| /external/XNNPACK/src/f32-dwconv2d-chw/gen/ |
| D | 3x3s2p1-minmax-wasmsimd-arm-loadsplat-4x4.c | 143 const v128_t vi7x89AB = wasm_v128_load(i7); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_loadsplat_4x4() local
164 const v128_t vi7x8ACE = wasm_v32x4_shuffle(vi7x89AB, vi7xCDEF, 0, 2, 4, 6); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_loadsplat_4x4()
165 const v128_t vi7x9BDF = wasm_v32x4_shuffle(vi7x89AB, vi7xCDEF, 1, 3, 5, 7); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_loadsplat_4x4()
270 const v128_t vi7x89AB = wasm_v128_load(i7); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_loadsplat_4x4() local
289 …const v128_t vi7x8ACE = wasm_v128_and(vmask_even, wasm_v32x4_shuffle(vi7x89AB, vi7xCDEF, 0, 2, 4, … in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_loadsplat_4x4()
290 …const v128_t vi7x9BDF = wasm_v128_and(vmask_odd, wasm_v32x4_shuffle(vi7x89AB, vi7xCDEF, 1, 3, 5, … in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_loadsplat_4x4()
|
| D | 3x3s2p1-minmax-wasmsimd-x86-loadsplat-4x4.c | 143 const v128_t vi7x89AB = wasm_v128_load(i7); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_4x4() local
164 const v128_t vi7x8ACE = wasm_v32x4_shuffle(vi7x89AB, vi7xCDEF, 0, 2, 4, 6); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_4x4()
165 const v128_t vi7x9BDF = wasm_v32x4_shuffle(vi7x89AB, vi7xCDEF, 1, 3, 5, 7); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_4x4()
270 const v128_t vi7x89AB = wasm_v128_load(i7); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_4x4() local
289 …const v128_t vi7x8ACE = wasm_v128_and(vmask_even, wasm_v32x4_shuffle(vi7x89AB, vi7xCDEF, 0, 2, 4, … in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_4x4()
290 …const v128_t vi7x9BDF = wasm_v128_and(vmask_odd, wasm_v32x4_shuffle(vi7x89AB, vi7xCDEF, 1, 3, 5, … in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_loadsplat_4x4()
|
| D | 3x3s2p1-minmax-wasmsimd-arm-splat-4x4.c | 133 const v128_t vi7x89AB = wasm_v128_load(i7); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_splat_4x4() local
154 const v128_t vi7x8ACE = wasm_v32x4_shuffle(vi7x89AB, vi7xCDEF, 0, 2, 4, 6); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_splat_4x4()
155 const v128_t vi7x9BDF = wasm_v32x4_shuffle(vi7x89AB, vi7xCDEF, 1, 3, 5, 7); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_splat_4x4()
260 const v128_t vi7x89AB = wasm_v128_load(i7); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_splat_4x4() local
279 …const v128_t vi7x8ACE = wasm_v128_and(vmask_even, wasm_v32x4_shuffle(vi7x89AB, vi7xCDEF, 0, 2, 4, … in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_splat_4x4()
280 …const v128_t vi7x9BDF = wasm_v128_and(vmask_odd, wasm_v32x4_shuffle(vi7x89AB, vi7xCDEF, 1, 3, 5, … in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_arm_splat_4x4()
|
| D | 3x3s2p1-minmax-wasmsimd-x86-splat-4x4.c | 133 const v128_t vi7x89AB = wasm_v128_load(i7); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_splat_4x4() local
154 const v128_t vi7x8ACE = wasm_v32x4_shuffle(vi7x89AB, vi7xCDEF, 0, 2, 4, 6); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_splat_4x4()
155 const v128_t vi7x9BDF = wasm_v32x4_shuffle(vi7x89AB, vi7xCDEF, 1, 3, 5, 7); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_splat_4x4()
260 const v128_t vi7x89AB = wasm_v128_load(i7); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_splat_4x4() local
279 …const v128_t vi7x8ACE = wasm_v128_and(vmask_even, wasm_v32x4_shuffle(vi7x89AB, vi7xCDEF, 0, 2, 4, … in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_splat_4x4()
280 …const v128_t vi7x9BDF = wasm_v128_and(vmask_odd, wasm_v32x4_shuffle(vi7x89AB, vi7xCDEF, 1, 3, 5, … in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__wasmsimd_x86_splat_4x4()
|
| D | 3x3s2p1-minmax-sse-4x4.c | 132 const __m128 vi7x89AB = _mm_loadu_ps(i7); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_4x4() local
153 const __m128 vi7x8ACE = _mm_shuffle_ps(vi7x89AB, vi7xCDEF, _MM_SHUFFLE(2, 0, 2, 0)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_4x4()
154 const __m128 vi7x9BDF = _mm_shuffle_ps(vi7x89AB, vi7xCDEF, _MM_SHUFFLE(3, 1, 3, 1)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_4x4()
264 const __m128 vi7x89AB = _mm_loadu_ps(i7); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_4x4() local
283 …const __m128 vi7x8ACE = _mm_and_ps(vmask_even, _mm_shuffle_ps(vi7x89AB, vi7xCDEF, _MM_SHUFFLE(2, 0… in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_4x4()
284 …const __m128 vi7x9BDF = _mm_and_ps(vmask_odd, _mm_shuffle_ps(vi7x89AB, vi7xCDEF, _MM_SHUFFLE(3, 1… in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_4x4()
|
| D | 5x5p2-minmax-neon-4x4.c | 114 const float32x4_t vi7x89AB = vld1q_f32(i7); i7 += 4; in xnn_f32_dwconv2d_chw_ukernel_5x5p2__neon_4x4() local
224 const float32x4_t vi7x5678 = vextq_f32(vi7x4567, vi7x89AB, 1); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__neon_4x4()
265 const float32x4_t vi7x6789 = vextq_f32(vi7x4567, vi7x89AB, 2); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__neon_4x4()
266 vi7x4567 = vi7x89AB; in xnn_f32_dwconv2d_chw_ukernel_5x5p2__neon_4x4()
323 float32x4_t vi7x89AB = vld1q_f32(i7); i7 += 4; in xnn_f32_dwconv2d_chw_ukernel_5x5p2__neon_4x4() local
332 vi7x89AB = vreinterpretq_f32_u32(vandq_u32(vmask, vreinterpretq_u32_f32(vi7x89AB))); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__neon_4x4()
442 const float32x4_t vi7x5678 = vextq_f32(vi7x4567, vi7x89AB, 1); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__neon_4x4()
483 const float32x4_t vi7x6789 = vextq_f32(vi7x4567, vi7x89AB, 2); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__neon_4x4()
484 vi7x4567 = vi7x89AB; in xnn_f32_dwconv2d_chw_ukernel_5x5p2__neon_4x4()
|
| D | 5x5p2-minmax-neon-4x4-acc2.c | 114 const float32x4_t vi7x89AB = vld1q_f32(i7); i7 += 4; in xnn_f32_dwconv2d_chw_ukernel_5x5p2__neon_4x4_acc2() local
224 const float32x4_t vi7x5678 = vextq_f32(vi7x4567, vi7x89AB, 1); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__neon_4x4_acc2()
265 const float32x4_t vi7x6789 = vextq_f32(vi7x4567, vi7x89AB, 2); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__neon_4x4_acc2()
266 vi7x4567 = vi7x89AB; in xnn_f32_dwconv2d_chw_ukernel_5x5p2__neon_4x4_acc2()
327 float32x4_t vi7x89AB = vld1q_f32(i7); i7 += 4; in xnn_f32_dwconv2d_chw_ukernel_5x5p2__neon_4x4_acc2() local
336 vi7x89AB = vreinterpretq_f32_u32(vandq_u32(vmask, vreinterpretq_u32_f32(vi7x89AB))); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__neon_4x4_acc2()
446 const float32x4_t vi7x5678 = vextq_f32(vi7x4567, vi7x89AB, 1); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__neon_4x4_acc2()
487 const float32x4_t vi7x6789 = vextq_f32(vi7x4567, vi7x89AB, 2); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__neon_4x4_acc2()
488 vi7x4567 = vi7x89AB; in xnn_f32_dwconv2d_chw_ukernel_5x5p2__neon_4x4_acc2()
|
| D | 5x5p2-minmax-neonfma-4x4.c | 114 const float32x4_t vi7x89AB = vld1q_f32(i7); i7 += 4; in xnn_f32_dwconv2d_chw_ukernel_5x5p2__neonfma_4x4() local
224 const float32x4_t vi7x5678 = vextq_f32(vi7x4567, vi7x89AB, 1); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__neonfma_4x4()
265 const float32x4_t vi7x6789 = vextq_f32(vi7x4567, vi7x89AB, 2); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__neonfma_4x4()
266 vi7x4567 = vi7x89AB; in xnn_f32_dwconv2d_chw_ukernel_5x5p2__neonfma_4x4()
323 float32x4_t vi7x89AB = vld1q_f32(i7); i7 += 4; in xnn_f32_dwconv2d_chw_ukernel_5x5p2__neonfma_4x4() local
332 vi7x89AB = vreinterpretq_f32_u32(vandq_u32(vmask, vreinterpretq_u32_f32(vi7x89AB))); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__neonfma_4x4()
442 const float32x4_t vi7x5678 = vextq_f32(vi7x4567, vi7x89AB, 1); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__neonfma_4x4()
483 const float32x4_t vi7x6789 = vextq_f32(vi7x4567, vi7x89AB, 2); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__neonfma_4x4()
484 vi7x4567 = vi7x89AB; in xnn_f32_dwconv2d_chw_ukernel_5x5p2__neonfma_4x4()
|
| D | 5x5p2-minmax-neonfma-4x4-acc2.c | 114 const float32x4_t vi7x89AB = vld1q_f32(i7); i7 += 4; in xnn_f32_dwconv2d_chw_ukernel_5x5p2__neonfma_4x4_acc2() local
224 const float32x4_t vi7x5678 = vextq_f32(vi7x4567, vi7x89AB, 1); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__neonfma_4x4_acc2()
265 const float32x4_t vi7x6789 = vextq_f32(vi7x4567, vi7x89AB, 2); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__neonfma_4x4_acc2()
266 vi7x4567 = vi7x89AB; in xnn_f32_dwconv2d_chw_ukernel_5x5p2__neonfma_4x4_acc2()
327 float32x4_t vi7x89AB = vld1q_f32(i7); i7 += 4; in xnn_f32_dwconv2d_chw_ukernel_5x5p2__neonfma_4x4_acc2() local
336 vi7x89AB = vreinterpretq_f32_u32(vandq_u32(vmask, vreinterpretq_u32_f32(vi7x89AB))); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__neonfma_4x4_acc2()
446 const float32x4_t vi7x5678 = vextq_f32(vi7x4567, vi7x89AB, 1); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__neonfma_4x4_acc2()
487 const float32x4_t vi7x6789 = vextq_f32(vi7x4567, vi7x89AB, 2); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__neonfma_4x4_acc2()
488 vi7x4567 = vi7x89AB; in xnn_f32_dwconv2d_chw_ukernel_5x5p2__neonfma_4x4_acc2()
|
| D | 5x5p2-minmax-wasmsimd-x86-loadsplat-4x4-acc2.c | 143 const v128_t vi7x89AB = wasm_v128_load(i7); i7 += 4; in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_x86_loadsplat_4x4_acc2() local
253 const v128_t vi7x5678 = wasm_v32x4_shuffle(vi7x4567, vi7x89AB, 1, 2, 3, 4); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_x86_loadsplat_4x4_acc2()
294 const v128_t vi7x6789 = wasm_v32x4_shuffle(vi7x4567, vi7x89AB, 2, 3, 4, 5); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_x86_loadsplat_4x4_acc2()
295 vi7x4567 = vi7x89AB; in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_x86_loadsplat_4x4_acc2()
355 v128_t vi7x89AB = wasm_v128_load(i7); i7 += 4; in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_x86_loadsplat_4x4_acc2() local
364 vi7x89AB = wasm_v128_and(vmask, vi7x89AB); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_x86_loadsplat_4x4_acc2()
474 const v128_t vi7x5678 = wasm_v32x4_shuffle(vi7x4567, vi7x89AB, 1, 2, 3, 4); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_x86_loadsplat_4x4_acc2()
515 const v128_t vi7x6789 = wasm_v32x4_shuffle(vi7x4567, vi7x89AB, 2, 3, 4, 5); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_x86_loadsplat_4x4_acc2()
516 vi7x4567 = vi7x89AB; in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_x86_loadsplat_4x4_acc2()
|
| D | 5x5p2-minmax-wasmsimd-arm-loadsplat-4x4.c | 143 const v128_t vi7x89AB = wasm_v128_load(i7); i7 += 4; in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_arm_loadsplat_4x4() local
253 const v128_t vi7x5678 = wasm_v32x4_shuffle(vi7x4567, vi7x89AB, 1, 2, 3, 4); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_arm_loadsplat_4x4()
294 const v128_t vi7x6789 = wasm_v32x4_shuffle(vi7x4567, vi7x89AB, 2, 3, 4, 5); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_arm_loadsplat_4x4()
295 vi7x4567 = vi7x89AB; in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_arm_loadsplat_4x4()
351 v128_t vi7x89AB = wasm_v128_load(i7); i7 += 4; in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_arm_loadsplat_4x4() local
360 vi7x89AB = wasm_v128_and(vmask, vi7x89AB); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_arm_loadsplat_4x4()
470 const v128_t vi7x5678 = wasm_v32x4_shuffle(vi7x4567, vi7x89AB, 1, 2, 3, 4); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_arm_loadsplat_4x4()
511 const v128_t vi7x6789 = wasm_v32x4_shuffle(vi7x4567, vi7x89AB, 2, 3, 4, 5); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_arm_loadsplat_4x4()
512 vi7x4567 = vi7x89AB; in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_arm_loadsplat_4x4()
|
| D | 5x5p2-minmax-wasmsimd-arm-loadsplat-4x4-acc2.c | 143 const v128_t vi7x89AB = wasm_v128_load(i7); i7 += 4; in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_arm_loadsplat_4x4_acc2() local
253 const v128_t vi7x5678 = wasm_v32x4_shuffle(vi7x4567, vi7x89AB, 1, 2, 3, 4); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_arm_loadsplat_4x4_acc2()
294 const v128_t vi7x6789 = wasm_v32x4_shuffle(vi7x4567, vi7x89AB, 2, 3, 4, 5); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_arm_loadsplat_4x4_acc2()
295 vi7x4567 = vi7x89AB; in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_arm_loadsplat_4x4_acc2()
355 v128_t vi7x89AB = wasm_v128_load(i7); i7 += 4; in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_arm_loadsplat_4x4_acc2() local
364 vi7x89AB = wasm_v128_and(vmask, vi7x89AB); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_arm_loadsplat_4x4_acc2()
474 const v128_t vi7x5678 = wasm_v32x4_shuffle(vi7x4567, vi7x89AB, 1, 2, 3, 4); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_arm_loadsplat_4x4_acc2()
515 const v128_t vi7x6789 = wasm_v32x4_shuffle(vi7x4567, vi7x89AB, 2, 3, 4, 5); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_arm_loadsplat_4x4_acc2()
516 vi7x4567 = vi7x89AB; in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_arm_loadsplat_4x4_acc2()
|
| D | 5x5p2-minmax-wasmsimd-x86-loadsplat-4x4.c | 143 const v128_t vi7x89AB = wasm_v128_load(i7); i7 += 4; in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_x86_loadsplat_4x4() local
253 const v128_t vi7x5678 = wasm_v32x4_shuffle(vi7x4567, vi7x89AB, 1, 2, 3, 4); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_x86_loadsplat_4x4()
294 const v128_t vi7x6789 = wasm_v32x4_shuffle(vi7x4567, vi7x89AB, 2, 3, 4, 5); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_x86_loadsplat_4x4()
295 vi7x4567 = vi7x89AB; in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_x86_loadsplat_4x4()
351 v128_t vi7x89AB = wasm_v128_load(i7); i7 += 4; in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_x86_loadsplat_4x4() local
360 vi7x89AB = wasm_v128_and(vmask, vi7x89AB); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_x86_loadsplat_4x4()
470 const v128_t vi7x5678 = wasm_v32x4_shuffle(vi7x4567, vi7x89AB, 1, 2, 3, 4); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_x86_loadsplat_4x4()
511 const v128_t vi7x6789 = wasm_v32x4_shuffle(vi7x4567, vi7x89AB, 2, 3, 4, 5); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_x86_loadsplat_4x4()
512 vi7x4567 = vi7x89AB; in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_x86_loadsplat_4x4()
|
| D | 3x3s2p1-minmax-sse-5x4.c | 144 const __m128 vi7x89AB = _mm_loadu_ps(i7); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_5x4() local
171 const __m128 vi7x8ACE = _mm_shuffle_ps(vi7x89AB, vi7xCDEF, _MM_SHUFFLE(2, 0, 2, 0)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_5x4()
172 const __m128 vi7x9BDF = _mm_shuffle_ps(vi7x89AB, vi7xCDEF, _MM_SHUFFLE(3, 1, 3, 1)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_5x4()
305 const __m128 vi7x89AB = _mm_loadu_ps(i7); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_5x4() local
328 …const __m128 vi7x8ACE = _mm_and_ps(vmask_even, _mm_shuffle_ps(vi7x89AB, vi7xCDEF, _MM_SHUFFLE(2, 0… in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_5x4()
329 …const __m128 vi7x9BDF = _mm_and_ps(vmask_odd, _mm_shuffle_ps(vi7x89AB, vi7xCDEF, _MM_SHUFFLE(3, 1… in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_5x4()
|
| D | 3x3s2p1-minmax-sse-6x4.c | 156 const __m128 vi7x89AB = _mm_loadu_ps(i7); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_6x4() local
189 const __m128 vi7x8ACE = _mm_shuffle_ps(vi7x89AB, vi7xCDEF, _MM_SHUFFLE(2, 0, 2, 0)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_6x4()
190 const __m128 vi7x9BDF = _mm_shuffle_ps(vi7x89AB, vi7xCDEF, _MM_SHUFFLE(3, 1, 3, 1)); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_6x4()
346 const __m128 vi7x89AB = _mm_loadu_ps(i7); in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_6x4() local
373 …const __m128 vi7x8ACE = _mm_and_ps(vmask_even, _mm_shuffle_ps(vi7x89AB, vi7xCDEF, _MM_SHUFFLE(2, 0… in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_6x4()
374 …const __m128 vi7x9BDF = _mm_and_ps(vmask_odd, _mm_shuffle_ps(vi7x89AB, vi7xCDEF, _MM_SHUFFLE(3, 1… in xnn_f32_dwconv2d_chw_ukernel_3x3s2p1__sse_6x4()
|
| D | 5x5p2-minmax-wasmsimd-arm-splat-4x4-acc2.c | 117 const v128_t vi7x89AB = wasm_v128_load(i7); i7 += 4; in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_arm_splat_4x4_acc2() local
227 const v128_t vi7x5678 = wasm_v32x4_shuffle(vi7x4567, vi7x89AB, 1, 2, 3, 4); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_arm_splat_4x4_acc2()
268 const v128_t vi7x6789 = wasm_v32x4_shuffle(vi7x4567, vi7x89AB, 2, 3, 4, 5); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_arm_splat_4x4_acc2()
269 vi7x4567 = vi7x89AB; in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_arm_splat_4x4_acc2()
329 v128_t vi7x89AB = wasm_v128_load(i7); i7 += 4; in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_arm_splat_4x4_acc2() local
338 vi7x89AB = wasm_v128_and(vmask, vi7x89AB); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_arm_splat_4x4_acc2()
448 const v128_t vi7x5678 = wasm_v32x4_shuffle(vi7x4567, vi7x89AB, 1, 2, 3, 4); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_arm_splat_4x4_acc2()
489 const v128_t vi7x6789 = wasm_v32x4_shuffle(vi7x4567, vi7x89AB, 2, 3, 4, 5); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_arm_splat_4x4_acc2()
490 vi7x4567 = vi7x89AB; in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_arm_splat_4x4_acc2()
|
| D | 5x5p2-minmax-wasmsimd-x86-loadsplat-5x4.c | 152 const v128_t vi7x89AB = wasm_v128_load(i7); i7 += 4; in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_x86_loadsplat_5x4() local
281 const v128_t vi7x5678 = wasm_v32x4_shuffle(vi7x4567, vi7x89AB, 1, 2, 3, 4); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_x86_loadsplat_5x4()
328 const v128_t vi7x6789 = wasm_v32x4_shuffle(vi7x4567, vi7x89AB, 2, 3, 4, 5); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_x86_loadsplat_5x4()
329 vi7x4567 = vi7x89AB; in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_x86_loadsplat_5x4()
396 v128_t vi7x89AB = wasm_v128_load(i7); i7 += 4; in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_x86_loadsplat_5x4() local
406 vi7x89AB = wasm_v128_and(vmask, vi7x89AB); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_x86_loadsplat_5x4()
535 const v128_t vi7x5678 = wasm_v32x4_shuffle(vi7x4567, vi7x89AB, 1, 2, 3, 4); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_x86_loadsplat_5x4()
582 const v128_t vi7x6789 = wasm_v32x4_shuffle(vi7x4567, vi7x89AB, 2, 3, 4, 5); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_x86_loadsplat_5x4()
583 vi7x4567 = vi7x89AB; in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_x86_loadsplat_5x4()
|
| D | 5x5p2-minmax-neonfma-5x4.c | 123 const float32x4_t vi7x89AB = vld1q_f32(i7); i7 += 4; in xnn_f32_dwconv2d_chw_ukernel_5x5p2__neonfma_5x4() local
252 const float32x4_t vi7x5678 = vextq_f32(vi7x4567, vi7x89AB, 1); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__neonfma_5x4()
299 const float32x4_t vi7x6789 = vextq_f32(vi7x4567, vi7x89AB, 2); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__neonfma_5x4()
300 vi7x4567 = vi7x89AB; in xnn_f32_dwconv2d_chw_ukernel_5x5p2__neonfma_5x4()
368 float32x4_t vi7x89AB = vld1q_f32(i7); i7 += 4; in xnn_f32_dwconv2d_chw_ukernel_5x5p2__neonfma_5x4() local
378 vi7x89AB = vreinterpretq_f32_u32(vandq_u32(vmask, vreinterpretq_u32_f32(vi7x89AB))); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__neonfma_5x4()
507 const float32x4_t vi7x5678 = vextq_f32(vi7x4567, vi7x89AB, 1); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__neonfma_5x4()
554 const float32x4_t vi7x6789 = vextq_f32(vi7x4567, vi7x89AB, 2); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__neonfma_5x4()
555 vi7x4567 = vi7x89AB; in xnn_f32_dwconv2d_chw_ukernel_5x5p2__neonfma_5x4()
|
| D | 5x5p2-minmax-neon-5x4.c | 123 const float32x4_t vi7x89AB = vld1q_f32(i7); i7 += 4; in xnn_f32_dwconv2d_chw_ukernel_5x5p2__neon_5x4() local
252 const float32x4_t vi7x5678 = vextq_f32(vi7x4567, vi7x89AB, 1); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__neon_5x4()
299 const float32x4_t vi7x6789 = vextq_f32(vi7x4567, vi7x89AB, 2); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__neon_5x4()
300 vi7x4567 = vi7x89AB; in xnn_f32_dwconv2d_chw_ukernel_5x5p2__neon_5x4()
368 float32x4_t vi7x89AB = vld1q_f32(i7); i7 += 4; in xnn_f32_dwconv2d_chw_ukernel_5x5p2__neon_5x4() local
378 vi7x89AB = vreinterpretq_f32_u32(vandq_u32(vmask, vreinterpretq_u32_f32(vi7x89AB))); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__neon_5x4()
507 const float32x4_t vi7x5678 = vextq_f32(vi7x4567, vi7x89AB, 1); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__neon_5x4()
554 const float32x4_t vi7x6789 = vextq_f32(vi7x4567, vi7x89AB, 2); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__neon_5x4()
555 vi7x4567 = vi7x89AB; in xnn_f32_dwconv2d_chw_ukernel_5x5p2__neon_5x4()
|
| D | 5x5p2-minmax-wasmsimd-arm-splat-4x4.c | 117 const v128_t vi7x89AB = wasm_v128_load(i7); i7 += 4; in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_arm_splat_4x4() local
227 const v128_t vi7x5678 = wasm_v32x4_shuffle(vi7x4567, vi7x89AB, 1, 2, 3, 4); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_arm_splat_4x4()
268 const v128_t vi7x6789 = wasm_v32x4_shuffle(vi7x4567, vi7x89AB, 2, 3, 4, 5); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_arm_splat_4x4()
269 vi7x4567 = vi7x89AB; in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_arm_splat_4x4()
325 v128_t vi7x89AB = wasm_v128_load(i7); i7 += 4; in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_arm_splat_4x4() local
334 vi7x89AB = wasm_v128_and(vmask, vi7x89AB); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_arm_splat_4x4()
444 const v128_t vi7x5678 = wasm_v32x4_shuffle(vi7x4567, vi7x89AB, 1, 2, 3, 4); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_arm_splat_4x4()
485 const v128_t vi7x6789 = wasm_v32x4_shuffle(vi7x4567, vi7x89AB, 2, 3, 4, 5); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_arm_splat_4x4()
486 vi7x4567 = vi7x89AB; in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_arm_splat_4x4()
|
| D | 5x5p2-minmax-wasmsimd-arm-loadsplat-5x4.c | 152 const v128_t vi7x89AB = wasm_v128_load(i7); i7 += 4; in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_arm_loadsplat_5x4() local
281 const v128_t vi7x5678 = wasm_v32x4_shuffle(vi7x4567, vi7x89AB, 1, 2, 3, 4); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_arm_loadsplat_5x4()
328 const v128_t vi7x6789 = wasm_v32x4_shuffle(vi7x4567, vi7x89AB, 2, 3, 4, 5); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_arm_loadsplat_5x4()
329 vi7x4567 = vi7x89AB; in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_arm_loadsplat_5x4()
396 v128_t vi7x89AB = wasm_v128_load(i7); i7 += 4; in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_arm_loadsplat_5x4() local
406 vi7x89AB = wasm_v128_and(vmask, vi7x89AB); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_arm_loadsplat_5x4()
535 const v128_t vi7x5678 = wasm_v32x4_shuffle(vi7x4567, vi7x89AB, 1, 2, 3, 4); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_arm_loadsplat_5x4()
582 const v128_t vi7x6789 = wasm_v32x4_shuffle(vi7x4567, vi7x89AB, 2, 3, 4, 5); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_arm_loadsplat_5x4()
583 vi7x4567 = vi7x89AB; in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_arm_loadsplat_5x4()
|
| D | 5x5p2-minmax-wasmsimd-x86-splat-4x4.c | 117 const v128_t vi7x89AB = wasm_v128_load(i7); i7 += 4; in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_x86_splat_4x4() local
227 const v128_t vi7x5678 = wasm_v32x4_shuffle(vi7x4567, vi7x89AB, 1, 2, 3, 4); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_x86_splat_4x4()
268 const v128_t vi7x6789 = wasm_v32x4_shuffle(vi7x4567, vi7x89AB, 2, 3, 4, 5); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_x86_splat_4x4()
269 vi7x4567 = vi7x89AB; in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_x86_splat_4x4()
325 v128_t vi7x89AB = wasm_v128_load(i7); i7 += 4; in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_x86_splat_4x4() local
334 vi7x89AB = wasm_v128_and(vmask, vi7x89AB); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_x86_splat_4x4()
444 const v128_t vi7x5678 = wasm_v32x4_shuffle(vi7x4567, vi7x89AB, 1, 2, 3, 4); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_x86_splat_4x4()
485 const v128_t vi7x6789 = wasm_v32x4_shuffle(vi7x4567, vi7x89AB, 2, 3, 4, 5); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_x86_splat_4x4()
486 vi7x4567 = vi7x89AB; in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_x86_splat_4x4()
|
| D | 5x5p2-minmax-wasmsimd-x86-splat-4x4-acc2.c | 117 const v128_t vi7x89AB = wasm_v128_load(i7); i7 += 4; in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_x86_splat_4x4_acc2() local
227 const v128_t vi7x5678 = wasm_v32x4_shuffle(vi7x4567, vi7x89AB, 1, 2, 3, 4); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_x86_splat_4x4_acc2()
268 const v128_t vi7x6789 = wasm_v32x4_shuffle(vi7x4567, vi7x89AB, 2, 3, 4, 5); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_x86_splat_4x4_acc2()
269 vi7x4567 = vi7x89AB; in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_x86_splat_4x4_acc2()
329 v128_t vi7x89AB = wasm_v128_load(i7); i7 += 4; in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_x86_splat_4x4_acc2() local
338 vi7x89AB = wasm_v128_and(vmask, vi7x89AB); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_x86_splat_4x4_acc2()
448 const v128_t vi7x5678 = wasm_v32x4_shuffle(vi7x4567, vi7x89AB, 1, 2, 3, 4); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_x86_splat_4x4_acc2()
489 const v128_t vi7x6789 = wasm_v32x4_shuffle(vi7x4567, vi7x89AB, 2, 3, 4, 5); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_x86_splat_4x4_acc2()
490 vi7x4567 = vi7x89AB; in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_x86_splat_4x4_acc2()
|
| D | 5x5p2-minmax-wasmsimd-x86-splat-5x4.c | 126 const v128_t vi7x89AB = wasm_v128_load(i7); i7 += 4; in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_x86_splat_5x4() local
255 const v128_t vi7x5678 = wasm_v32x4_shuffle(vi7x4567, vi7x89AB, 1, 2, 3, 4); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_x86_splat_5x4()
302 const v128_t vi7x6789 = wasm_v32x4_shuffle(vi7x4567, vi7x89AB, 2, 3, 4, 5); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_x86_splat_5x4()
303 vi7x4567 = vi7x89AB; in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_x86_splat_5x4()
370 v128_t vi7x89AB = wasm_v128_load(i7); i7 += 4; in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_x86_splat_5x4() local
380 vi7x89AB = wasm_v128_and(vmask, vi7x89AB); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_x86_splat_5x4()
509 const v128_t vi7x5678 = wasm_v32x4_shuffle(vi7x4567, vi7x89AB, 1, 2, 3, 4); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_x86_splat_5x4()
556 const v128_t vi7x6789 = wasm_v32x4_shuffle(vi7x4567, vi7x89AB, 2, 3, 4, 5); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_x86_splat_5x4()
557 vi7x4567 = vi7x89AB; in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_x86_splat_5x4()
|
| D | 5x5p2-minmax-wasmsimd-arm-splat-5x4.c | 126 const v128_t vi7x89AB = wasm_v128_load(i7); i7 += 4; in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_arm_splat_5x4() local
255 const v128_t vi7x5678 = wasm_v32x4_shuffle(vi7x4567, vi7x89AB, 1, 2, 3, 4); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_arm_splat_5x4()
302 const v128_t vi7x6789 = wasm_v32x4_shuffle(vi7x4567, vi7x89AB, 2, 3, 4, 5); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_arm_splat_5x4()
303 vi7x4567 = vi7x89AB; in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_arm_splat_5x4()
370 v128_t vi7x89AB = wasm_v128_load(i7); i7 += 4; in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_arm_splat_5x4() local
380 vi7x89AB = wasm_v128_and(vmask, vi7x89AB); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_arm_splat_5x4()
509 const v128_t vi7x5678 = wasm_v32x4_shuffle(vi7x4567, vi7x89AB, 1, 2, 3, 4); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_arm_splat_5x4()
556 const v128_t vi7x6789 = wasm_v32x4_shuffle(vi7x4567, vi7x89AB, 2, 3, 4, 5); in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_arm_splat_5x4()
557 vi7x4567 = vi7x89AB; in xnn_f32_dwconv2d_chw_ukernel_5x5p2__wasmsimd_arm_splat_5x4()
|