/external/XNNPACK/src/f32-dwconv2d-chw/gen/ |
D | 3x3s2p1-minmax-sse-1x4.c | 89 const __m128 vi2x9BDF = _mm_shuffle_ps(vi2x89AB, vi2xCDEF, _MM_SHUFFLE(3, 1, 3, 1)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_1x4() local 97 const __m128 vi2xF9BD = _mm_shuffle_ps(vi2x9BDF, vi2x9BDF, _MM_SHUFFLE(2, 1, 0, 3)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_1x4() 101 vo0p0 = _mm_add_ps(vo0p0, _mm_mul_ps(vi2x9BDF, vk22)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_1x4() 138 …const __m128 vi2x9BDF = _mm_and_ps(vmask_odd, _mm_shuffle_ps(vi2x89AB, vi2xCDEF, _MM_SHUFFLE(3, 1… in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_1x4() local 146 const __m128 vi2xF9BD = _mm_shuffle_ps(vi2x9BDF, vi2x9BDF, _MM_SHUFFLE(2, 1, 0, 3)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_1x4() 150 vo0p0 = _mm_add_ps(vo0p0, _mm_mul_ps(vi2x9BDF, vk22)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_1x4()
|
D | 3x3s2p1-minmax-sse-1x4-acc3.c | 89 const __m128 vi2x9BDF = _mm_shuffle_ps(vi2x89AB, vi2xCDEF, _MM_SHUFFLE(3, 1, 3, 1)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_1x4_acc3() local 97 const __m128 vi2xF9BD = _mm_shuffle_ps(vi2x9BDF, vi2x9BDF, _MM_SHUFFLE(2, 1, 0, 3)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_1x4_acc3() 101 vo0p2 = _mm_add_ps(vo0p2, _mm_mul_ps(vi2x9BDF, vk22)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_1x4_acc3() 140 …const __m128 vi2x9BDF = _mm_and_ps(vmask_odd, _mm_shuffle_ps(vi2x89AB, vi2xCDEF, _MM_SHUFFLE(3, 1… in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_1x4_acc3() local 148 const __m128 vi2xF9BD = _mm_shuffle_ps(vi2x9BDF, vi2x9BDF, _MM_SHUFFLE(2, 1, 0, 3)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_1x4_acc3() 152 vo0p2 = _mm_add_ps(vo0p2, _mm_mul_ps(vi2x9BDF, vk22)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_1x4_acc3()
|
D | 3x3s2p1-minmax-sse-1x4-acc2.c | 89 const __m128 vi2x9BDF = _mm_shuffle_ps(vi2x89AB, vi2xCDEF, _MM_SHUFFLE(3, 1, 3, 1)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_1x4_acc2() local 97 const __m128 vi2xF9BD = _mm_shuffle_ps(vi2x9BDF, vi2x9BDF, _MM_SHUFFLE(2, 1, 0, 3)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_1x4_acc2() 101 vo0p1 = _mm_add_ps(vo0p1, _mm_mul_ps(vi2x9BDF, vk22)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_1x4_acc2() 139 …const __m128 vi2x9BDF = _mm_and_ps(vmask_odd, _mm_shuffle_ps(vi2x89AB, vi2xCDEF, _MM_SHUFFLE(3, 1… in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_1x4_acc2() local 147 const __m128 vi2xF9BD = _mm_shuffle_ps(vi2x9BDF, vi2x9BDF, _MM_SHUFFLE(2, 1, 0, 3)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_1x4_acc2() 151 vo0p1 = _mm_add_ps(vo0p1, _mm_mul_ps(vi2x9BDF, vk22)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_1x4_acc2()
|
D | 3x3s2p1-minmax-sse-1x4-acc4.c | 89 const __m128 vi2x9BDF = _mm_shuffle_ps(vi2x89AB, vi2xCDEF, _MM_SHUFFLE(3, 1, 3, 1)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_1x4_acc4() local 97 const __m128 vi2xF9BD = _mm_shuffle_ps(vi2x9BDF, vi2x9BDF, _MM_SHUFFLE(2, 1, 0, 3)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_1x4_acc4() 101 vo0p1 = _mm_add_ps(vo0p1, _mm_mul_ps(vi2x9BDF, vk22)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_1x4_acc4() 141 …const __m128 vi2x9BDF = _mm_and_ps(vmask_odd, _mm_shuffle_ps(vi2x89AB, vi2xCDEF, _MM_SHUFFLE(3, 1… in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_1x4_acc4() local 149 const __m128 vi2xF9BD = _mm_shuffle_ps(vi2x9BDF, vi2x9BDF, _MM_SHUFFLE(2, 1, 0, 3)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_1x4_acc4() 153 vo0p1 = _mm_add_ps(vo0p1, _mm_mul_ps(vi2x9BDF, vk22)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_1x4_acc4()
|
D | 3x3s2p1-minmax-wasmsimd-arm-loadsplat-1x4.c | 97 const v128_t vi2x9BDF = wasm_v32x4_shuffle(vi2x89AB, vi2xCDEF, 1, 3, 5, 7); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_loadsplat_1x4() local 109 const v128_t vi2x7BDF = wasm_v32x4_shuffle(vi2x1357, vi2x9BDF, 3, 4, 5, 6); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_loadsplat_1x4() 110 vi2x1357 = vi2x9BDF; in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_loadsplat_1x4() 122 vo0p0 = wasm_f32x4_add(vo0p0, wasm_f32x4_mul(vi2x9BDF, vk22)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_loadsplat_1x4() 147 …const v128_t vi2x9BDF = wasm_v128_and(vmask_odd, wasm_v32x4_shuffle(vi2x89AB, vi2xCDEF, 1, 3, 5, … in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_loadsplat_1x4() local 157 const v128_t vi2x7BDF = wasm_v32x4_shuffle(vi2x1357, vi2x9BDF, 3, 4, 5, 6); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_loadsplat_1x4() 169 vo0p0 = wasm_f32x4_add(vo0p0, wasm_f32x4_mul(vi2x9BDF, vk22)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_loadsplat_1x4()
|
D | 3x3s2p1-minmax-wasmsimd-arm-loadsplat-1x4-acc2.c | 97 const v128_t vi2x9BDF = wasm_v32x4_shuffle(vi2x89AB, vi2xCDEF, 1, 3, 5, 7); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_loadsplat_1x4_acc2() local 109 const v128_t vi2x7BDF = wasm_v32x4_shuffle(vi2x1357, vi2x9BDF, 3, 4, 5, 6); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_loadsplat_1x4_acc2() 110 vi2x1357 = vi2x9BDF; in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_loadsplat_1x4_acc2() 122 vo0p0 = wasm_f32x4_add(vo0p0, wasm_f32x4_mul(vi2x9BDF, vk22)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_loadsplat_1x4_acc2() 148 …const v128_t vi2x9BDF = wasm_v128_and(vmask_odd, wasm_v32x4_shuffle(vi2x89AB, vi2xCDEF, 1, 3, 5, … in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_loadsplat_1x4_acc2() local 158 const v128_t vi2x7BDF = wasm_v32x4_shuffle(vi2x1357, vi2x9BDF, 3, 4, 5, 6); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_loadsplat_1x4_acc2() 170 vo0p0 = wasm_f32x4_add(vo0p0, wasm_f32x4_mul(vi2x9BDF, vk22)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_loadsplat_1x4_acc2()
|
D | 3x3s2p1-minmax-wasmsimd-x86-splat-1x4.c | 87 const v128_t vi2x9BDF = wasm_v32x4_shuffle(vi2x89AB, vi2xCDEF, 1, 3, 5, 7); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_splat_1x4() local 99 const v128_t vi2x7BDF = wasm_v32x4_shuffle(vi2x1357, vi2x9BDF, 3, 4, 5, 6); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_splat_1x4() 100 vi2x1357 = vi2x9BDF; in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_splat_1x4() 112 …vo0p0 = wasm_f32x4_add(vo0p0, wasm_f32x4_mul(vi2x9BDF, wasm_v32x4_shuffle(vw89, vw89, 1, 1, 1, 1))… in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_splat_1x4() 137 …const v128_t vi2x9BDF = wasm_v128_and(vmask_odd, wasm_v32x4_shuffle(vi2x89AB, vi2xCDEF, 1, 3, 5, … in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_splat_1x4() local 147 const v128_t vi2x7BDF = wasm_v32x4_shuffle(vi2x1357, vi2x9BDF, 3, 4, 5, 6); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_splat_1x4() 159 …vo0p0 = wasm_f32x4_add(vo0p0, wasm_f32x4_mul(vi2x9BDF, wasm_v32x4_shuffle(vw89, vw89, 1, 1, 1, 1))… in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_splat_1x4()
|
D | 3x3s2p1-minmax-wasmsimd-arm-splat-1x4.c | 87 const v128_t vi2x9BDF = wasm_v32x4_shuffle(vi2x89AB, vi2xCDEF, 1, 3, 5, 7); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_splat_1x4() local 99 const v128_t vi2x7BDF = wasm_v32x4_shuffle(vi2x1357, vi2x9BDF, 3, 4, 5, 6); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_splat_1x4() 100 vi2x1357 = vi2x9BDF; in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_splat_1x4() 112 …vo0p0 = wasm_f32x4_add(vo0p0, wasm_f32x4_mul(vi2x9BDF, wasm_v32x4_shuffle(vw89, vw89, 1, 1, 1, 1))… in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_splat_1x4() 137 …const v128_t vi2x9BDF = wasm_v128_and(vmask_odd, wasm_v32x4_shuffle(vi2x89AB, vi2xCDEF, 1, 3, 5, … in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_splat_1x4() local 147 const v128_t vi2x7BDF = wasm_v32x4_shuffle(vi2x1357, vi2x9BDF, 3, 4, 5, 6); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_splat_1x4() 159 …vo0p0 = wasm_f32x4_add(vo0p0, wasm_f32x4_mul(vi2x9BDF, wasm_v32x4_shuffle(vw89, vw89, 1, 1, 1, 1))… in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_splat_1x4()
|
D | 3x3s2p1-minmax-wasmsimd-arm-splat-1x4-acc3.c | 87 const v128_t vi2x9BDF = wasm_v32x4_shuffle(vi2x89AB, vi2xCDEF, 1, 3, 5, 7); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_splat_1x4_acc3() local 99 const v128_t vi2x7BDF = wasm_v32x4_shuffle(vi2x1357, vi2x9BDF, 3, 4, 5, 6); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_splat_1x4_acc3() 100 vi2x1357 = vi2x9BDF; in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_splat_1x4_acc3() 112 …vo0p1 = wasm_f32x4_add(vo0p1, wasm_f32x4_mul(vi2x9BDF, wasm_v32x4_shuffle(vw89, vw89, 1, 1, 1, 1))… in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_splat_1x4_acc3() 139 …const v128_t vi2x9BDF = wasm_v128_and(vmask_odd, wasm_v32x4_shuffle(vi2x89AB, vi2xCDEF, 1, 3, 5, … in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_splat_1x4_acc3() local 149 const v128_t vi2x7BDF = wasm_v32x4_shuffle(vi2x1357, vi2x9BDF, 3, 4, 5, 6); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_splat_1x4_acc3() 161 …vo0p1 = wasm_f32x4_add(vo0p1, wasm_f32x4_mul(vi2x9BDF, wasm_v32x4_shuffle(vw89, vw89, 1, 1, 1, 1))… in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_splat_1x4_acc3()
|
D | 3x3s2p1-minmax-wasmsimd-arm-splat-1x4-acc2.c | 87 const v128_t vi2x9BDF = wasm_v32x4_shuffle(vi2x89AB, vi2xCDEF, 1, 3, 5, 7); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_splat_1x4_acc2() local 99 const v128_t vi2x7BDF = wasm_v32x4_shuffle(vi2x1357, vi2x9BDF, 3, 4, 5, 6); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_splat_1x4_acc2() 100 vi2x1357 = vi2x9BDF; in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_splat_1x4_acc2() 112 …vo0p0 = wasm_f32x4_add(vo0p0, wasm_f32x4_mul(vi2x9BDF, wasm_v32x4_shuffle(vw89, vw89, 1, 1, 1, 1))… in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_splat_1x4_acc2() 138 …const v128_t vi2x9BDF = wasm_v128_and(vmask_odd, wasm_v32x4_shuffle(vi2x89AB, vi2xCDEF, 1, 3, 5, … in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_splat_1x4_acc2() local 148 const v128_t vi2x7BDF = wasm_v32x4_shuffle(vi2x1357, vi2x9BDF, 3, 4, 5, 6); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_splat_1x4_acc2() 160 …vo0p0 = wasm_f32x4_add(vo0p0, wasm_f32x4_mul(vi2x9BDF, wasm_v32x4_shuffle(vw89, vw89, 1, 1, 1, 1))… in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_splat_1x4_acc2()
|
D | 3x3s2p1-minmax-wasmsimd-x86-splat-1x4-acc2.c | 87 const v128_t vi2x9BDF = wasm_v32x4_shuffle(vi2x89AB, vi2xCDEF, 1, 3, 5, 7); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_splat_1x4_acc2() local 99 const v128_t vi2x7BDF = wasm_v32x4_shuffle(vi2x1357, vi2x9BDF, 3, 4, 5, 6); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_splat_1x4_acc2() 100 vi2x1357 = vi2x9BDF; in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_splat_1x4_acc2() 112 …vo0p0 = wasm_f32x4_add(vo0p0, wasm_f32x4_mul(vi2x9BDF, wasm_v32x4_shuffle(vw89, vw89, 1, 1, 1, 1))… in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_splat_1x4_acc2() 138 …const v128_t vi2x9BDF = wasm_v128_and(vmask_odd, wasm_v32x4_shuffle(vi2x89AB, vi2xCDEF, 1, 3, 5, … in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_splat_1x4_acc2() local 148 const v128_t vi2x7BDF = wasm_v32x4_shuffle(vi2x1357, vi2x9BDF, 3, 4, 5, 6); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_splat_1x4_acc2() 160 …vo0p0 = wasm_f32x4_add(vo0p0, wasm_f32x4_mul(vi2x9BDF, wasm_v32x4_shuffle(vw89, vw89, 1, 1, 1, 1))… in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_splat_1x4_acc2()
|
D | 3x3s2p1-minmax-wasmsimd-arm-splat-1x4-acc4.c | 87 const v128_t vi2x9BDF = wasm_v32x4_shuffle(vi2x89AB, vi2xCDEF, 1, 3, 5, 7); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_splat_1x4_acc4() local 99 const v128_t vi2x7BDF = wasm_v32x4_shuffle(vi2x1357, vi2x9BDF, 3, 4, 5, 6); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_splat_1x4_acc4() 100 vi2x1357 = vi2x9BDF; in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_splat_1x4_acc4() 112 …vo0p2 = wasm_f32x4_add(vo0p2, wasm_f32x4_mul(vi2x9BDF, wasm_v32x4_shuffle(vw89, vw89, 1, 1, 1, 1))… in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_splat_1x4_acc4() 140 …const v128_t vi2x9BDF = wasm_v128_and(vmask_odd, wasm_v32x4_shuffle(vi2x89AB, vi2xCDEF, 1, 3, 5, … in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_splat_1x4_acc4() local 150 const v128_t vi2x7BDF = wasm_v32x4_shuffle(vi2x1357, vi2x9BDF, 3, 4, 5, 6); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_splat_1x4_acc4() 162 …vo0p2 = wasm_f32x4_add(vo0p2, wasm_f32x4_mul(vi2x9BDF, wasm_v32x4_shuffle(vw89, vw89, 1, 1, 1, 1))… in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_splat_1x4_acc4()
|
D | 3x3s2p1-minmax-wasmsimd-arm-loadsplat-1x4-acc4.c | 97 const v128_t vi2x9BDF = wasm_v32x4_shuffle(vi2x89AB, vi2xCDEF, 1, 3, 5, 7); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_loadsplat_1x4_acc4() local 109 const v128_t vi2x7BDF = wasm_v32x4_shuffle(vi2x1357, vi2x9BDF, 3, 4, 5, 6); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_loadsplat_1x4_acc4() 110 vi2x1357 = vi2x9BDF; in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_loadsplat_1x4_acc4() 122 vo0p2 = wasm_f32x4_add(vo0p2, wasm_f32x4_mul(vi2x9BDF, vk22)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_loadsplat_1x4_acc4() 150 …const v128_t vi2x9BDF = wasm_v128_and(vmask_odd, wasm_v32x4_shuffle(vi2x89AB, vi2xCDEF, 1, 3, 5, … in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_loadsplat_1x4_acc4() local 160 const v128_t vi2x7BDF = wasm_v32x4_shuffle(vi2x1357, vi2x9BDF, 3, 4, 5, 6); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_loadsplat_1x4_acc4() 172 vo0p2 = wasm_f32x4_add(vo0p2, wasm_f32x4_mul(vi2x9BDF, vk22)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_loadsplat_1x4_acc4()
|
D | 3x3s2p1-minmax-wasmsimd-x86-loadsplat-1x4-acc3.c | 97 const v128_t vi2x9BDF = wasm_v32x4_shuffle(vi2x89AB, vi2xCDEF, 1, 3, 5, 7); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_1x4_acc3() local 109 const v128_t vi2x7BDF = wasm_v32x4_shuffle(vi2x1357, vi2x9BDF, 3, 4, 5, 6); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_1x4_acc3() 110 vi2x1357 = vi2x9BDF; in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_1x4_acc3() 122 vo0p1 = wasm_f32x4_add(vo0p1, wasm_f32x4_mul(vi2x9BDF, vk22)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_1x4_acc3() 149 …const v128_t vi2x9BDF = wasm_v128_and(vmask_odd, wasm_v32x4_shuffle(vi2x89AB, vi2xCDEF, 1, 3, 5, … in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_1x4_acc3() local 159 const v128_t vi2x7BDF = wasm_v32x4_shuffle(vi2x1357, vi2x9BDF, 3, 4, 5, 6); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_1x4_acc3() 171 vo0p1 = wasm_f32x4_add(vo0p1, wasm_f32x4_mul(vi2x9BDF, vk22)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_1x4_acc3()
|
D | 3x3s2p1-minmax-wasmsimd-x86-loadsplat-1x4-acc2.c | 97 const v128_t vi2x9BDF = wasm_v32x4_shuffle(vi2x89AB, vi2xCDEF, 1, 3, 5, 7); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_1x4_acc2() local 109 const v128_t vi2x7BDF = wasm_v32x4_shuffle(vi2x1357, vi2x9BDF, 3, 4, 5, 6); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_1x4_acc2() 110 vi2x1357 = vi2x9BDF; in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_1x4_acc2() 122 vo0p0 = wasm_f32x4_add(vo0p0, wasm_f32x4_mul(vi2x9BDF, vk22)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_1x4_acc2() 148 …const v128_t vi2x9BDF = wasm_v128_and(vmask_odd, wasm_v32x4_shuffle(vi2x89AB, vi2xCDEF, 1, 3, 5, … in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_1x4_acc2() local 158 const v128_t vi2x7BDF = wasm_v32x4_shuffle(vi2x1357, vi2x9BDF, 3, 4, 5, 6); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_1x4_acc2() 170 vo0p0 = wasm_f32x4_add(vo0p0, wasm_f32x4_mul(vi2x9BDF, vk22)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_1x4_acc2()
|
D | 3x3s2p1-minmax-wasmsimd-x86-splat-1x4-acc4.c | 87 const v128_t vi2x9BDF = wasm_v32x4_shuffle(vi2x89AB, vi2xCDEF, 1, 3, 5, 7); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_splat_1x4_acc4() local 99 const v128_t vi2x7BDF = wasm_v32x4_shuffle(vi2x1357, vi2x9BDF, 3, 4, 5, 6); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_splat_1x4_acc4() 100 vi2x1357 = vi2x9BDF; in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_splat_1x4_acc4() 112 …vo0p2 = wasm_f32x4_add(vo0p2, wasm_f32x4_mul(vi2x9BDF, wasm_v32x4_shuffle(vw89, vw89, 1, 1, 1, 1))… in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_splat_1x4_acc4() 140 …const v128_t vi2x9BDF = wasm_v128_and(vmask_odd, wasm_v32x4_shuffle(vi2x89AB, vi2xCDEF, 1, 3, 5, … in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_splat_1x4_acc4() local 150 const v128_t vi2x7BDF = wasm_v32x4_shuffle(vi2x1357, vi2x9BDF, 3, 4, 5, 6); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_splat_1x4_acc4() 162 …vo0p2 = wasm_f32x4_add(vo0p2, wasm_f32x4_mul(vi2x9BDF, wasm_v32x4_shuffle(vw89, vw89, 1, 1, 1, 1))… in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_splat_1x4_acc4()
|
D | 3x3s2p1-minmax-wasmsimd-x86-splat-1x4-acc3.c | 87 const v128_t vi2x9BDF = wasm_v32x4_shuffle(vi2x89AB, vi2xCDEF, 1, 3, 5, 7); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_splat_1x4_acc3() local 99 const v128_t vi2x7BDF = wasm_v32x4_shuffle(vi2x1357, vi2x9BDF, 3, 4, 5, 6); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_splat_1x4_acc3() 100 vi2x1357 = vi2x9BDF; in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_splat_1x4_acc3() 112 …vo0p1 = wasm_f32x4_add(vo0p1, wasm_f32x4_mul(vi2x9BDF, wasm_v32x4_shuffle(vw89, vw89, 1, 1, 1, 1))… in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_splat_1x4_acc3() 139 …const v128_t vi2x9BDF = wasm_v128_and(vmask_odd, wasm_v32x4_shuffle(vi2x89AB, vi2xCDEF, 1, 3, 5, … in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_splat_1x4_acc3() local 149 const v128_t vi2x7BDF = wasm_v32x4_shuffle(vi2x1357, vi2x9BDF, 3, 4, 5, 6); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_splat_1x4_acc3() 161 …vo0p1 = wasm_f32x4_add(vo0p1, wasm_f32x4_mul(vi2x9BDF, wasm_v32x4_shuffle(vw89, vw89, 1, 1, 1, 1))… in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_splat_1x4_acc3()
|
D | 3x3s2p1-minmax-wasmsimd-x86-loadsplat-1x4.c | 97 const v128_t vi2x9BDF = wasm_v32x4_shuffle(vi2x89AB, vi2xCDEF, 1, 3, 5, 7); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_1x4() local 109 const v128_t vi2x7BDF = wasm_v32x4_shuffle(vi2x1357, vi2x9BDF, 3, 4, 5, 6); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_1x4() 110 vi2x1357 = vi2x9BDF; in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_1x4() 122 vo0p0 = wasm_f32x4_add(vo0p0, wasm_f32x4_mul(vi2x9BDF, vk22)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_1x4() 147 …const v128_t vi2x9BDF = wasm_v128_and(vmask_odd, wasm_v32x4_shuffle(vi2x89AB, vi2xCDEF, 1, 3, 5, … in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_1x4() local 157 const v128_t vi2x7BDF = wasm_v32x4_shuffle(vi2x1357, vi2x9BDF, 3, 4, 5, 6); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_1x4() 169 vo0p0 = wasm_f32x4_add(vo0p0, wasm_f32x4_mul(vi2x9BDF, vk22)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_1x4()
|
D | 3x3s2p1-minmax-wasmsimd-x86-loadsplat-1x4-acc4.c | 97 const v128_t vi2x9BDF = wasm_v32x4_shuffle(vi2x89AB, vi2xCDEF, 1, 3, 5, 7); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_1x4_acc4() local 109 const v128_t vi2x7BDF = wasm_v32x4_shuffle(vi2x1357, vi2x9BDF, 3, 4, 5, 6); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_1x4_acc4() 110 vi2x1357 = vi2x9BDF; in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_1x4_acc4() 122 vo0p2 = wasm_f32x4_add(vo0p2, wasm_f32x4_mul(vi2x9BDF, vk22)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_1x4_acc4() 150 …const v128_t vi2x9BDF = wasm_v128_and(vmask_odd, wasm_v32x4_shuffle(vi2x89AB, vi2xCDEF, 1, 3, 5, … in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_1x4_acc4() local 160 const v128_t vi2x7BDF = wasm_v32x4_shuffle(vi2x1357, vi2x9BDF, 3, 4, 5, 6); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_1x4_acc4() 172 vo0p2 = wasm_f32x4_add(vo0p2, wasm_f32x4_mul(vi2x9BDF, vk22)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_1x4_acc4()
|
D | 3x3s2p1-minmax-wasmsimd-arm-loadsplat-1x4-acc3.c | 97 const v128_t vi2x9BDF = wasm_v32x4_shuffle(vi2x89AB, vi2xCDEF, 1, 3, 5, 7); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_loadsplat_1x4_acc3() local 109 const v128_t vi2x7BDF = wasm_v32x4_shuffle(vi2x1357, vi2x9BDF, 3, 4, 5, 6); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_loadsplat_1x4_acc3() 110 vi2x1357 = vi2x9BDF; in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_loadsplat_1x4_acc3() 122 vo0p1 = wasm_f32x4_add(vo0p1, wasm_f32x4_mul(vi2x9BDF, vk22)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_loadsplat_1x4_acc3() 149 …const v128_t vi2x9BDF = wasm_v128_and(vmask_odd, wasm_v32x4_shuffle(vi2x89AB, vi2xCDEF, 1, 3, 5, … in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_loadsplat_1x4_acc3() local 159 const v128_t vi2x7BDF = wasm_v32x4_shuffle(vi2x1357, vi2x9BDF, 3, 4, 5, 6); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_loadsplat_1x4_acc3() 171 vo0p1 = wasm_f32x4_add(vo0p1, wasm_f32x4_mul(vi2x9BDF, vk22)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_loadsplat_1x4_acc3()
|
D | 3x3s2p1-minmax-sse-2x4-acc2.c | 108 const __m128 vi2x9BDF = _mm_shuffle_ps(vi2x89AB, vi2xCDEF, _MM_SHUFFLE(3, 1, 3, 1)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_2x4_acc2() local 123 const __m128 vi2xF9BD = _mm_shuffle_ps(vi2x9BDF, vi2x9BDF, _MM_SHUFFLE(2, 1, 0, 3)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_2x4_acc2() 128 vo1p1 = _mm_add_ps(vo1p1, _mm_mul_ps(vi2x9BDF, vk02)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_2x4_acc2() 131 vo0p1 = _mm_add_ps(vo0p1, _mm_mul_ps(vi2x9BDF, vk22)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_2x4_acc2() 186 …const __m128 vi2x9BDF = _mm_and_ps(vmask_odd, _mm_shuffle_ps(vi2x89AB, vi2xCDEF, _MM_SHUFFLE(3, 1… in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_2x4_acc2() local 201 const __m128 vi2xF9BD = _mm_shuffle_ps(vi2x9BDF, vi2x9BDF, _MM_SHUFFLE(2, 1, 0, 3)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_2x4_acc2() 206 vo1p1 = _mm_add_ps(vo1p1, _mm_mul_ps(vi2x9BDF, vk02)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_2x4_acc2() 209 vo0p1 = _mm_add_ps(vo0p1, _mm_mul_ps(vi2x9BDF, vk22)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_2x4_acc2()
|
D | 3x3s2p1-minmax-sse-2x4.c | 108 const __m128 vi2x9BDF = _mm_shuffle_ps(vi2x89AB, vi2xCDEF, _MM_SHUFFLE(3, 1, 3, 1)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_2x4() local 123 const __m128 vi2xF9BD = _mm_shuffle_ps(vi2x9BDF, vi2x9BDF, _MM_SHUFFLE(2, 1, 0, 3)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_2x4() 128 vo1p0 = _mm_add_ps(vo1p0, _mm_mul_ps(vi2x9BDF, vk02)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_2x4() 131 vo0p0 = _mm_add_ps(vo0p0, _mm_mul_ps(vi2x9BDF, vk22)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_2x4() 184 …const __m128 vi2x9BDF = _mm_and_ps(vmask_odd, _mm_shuffle_ps(vi2x89AB, vi2xCDEF, _MM_SHUFFLE(3, 1… in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_2x4() local 199 const __m128 vi2xF9BD = _mm_shuffle_ps(vi2x9BDF, vi2x9BDF, _MM_SHUFFLE(2, 1, 0, 3)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_2x4() 204 vo1p0 = _mm_add_ps(vo1p0, _mm_mul_ps(vi2x9BDF, vk02)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_2x4() 207 vo0p0 = _mm_add_ps(vo0p0, _mm_mul_ps(vi2x9BDF, vk22)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_2x4()
|
D | 3x3s2p1-minmax-wasmsimd-x86-loadsplat-2x4.c | 117 const v128_t vi2x9BDF = wasm_v32x4_shuffle(vi2x89AB, vi2xCDEF, 1, 3, 5, 7); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_2x4() local 136 const v128_t vi2x7BDF = wasm_v32x4_shuffle(vi2x1357, vi2x9BDF, 3, 4, 5, 6); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_2x4() 137 vi2x1357 = vi2x9BDF; in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_2x4() 153 vo1p0 = wasm_f32x4_add(vo1p0, wasm_f32x4_mul(vi2x9BDF, vk02)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_2x4() 158 vo0p0 = wasm_f32x4_add(vo0p0, wasm_f32x4_mul(vi2x9BDF, vk22)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_2x4() 192 …const v128_t vi2x9BDF = wasm_v128_and(vmask_odd, wasm_v32x4_shuffle(vi2x89AB, vi2xCDEF, 1, 3, 5, … in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_2x4() local 209 const v128_t vi2x7BDF = wasm_v32x4_shuffle(vi2x1357, vi2x9BDF, 3, 4, 5, 6); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_2x4() 223 vo1p0 = wasm_f32x4_add(vo1p0, wasm_f32x4_mul(vi2x9BDF, vk02)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_2x4() 228 vo0p0 = wasm_f32x4_add(vo0p0, wasm_f32x4_mul(vi2x9BDF, vk22)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_2x4()
|
D | 3x3s2p1-minmax-wasmsimd-x86-loadsplat-2x4-acc2.c | 117 const v128_t vi2x9BDF = wasm_v32x4_shuffle(vi2x89AB, vi2xCDEF, 1, 3, 5, 7); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_2x4_acc2() local 136 const v128_t vi2x7BDF = wasm_v32x4_shuffle(vi2x1357, vi2x9BDF, 3, 4, 5, 6); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_2x4_acc2() 137 vi2x1357 = vi2x9BDF; in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_2x4_acc2() 153 vo1p0 = wasm_f32x4_add(vo1p0, wasm_f32x4_mul(vi2x9BDF, vk02)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_2x4_acc2() 158 vo0p0 = wasm_f32x4_add(vo0p0, wasm_f32x4_mul(vi2x9BDF, vk22)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_2x4_acc2() 194 …const v128_t vi2x9BDF = wasm_v128_and(vmask_odd, wasm_v32x4_shuffle(vi2x89AB, vi2xCDEF, 1, 3, 5, … in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_2x4_acc2() local 211 const v128_t vi2x7BDF = wasm_v32x4_shuffle(vi2x1357, vi2x9BDF, 3, 4, 5, 6); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_2x4_acc2() 225 vo1p0 = wasm_f32x4_add(vo1p0, wasm_f32x4_mul(vi2x9BDF, vk02)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_2x4_acc2() 230 vo0p0 = wasm_f32x4_add(vo0p0, wasm_f32x4_mul(vi2x9BDF, vk22)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_2x4_acc2()
|
D | 3x3s2p1-minmax-wasmsimd-arm-loadsplat-2x4-acc2.c | 117 const v128_t vi2x9BDF = wasm_v32x4_shuffle(vi2x89AB, vi2xCDEF, 1, 3, 5, 7); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_loadsplat_2x4_acc2() local 136 const v128_t vi2x7BDF = wasm_v32x4_shuffle(vi2x1357, vi2x9BDF, 3, 4, 5, 6); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_loadsplat_2x4_acc2() 137 vi2x1357 = vi2x9BDF; in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_loadsplat_2x4_acc2() 153 vo1p0 = wasm_f32x4_add(vo1p0, wasm_f32x4_mul(vi2x9BDF, vk02)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_loadsplat_2x4_acc2() 158 vo0p0 = wasm_f32x4_add(vo0p0, wasm_f32x4_mul(vi2x9BDF, vk22)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_loadsplat_2x4_acc2() 194 …const v128_t vi2x9BDF = wasm_v128_and(vmask_odd, wasm_v32x4_shuffle(vi2x89AB, vi2xCDEF, 1, 3, 5, … in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_loadsplat_2x4_acc2() local 211 const v128_t vi2x7BDF = wasm_v32x4_shuffle(vi2x1357, vi2x9BDF, 3, 4, 5, 6); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_loadsplat_2x4_acc2() 225 vo1p0 = wasm_f32x4_add(vo1p0, wasm_f32x4_mul(vi2x9BDF, vk02)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_loadsplat_2x4_acc2() 230 vo0p0 = wasm_f32x4_add(vo0p0, wasm_f32x4_mul(vi2x9BDF, vk22)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_loadsplat_2x4_acc2()
|