/external/XNNPACK/src/f32-dwconv2d-chw/gen/ |
D | 3x3s2p1-minmax-wasmsimd-arm-loadsplat-1x4.c | 86 const v128_t vi1xCDEF = wasm_v128_load(i1 + 4); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_loadsplat_1x4() local 94 const v128_t vi1x8ACE = wasm_v32x4_shuffle(vi1x89AB, vi1xCDEF, 0, 2, 4, 6); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_loadsplat_1x4() 95 const v128_t vi1x9BDF = wasm_v32x4_shuffle(vi1x89AB, vi1xCDEF, 1, 3, 5, 7); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_loadsplat_1x4() 138 const v128_t vi1xCDEF = wasm_v128_load(i1 + 4); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_loadsplat_1x4() local 144 …const v128_t vi1x8ACE = wasm_v128_and(vmask_even, wasm_v32x4_shuffle(vi1x89AB, vi1xCDEF, 0, 2, 4, … in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_loadsplat_1x4() 145 …const v128_t vi1x9BDF = wasm_v128_and(vmask_odd, wasm_v32x4_shuffle(vi1x89AB, vi1xCDEF, 1, 3, 5, … in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_loadsplat_1x4()
|
D | 3x3s2p1-minmax-wasmsimd-arm-loadsplat-1x4-acc2.c | 86 const v128_t vi1xCDEF = wasm_v128_load(i1 + 4); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_loadsplat_1x4_acc2() local 94 const v128_t vi1x8ACE = wasm_v32x4_shuffle(vi1x89AB, vi1xCDEF, 0, 2, 4, 6); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_loadsplat_1x4_acc2() 95 const v128_t vi1x9BDF = wasm_v32x4_shuffle(vi1x89AB, vi1xCDEF, 1, 3, 5, 7); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_loadsplat_1x4_acc2() 139 const v128_t vi1xCDEF = wasm_v128_load(i1 + 4); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_loadsplat_1x4_acc2() local 145 …const v128_t vi1x8ACE = wasm_v128_and(vmask_even, wasm_v32x4_shuffle(vi1x89AB, vi1xCDEF, 0, 2, 4, … in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_loadsplat_1x4_acc2() 146 …const v128_t vi1x9BDF = wasm_v128_and(vmask_odd, wasm_v32x4_shuffle(vi1x89AB, vi1xCDEF, 1, 3, 5, … in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_loadsplat_1x4_acc2()
|
D | 3x3s2p1-minmax-wasmsimd-x86-splat-1x4.c | 76 const v128_t vi1xCDEF = wasm_v128_load(i1 + 4); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_splat_1x4() local 84 const v128_t vi1x8ACE = wasm_v32x4_shuffle(vi1x89AB, vi1xCDEF, 0, 2, 4, 6); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_splat_1x4() 85 const v128_t vi1x9BDF = wasm_v32x4_shuffle(vi1x89AB, vi1xCDEF, 1, 3, 5, 7); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_splat_1x4() 128 const v128_t vi1xCDEF = wasm_v128_load(i1 + 4); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_splat_1x4() local 134 …const v128_t vi1x8ACE = wasm_v128_and(vmask_even, wasm_v32x4_shuffle(vi1x89AB, vi1xCDEF, 0, 2, 4, … in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_splat_1x4() 135 …const v128_t vi1x9BDF = wasm_v128_and(vmask_odd, wasm_v32x4_shuffle(vi1x89AB, vi1xCDEF, 1, 3, 5, … in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_splat_1x4()
|
D | 3x3s2p1-minmax-wasmsimd-arm-splat-1x4.c | 76 const v128_t vi1xCDEF = wasm_v128_load(i1 + 4); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_splat_1x4() local 84 const v128_t vi1x8ACE = wasm_v32x4_shuffle(vi1x89AB, vi1xCDEF, 0, 2, 4, 6); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_splat_1x4() 85 const v128_t vi1x9BDF = wasm_v32x4_shuffle(vi1x89AB, vi1xCDEF, 1, 3, 5, 7); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_splat_1x4() 128 const v128_t vi1xCDEF = wasm_v128_load(i1 + 4); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_splat_1x4() local 134 …const v128_t vi1x8ACE = wasm_v128_and(vmask_even, wasm_v32x4_shuffle(vi1x89AB, vi1xCDEF, 0, 2, 4, … in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_splat_1x4() 135 …const v128_t vi1x9BDF = wasm_v128_and(vmask_odd, wasm_v32x4_shuffle(vi1x89AB, vi1xCDEF, 1, 3, 5, … in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_splat_1x4()
|
D | 3x3s2p1-minmax-wasmsimd-arm-splat-1x4-acc3.c | 76 const v128_t vi1xCDEF = wasm_v128_load(i1 + 4); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_splat_1x4_acc3() local 84 const v128_t vi1x8ACE = wasm_v32x4_shuffle(vi1x89AB, vi1xCDEF, 0, 2, 4, 6); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_splat_1x4_acc3() 85 const v128_t vi1x9BDF = wasm_v32x4_shuffle(vi1x89AB, vi1xCDEF, 1, 3, 5, 7); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_splat_1x4_acc3() 130 const v128_t vi1xCDEF = wasm_v128_load(i1 + 4); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_splat_1x4_acc3() local 136 …const v128_t vi1x8ACE = wasm_v128_and(vmask_even, wasm_v32x4_shuffle(vi1x89AB, vi1xCDEF, 0, 2, 4, … in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_splat_1x4_acc3() 137 …const v128_t vi1x9BDF = wasm_v128_and(vmask_odd, wasm_v32x4_shuffle(vi1x89AB, vi1xCDEF, 1, 3, 5, … in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_splat_1x4_acc3()
|
D | 3x3s2p1-minmax-wasmsimd-arm-splat-1x4-acc2.c | 76 const v128_t vi1xCDEF = wasm_v128_load(i1 + 4); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_splat_1x4_acc2() local 84 const v128_t vi1x8ACE = wasm_v32x4_shuffle(vi1x89AB, vi1xCDEF, 0, 2, 4, 6); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_splat_1x4_acc2() 85 const v128_t vi1x9BDF = wasm_v32x4_shuffle(vi1x89AB, vi1xCDEF, 1, 3, 5, 7); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_splat_1x4_acc2() 129 const v128_t vi1xCDEF = wasm_v128_load(i1 + 4); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_splat_1x4_acc2() local 135 …const v128_t vi1x8ACE = wasm_v128_and(vmask_even, wasm_v32x4_shuffle(vi1x89AB, vi1xCDEF, 0, 2, 4, … in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_splat_1x4_acc2() 136 …const v128_t vi1x9BDF = wasm_v128_and(vmask_odd, wasm_v32x4_shuffle(vi1x89AB, vi1xCDEF, 1, 3, 5, … in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_splat_1x4_acc2()
|
D | 3x3s2p1-minmax-wasmsimd-x86-splat-1x4-acc2.c | 76 const v128_t vi1xCDEF = wasm_v128_load(i1 + 4); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_splat_1x4_acc2() local 84 const v128_t vi1x8ACE = wasm_v32x4_shuffle(vi1x89AB, vi1xCDEF, 0, 2, 4, 6); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_splat_1x4_acc2() 85 const v128_t vi1x9BDF = wasm_v32x4_shuffle(vi1x89AB, vi1xCDEF, 1, 3, 5, 7); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_splat_1x4_acc2() 129 const v128_t vi1xCDEF = wasm_v128_load(i1 + 4); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_splat_1x4_acc2() local 135 …const v128_t vi1x8ACE = wasm_v128_and(vmask_even, wasm_v32x4_shuffle(vi1x89AB, vi1xCDEF, 0, 2, 4, … in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_splat_1x4_acc2() 136 …const v128_t vi1x9BDF = wasm_v128_and(vmask_odd, wasm_v32x4_shuffle(vi1x89AB, vi1xCDEF, 1, 3, 5, … in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_splat_1x4_acc2()
|
D | 3x3s2p1-minmax-wasmsimd-arm-splat-1x4-acc4.c | 76 const v128_t vi1xCDEF = wasm_v128_load(i1 + 4); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_splat_1x4_acc4() local 84 const v128_t vi1x8ACE = wasm_v32x4_shuffle(vi1x89AB, vi1xCDEF, 0, 2, 4, 6); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_splat_1x4_acc4() 85 const v128_t vi1x9BDF = wasm_v32x4_shuffle(vi1x89AB, vi1xCDEF, 1, 3, 5, 7); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_splat_1x4_acc4() 131 const v128_t vi1xCDEF = wasm_v128_load(i1 + 4); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_splat_1x4_acc4() local 137 …const v128_t vi1x8ACE = wasm_v128_and(vmask_even, wasm_v32x4_shuffle(vi1x89AB, vi1xCDEF, 0, 2, 4, … in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_splat_1x4_acc4() 138 …const v128_t vi1x9BDF = wasm_v128_and(vmask_odd, wasm_v32x4_shuffle(vi1x89AB, vi1xCDEF, 1, 3, 5, … in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_splat_1x4_acc4()
|
D | 3x3s2p1-minmax-wasmsimd-arm-loadsplat-1x4-acc4.c | 86 const v128_t vi1xCDEF = wasm_v128_load(i1 + 4); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_loadsplat_1x4_acc4() local 94 const v128_t vi1x8ACE = wasm_v32x4_shuffle(vi1x89AB, vi1xCDEF, 0, 2, 4, 6); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_loadsplat_1x4_acc4() 95 const v128_t vi1x9BDF = wasm_v32x4_shuffle(vi1x89AB, vi1xCDEF, 1, 3, 5, 7); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_loadsplat_1x4_acc4() 141 const v128_t vi1xCDEF = wasm_v128_load(i1 + 4); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_loadsplat_1x4_acc4() local 147 …const v128_t vi1x8ACE = wasm_v128_and(vmask_even, wasm_v32x4_shuffle(vi1x89AB, vi1xCDEF, 0, 2, 4, … in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_loadsplat_1x4_acc4() 148 …const v128_t vi1x9BDF = wasm_v128_and(vmask_odd, wasm_v32x4_shuffle(vi1x89AB, vi1xCDEF, 1, 3, 5, … in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_loadsplat_1x4_acc4()
|
D | 3x3s2p1-minmax-wasmsimd-x86-loadsplat-1x4-acc3.c | 86 const v128_t vi1xCDEF = wasm_v128_load(i1 + 4); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_1x4_acc3() local 94 const v128_t vi1x8ACE = wasm_v32x4_shuffle(vi1x89AB, vi1xCDEF, 0, 2, 4, 6); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_1x4_acc3() 95 const v128_t vi1x9BDF = wasm_v32x4_shuffle(vi1x89AB, vi1xCDEF, 1, 3, 5, 7); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_1x4_acc3() 140 const v128_t vi1xCDEF = wasm_v128_load(i1 + 4); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_1x4_acc3() local 146 …const v128_t vi1x8ACE = wasm_v128_and(vmask_even, wasm_v32x4_shuffle(vi1x89AB, vi1xCDEF, 0, 2, 4, … in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_1x4_acc3() 147 …const v128_t vi1x9BDF = wasm_v128_and(vmask_odd, wasm_v32x4_shuffle(vi1x89AB, vi1xCDEF, 1, 3, 5, … in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_1x4_acc3()
|
D | 3x3s2p1-minmax-wasmsimd-x86-loadsplat-1x4-acc2.c | 86 const v128_t vi1xCDEF = wasm_v128_load(i1 + 4); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_1x4_acc2() local 94 const v128_t vi1x8ACE = wasm_v32x4_shuffle(vi1x89AB, vi1xCDEF, 0, 2, 4, 6); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_1x4_acc2() 95 const v128_t vi1x9BDF = wasm_v32x4_shuffle(vi1x89AB, vi1xCDEF, 1, 3, 5, 7); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_1x4_acc2() 139 const v128_t vi1xCDEF = wasm_v128_load(i1 + 4); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_1x4_acc2() local 145 …const v128_t vi1x8ACE = wasm_v128_and(vmask_even, wasm_v32x4_shuffle(vi1x89AB, vi1xCDEF, 0, 2, 4, … in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_1x4_acc2() 146 …const v128_t vi1x9BDF = wasm_v128_and(vmask_odd, wasm_v32x4_shuffle(vi1x89AB, vi1xCDEF, 1, 3, 5, … in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_1x4_acc2()
|
D | 3x3s2p1-minmax-wasmsimd-x86-splat-1x4-acc4.c | 76 const v128_t vi1xCDEF = wasm_v128_load(i1 + 4); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_splat_1x4_acc4() local 84 const v128_t vi1x8ACE = wasm_v32x4_shuffle(vi1x89AB, vi1xCDEF, 0, 2, 4, 6); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_splat_1x4_acc4() 85 const v128_t vi1x9BDF = wasm_v32x4_shuffle(vi1x89AB, vi1xCDEF, 1, 3, 5, 7); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_splat_1x4_acc4() 131 const v128_t vi1xCDEF = wasm_v128_load(i1 + 4); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_splat_1x4_acc4() local 137 …const v128_t vi1x8ACE = wasm_v128_and(vmask_even, wasm_v32x4_shuffle(vi1x89AB, vi1xCDEF, 0, 2, 4, … in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_splat_1x4_acc4() 138 …const v128_t vi1x9BDF = wasm_v128_and(vmask_odd, wasm_v32x4_shuffle(vi1x89AB, vi1xCDEF, 1, 3, 5, … in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_splat_1x4_acc4()
|
D | 3x3s2p1-minmax-sse-1x4.c | 78 const __m128 vi1xCDEF = _mm_loadu_ps(i1 + 4); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_1x4() local 86 const __m128 vi1x8ACE = _mm_shuffle_ps(vi1x89AB, vi1xCDEF, _MM_SHUFFLE(2, 0, 2, 0)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_1x4() 87 const __m128 vi1x9BDF = _mm_shuffle_ps(vi1x89AB, vi1xCDEF, _MM_SHUFFLE(3, 1, 3, 1)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_1x4() 129 const __m128 vi1xCDEF = _mm_loadu_ps(i1 + 4); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_1x4() local 135 …const __m128 vi1x8ACE = _mm_and_ps(vmask_even, _mm_shuffle_ps(vi1x89AB, vi1xCDEF, _MM_SHUFFLE(2, 0… in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_1x4() 136 …const __m128 vi1x9BDF = _mm_and_ps(vmask_odd, _mm_shuffle_ps(vi1x89AB, vi1xCDEF, _MM_SHUFFLE(3, 1… in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_1x4()
|
D | 3x3s2p1-minmax-sse-1x4-acc3.c | 78 const __m128 vi1xCDEF = _mm_loadu_ps(i1 + 4); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_1x4_acc3() local 86 const __m128 vi1x8ACE = _mm_shuffle_ps(vi1x89AB, vi1xCDEF, _MM_SHUFFLE(2, 0, 2, 0)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_1x4_acc3() 87 const __m128 vi1x9BDF = _mm_shuffle_ps(vi1x89AB, vi1xCDEF, _MM_SHUFFLE(3, 1, 3, 1)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_1x4_acc3() 131 const __m128 vi1xCDEF = _mm_loadu_ps(i1 + 4); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_1x4_acc3() local 137 …const __m128 vi1x8ACE = _mm_and_ps(vmask_even, _mm_shuffle_ps(vi1x89AB, vi1xCDEF, _MM_SHUFFLE(2, 0… in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_1x4_acc3() 138 …const __m128 vi1x9BDF = _mm_and_ps(vmask_odd, _mm_shuffle_ps(vi1x89AB, vi1xCDEF, _MM_SHUFFLE(3, 1… in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_1x4_acc3()
|
D | 3x3s2p1-minmax-wasmsimd-x86-splat-1x4-acc3.c | 76 const v128_t vi1xCDEF = wasm_v128_load(i1 + 4); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_splat_1x4_acc3() local 84 const v128_t vi1x8ACE = wasm_v32x4_shuffle(vi1x89AB, vi1xCDEF, 0, 2, 4, 6); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_splat_1x4_acc3() 85 const v128_t vi1x9BDF = wasm_v32x4_shuffle(vi1x89AB, vi1xCDEF, 1, 3, 5, 7); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_splat_1x4_acc3() 130 const v128_t vi1xCDEF = wasm_v128_load(i1 + 4); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_splat_1x4_acc3() local 136 …const v128_t vi1x8ACE = wasm_v128_and(vmask_even, wasm_v32x4_shuffle(vi1x89AB, vi1xCDEF, 0, 2, 4, … in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_splat_1x4_acc3() 137 …const v128_t vi1x9BDF = wasm_v128_and(vmask_odd, wasm_v32x4_shuffle(vi1x89AB, vi1xCDEF, 1, 3, 5, … in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_splat_1x4_acc3()
|
D | 3x3s2p1-minmax-wasmsimd-x86-loadsplat-1x4.c | 86 const v128_t vi1xCDEF = wasm_v128_load(i1 + 4); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_1x4() local 94 const v128_t vi1x8ACE = wasm_v32x4_shuffle(vi1x89AB, vi1xCDEF, 0, 2, 4, 6); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_1x4() 95 const v128_t vi1x9BDF = wasm_v32x4_shuffle(vi1x89AB, vi1xCDEF, 1, 3, 5, 7); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_1x4() 138 const v128_t vi1xCDEF = wasm_v128_load(i1 + 4); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_1x4() local 144 …const v128_t vi1x8ACE = wasm_v128_and(vmask_even, wasm_v32x4_shuffle(vi1x89AB, vi1xCDEF, 0, 2, 4, … in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_1x4() 145 …const v128_t vi1x9BDF = wasm_v128_and(vmask_odd, wasm_v32x4_shuffle(vi1x89AB, vi1xCDEF, 1, 3, 5, … in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_1x4()
|
D | 3x3s2p1-minmax-wasmsimd-x86-loadsplat-1x4-acc4.c | 86 const v128_t vi1xCDEF = wasm_v128_load(i1 + 4); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_1x4_acc4() local 94 const v128_t vi1x8ACE = wasm_v32x4_shuffle(vi1x89AB, vi1xCDEF, 0, 2, 4, 6); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_1x4_acc4() 95 const v128_t vi1x9BDF = wasm_v32x4_shuffle(vi1x89AB, vi1xCDEF, 1, 3, 5, 7); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_1x4_acc4() 141 const v128_t vi1xCDEF = wasm_v128_load(i1 + 4); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_1x4_acc4() local 147 …const v128_t vi1x8ACE = wasm_v128_and(vmask_even, wasm_v32x4_shuffle(vi1x89AB, vi1xCDEF, 0, 2, 4, … in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_1x4_acc4() 148 …const v128_t vi1x9BDF = wasm_v128_and(vmask_odd, wasm_v32x4_shuffle(vi1x89AB, vi1xCDEF, 1, 3, 5, … in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_1x4_acc4()
|
D | 3x3s2p1-minmax-sse-1x4-acc2.c | 78 const __m128 vi1xCDEF = _mm_loadu_ps(i1 + 4); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_1x4_acc2() local 86 const __m128 vi1x8ACE = _mm_shuffle_ps(vi1x89AB, vi1xCDEF, _MM_SHUFFLE(2, 0, 2, 0)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_1x4_acc2() 87 const __m128 vi1x9BDF = _mm_shuffle_ps(vi1x89AB, vi1xCDEF, _MM_SHUFFLE(3, 1, 3, 1)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_1x4_acc2() 130 const __m128 vi1xCDEF = _mm_loadu_ps(i1 + 4); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_1x4_acc2() local 136 …const __m128 vi1x8ACE = _mm_and_ps(vmask_even, _mm_shuffle_ps(vi1x89AB, vi1xCDEF, _MM_SHUFFLE(2, 0… in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_1x4_acc2() 137 …const __m128 vi1x9BDF = _mm_and_ps(vmask_odd, _mm_shuffle_ps(vi1x89AB, vi1xCDEF, _MM_SHUFFLE(3, 1… in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_1x4_acc2()
|
D | 3x3s2p1-minmax-wasmsimd-arm-loadsplat-1x4-acc3.c | 86 const v128_t vi1xCDEF = wasm_v128_load(i1 + 4); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_loadsplat_1x4_acc3() local 94 const v128_t vi1x8ACE = wasm_v32x4_shuffle(vi1x89AB, vi1xCDEF, 0, 2, 4, 6); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_loadsplat_1x4_acc3() 95 const v128_t vi1x9BDF = wasm_v32x4_shuffle(vi1x89AB, vi1xCDEF, 1, 3, 5, 7); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_loadsplat_1x4_acc3() 140 const v128_t vi1xCDEF = wasm_v128_load(i1 + 4); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_loadsplat_1x4_acc3() local 146 …const v128_t vi1x8ACE = wasm_v128_and(vmask_even, wasm_v32x4_shuffle(vi1x89AB, vi1xCDEF, 0, 2, 4, … in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_loadsplat_1x4_acc3() 147 …const v128_t vi1x9BDF = wasm_v128_and(vmask_odd, wasm_v32x4_shuffle(vi1x89AB, vi1xCDEF, 1, 3, 5, … in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_loadsplat_1x4_acc3()
|
D | 3x3s2p1-minmax-sse-1x4-acc4.c | 78 const __m128 vi1xCDEF = _mm_loadu_ps(i1 + 4); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_1x4_acc4() local 86 const __m128 vi1x8ACE = _mm_shuffle_ps(vi1x89AB, vi1xCDEF, _MM_SHUFFLE(2, 0, 2, 0)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_1x4_acc4() 87 const __m128 vi1x9BDF = _mm_shuffle_ps(vi1x89AB, vi1xCDEF, _MM_SHUFFLE(3, 1, 3, 1)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_1x4_acc4() 132 const __m128 vi1xCDEF = _mm_loadu_ps(i1 + 4); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_1x4_acc4() local 138 …const __m128 vi1x8ACE = _mm_and_ps(vmask_even, _mm_shuffle_ps(vi1x89AB, vi1xCDEF, _MM_SHUFFLE(2, 0… in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_1x4_acc4() 139 …const __m128 vi1x9BDF = _mm_and_ps(vmask_odd, _mm_shuffle_ps(vi1x89AB, vi1xCDEF, _MM_SHUFFLE(3, 1… in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_1x4_acc4()
|
D | 3x3s2p1-minmax-wasmsimd-x86-loadsplat-2x4.c | 100 const v128_t vi1xCDEF = wasm_v128_load(i1 + 4); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_2x4() local 114 const v128_t vi1x8ACE = wasm_v32x4_shuffle(vi1x89AB, vi1xCDEF, 0, 2, 4, 6); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_2x4() 115 const v128_t vi1x9BDF = wasm_v32x4_shuffle(vi1x89AB, vi1xCDEF, 1, 3, 5, 7); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_2x4() 179 const v128_t vi1xCDEF = wasm_v128_load(i1 + 4); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_2x4() local 189 …const v128_t vi1x8ACE = wasm_v128_and(vmask_even, wasm_v32x4_shuffle(vi1x89AB, vi1xCDEF, 0, 2, 4, … in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_2x4() 190 …const v128_t vi1x9BDF = wasm_v128_and(vmask_odd, wasm_v32x4_shuffle(vi1x89AB, vi1xCDEF, 1, 3, 5, … in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_2x4()
|
D | 3x3s2p1-minmax-wasmsimd-x86-loadsplat-2x4-acc2.c | 100 const v128_t vi1xCDEF = wasm_v128_load(i1 + 4); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_2x4_acc2() local 114 const v128_t vi1x8ACE = wasm_v32x4_shuffle(vi1x89AB, vi1xCDEF, 0, 2, 4, 6); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_2x4_acc2() 115 const v128_t vi1x9BDF = wasm_v32x4_shuffle(vi1x89AB, vi1xCDEF, 1, 3, 5, 7); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_2x4_acc2() 181 const v128_t vi1xCDEF = wasm_v128_load(i1 + 4); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_2x4_acc2() local 191 …const v128_t vi1x8ACE = wasm_v128_and(vmask_even, wasm_v32x4_shuffle(vi1x89AB, vi1xCDEF, 0, 2, 4, … in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_2x4_acc2() 192 …const v128_t vi1x9BDF = wasm_v128_and(vmask_odd, wasm_v32x4_shuffle(vi1x89AB, vi1xCDEF, 1, 3, 5, … in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_2x4_acc2()
|
/external/XNNPACK/src/f32-prelu/gen/ |
D | wasmsimd-minmax-2x16.c | 62 v128_t vi1xCDEF = wasm_v128_load(i1 + 12); in xnn_f32_prelu_ukernel__wasmsimd_minmax_2x16() local 79 v128_t vacc1xCDEF = wasm_i32x4_max(vi1xCDEF, vzero); in xnn_f32_prelu_ukernel__wasmsimd_minmax_2x16() 80 vi1xCDEF = wasm_i32x4_min(vi1xCDEF, vzero); in xnn_f32_prelu_ukernel__wasmsimd_minmax_2x16() 89 vacc1xCDEF = wasm_f32x4_add(vacc1xCDEF, wasm_f32x4_mul(vi1xCDEF, vwCDEF)); in xnn_f32_prelu_ukernel__wasmsimd_minmax_2x16()
|
D | wasmsimd-bitselect-2x16.c | 62 const v128_t vi1xCDEF = wasm_v128_load(i1 + 12); in xnn_f32_prelu_ukernel__wasmsimd_bitselect_2x16() local 79 v128_t vacc1xCDEF = wasm_f32x4_mul(vi1xCDEF, vwCDEF); in xnn_f32_prelu_ukernel__wasmsimd_bitselect_2x16() 80 const v128_t vmask1xCDEF = wasm_i32x4_lt(vi1xCDEF, vzero); in xnn_f32_prelu_ukernel__wasmsimd_bitselect_2x16() 89 vacc1xCDEF = wasm_v128_bitselect(vacc1xCDEF, vi1xCDEF, vmask1xCDEF); in xnn_f32_prelu_ukernel__wasmsimd_bitselect_2x16()
|
D | neon-2x16.c | 59 const float32x4_t vi1xCDEF = vld1q_f32(i1); i1 += 4; in xnn_f32_prelu_ukernel__neon_2x16() local 75 float32x4_t vacc1xCDEF = vmulq_f32(vi1xCDEF, vwCDEF); in xnn_f32_prelu_ukernel__neon_2x16() 76 const uint32x4_t vm1xCDEF = vcltq_s32(vreinterpretq_s32_f32(vi1xCDEF), vmovq_n_s32(0)); in xnn_f32_prelu_ukernel__neon_2x16() 85 vacc1xCDEF = vbslq_f32(vm1xCDEF, vacc1xCDEF, vi1xCDEF); in xnn_f32_prelu_ukernel__neon_2x16()
|