_f32(w); w += 4; in xnn_f32_dwconv_minmax_ukernel_up4x4__neonfma_acc2() local
66 vacc0123p0 = vfmaq_f32(vacc0123p0, vi0x0123, vk0x0123); in xnn_f32_dwconv_minmax_ukernel_up4x4__neonfma_acc2()
93 const float32x4_t vk0x0123 = vld1q_f32(w); w += 4; in xnn_f32_dwconv_minmax_ukernel_up4x4__neonfma_acc2() local
94 vacc0123p0 = vfmaq_f32(vacc0123p0, vi0x0123, vk0x0123); in xnn_f32_dwconv_minmax_ukernel_up4x4__neonfma_acc2()
Dup4x4-minmax-wasmsimd-x86.c66 const v128_t vk0x0123 = wasm_v128_load(w + 4); in xnn_f32_dwconv_minmax_ukernel_up4x4__wasmsimd_x86() local
67 vacc0123p0 = wasm_f32x4_add(vacc0123p0, wasm_f32x4_mul(vi0x0123, vk0x0123)); in xnn_f32_dwconv_minmax_ukernel_up4x4__wasmsimd_x86()
101 const v128_t vk0x0123 = wasm_v128_load(w + 4); in xnn_f32_dwconv_minmax_ukernel_up4x4__wasmsimd_x86() local
102 vacc0123p0 = wasm_f32x4_add(vacc0123p0, wasm_f32x4_mul(vi0x0123, vk0x0123)); in xnn_f32_dwconv_minmax_ukernel_up4x4__wasmsimd_x86()
Dup4x4-minmax-neonfma.c65 const float32x4_t vk0x0123 = vld1q_f32(w); w += 4; in xnn_f32_dwconv_minmax_ukernel_up4x4__neonfma() local
66 vacc0123p0 = vfmaq_f32(vacc0123p0, vi0x0123, vk0x0123); in xnn_f32_dwconv_minmax_ukernel_up4x4__neonfma()
91 const float32x4_t vk0x0123 = vld1q_f32(w); w += 4; in xnn_f32_dwconv_minmax_ukernel_up4x4__neonfma() local
92 vacc0123p0 = vfmaq_f32(vacc0123p0, vi0x0123, vk0x0123); in xnn_f32_dwconv_minmax_ukernel_up4x4__neonfma()
Dup4x4-minmax-neon-acc2.c65 const float32x4_t vk0x0123 = vld1q_f32(w); w += 4; in xnn_f32_dwconv_minmax_ukernel_up4x4__neon_acc2() local
66 vacc0123p0 = vmlaq_f32(vacc0123p0, vi0x0123, vk0x0123); in xnn_f32_dwconv_minmax_ukernel_up4x4__neon_acc2()
93 const float32x4_t vk0x0123 = vld1q_f32(w); w += 4; in xnn_f32_dwconv_minmax_ukernel_up4x4__neon_acc2() local
94 vacc0123p0 = vmlaq_f32(vacc0123p0, vi0x0123, vk0x0123); in xnn_f32_dwconv_minmax_ukernel_up4x4__neon_acc2()
Dup4x4-minmax-wasmsimd-x86-acc2.c66 const v128_t vk0x0123 = wasm_v128_load(w + 4); in xnn_f32_dwconv_minmax_ukernel_up4x4__wasmsimd_x86_acc2() local
67 vacc0123p0 = wasm_f32x4_add(vacc0123p0, wasm_f32x4_mul(vi0x0123, vk0x0123)); in xnn_f32_dwconv_minmax_ukernel_up4x4__wasmsimd_x86_acc2()
103 const v128_t vk0x0123 = wasm_v128_load(w + 4); in xnn_f32_dwconv_minmax_ukernel_up4x4__wasmsimd_x86_acc2() local
104 vacc0123p0 = wasm_f32x4_add(vacc0123p0, wasm_f32x4_mul(vi0x0123, vk0x0123)); in xnn_f32_dwconv_minmax_ukernel_up4x4__wasmsimd_x86_acc2()
Dup8x9-minmax-sse.c93 const __m128 vk0x0123 = _mm_load_ps(w + 8); in xnn_f32_dwconv_minmax_ukernel_up8x9__sse() local
95 vacc0123p0 = _mm_add_ps(vacc0123p0, _mm_mul_ps(vi0x0123, vk0x0123)); in xnn_f32_dwconv_minmax_ukernel_up8x9__sse()
188 const __m128 vk0x0123 = _mm_load_ps(w + 8); in xnn_f32_dwconv_minmax_ukernel_up8x9__sse() local
189 vacc0123p0 = _mm_add_ps(vacc0123p0, _mm_mul_ps(vi0x0123, vk0x0123)); in xnn_f32_dwconv_minmax_ukernel_up8x9__sse()
252 const __m128 vk0x0123 = _mm_load_ps(w + 8); in xnn_f32_dwconv_minmax_ukernel_up8x9__sse() local
253 vacc0123p0 = _mm_add_ps(vacc0123p0, _mm_mul_ps(vi0x0123, vk0x0123)); in xnn_f32_dwconv_minmax_ukernel_up8x9__sse()
Dup8x9-wasmsimd.c91 const v128_t vk0x0123 = wasm_v128_load(w + 8); in xnn_f32_dwconv_ukernel_up8x9__wasmsimd() local
93 vacc0123p0 = wasm_f32x4_add(vacc0123p0, wasm_f32x4_mul(vi0x0123, vk0x0123)); in xnn_f32_dwconv_ukernel_up8x9__wasmsimd()
184 const v128_t vk0x0123 = wasm_v128_load(w + 8); in xnn_f32_dwconv_ukernel_up8x9__wasmsimd() local
185 vacc0123p0 = wasm_f32x4_add(vacc0123p0, wasm_f32x4_mul(vi0x0123, vk0x0123)); in xnn_f32_dwconv_ukernel_up8x9__wasmsimd()
247 const v128_t vk0x0123 = wasm_v128_load(w + 8); in xnn_f32_dwconv_ukernel_up8x9__wasmsimd() local
248 vacc0123p0 = wasm_f32x4_add(vacc0123p0, wasm_f32x4_mul(vi0x0123, vk0x0123)); in xnn_f32_dwconv_ukernel_up8x9__wasmsimd()
Dup8x9-minmax-neon.c92 const float32x4_t vk0x0123 = vld1q_f32(w); w += 4; in xnn_f32_dwconv_minmax_ukernel_up8x9__neon() local
94 vacc0123p0 = vmlaq_f32(vacc0123p0, vi0x0123, vk0x0123); in xnn_f32_dwconv_minmax_ukernel_up8x9__neon()
167 const float32x4_t vk0x0123 = vld1q_f32(w + 4); in xnn_f32_dwconv_minmax_ukernel_up8x9__neon() local
168 vacc0123p0 = vmlaq_f32(vacc0123p0, vi0x0123, vk0x0123); in xnn_f32_dwconv_minmax_ukernel_up8x9__neon()
213 const float32x4_t vk0x0123 = vld1q_f32(w + 8); in xnn_f32_dwconv_minmax_ukernel_up8x9__neon() local
214 vacc0123p0 = vmlaq_f32(vacc0123p0, vi0x0123, vk0x0123); in xnn_f32_dwconv_minmax_ukernel_up8x9__neon()

123