/external/XNNPACK/src/f32-conv-hwc2chw/ |
D | 3x3s2p1c3x4-neonfma-2x2.c | 312 const float32x4_t vk22c2 = vld1q_f32(w + 108); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__neonfma_2x2() local 314 vo0x0 = vfmaq_laneq_f32(vo0x0, vk22c2, vi2x2, 1); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__neonfma_2x2() 315 vo1x0 = vfmaq_laneq_f32(vo1x0, vk22c2, vi4x2, 1); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__neonfma_2x2() 316 vo0x1 = vfmaq_laneq_f32(vo0x1, vk22c2, vi2x3, 3); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__neonfma_2x2() 317 vo1x1 = vfmaq_laneq_f32(vo1x1, vk22c2, vi4x3, 3); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__neonfma_2x2() 578 const float32x4_t vk22c2 = vld1q_f32(w + 108); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__neonfma_2x2() local 580 vo0x0 = vfmaq_laneq_f32(vo0x0, vk22c2, vi2x2, 1); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__neonfma_2x2() 581 vo1x0 = vfmaq_laneq_f32(vo1x0, vk22c2, vi4x2, 1); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__neonfma_2x2()
|
D | 3x3s2p1c3x4-neon-2x2.c | 312 const float32x4_t vk22c2 = vld1q_f32(w + 108); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__neon_2x2() local 314 vo0x0 = vmlaq_lane_f32(vo0x0, vk22c2, vget_low_f32(vi2x2), 1); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__neon_2x2() 315 vo1x0 = vmlaq_lane_f32(vo1x0, vk22c2, vget_low_f32(vi4x2), 1); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__neon_2x2() 316 vo0x1 = vmlaq_lane_f32(vo0x1, vk22c2, vget_high_f32(vi2x3), 1); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__neon_2x2() 317 vo1x1 = vmlaq_lane_f32(vo1x1, vk22c2, vget_high_f32(vi4x3), 1); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__neon_2x2() 576 const float32x4_t vk22c2 = vld1q_f32(w + 108); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__neon_2x2() local 578 vo0x0 = vmlaq_lane_f32(vo0x0, vk22c2, vget_low_f32(vi2x2), 1); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__neon_2x2() 579 vo1x0 = vmlaq_lane_f32(vo1x0, vk22c2, vget_low_f32(vi4x2), 1); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__neon_2x2()
|
D | 3x3s2p1c3x4-wasmsimd-2x2.c | 313 const v128_t vk22c2 = wasm_v128_load(w + 108); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__wasmsimd_2x2() local 315 …vo0x0 = wasm_f32x4_add(vo0x0, wasm_f32x4_mul(vk22c2, wasm_v32x4_shuffle(vi2x2, vi2x2, 1, 1, 1, 1))… in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__wasmsimd_2x2() 316 …vo1x0 = wasm_f32x4_add(vo1x0, wasm_f32x4_mul(vk22c2, wasm_v32x4_shuffle(vi4x2, vi4x2, 1, 1, 1, 1))… in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__wasmsimd_2x2() 317 …vo0x1 = wasm_f32x4_add(vo0x1, wasm_f32x4_mul(vk22c2, wasm_v32x4_shuffle(vi2x3, vi2x3, 3, 3, 3, 3))… in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__wasmsimd_2x2() 318 …vo1x1 = wasm_f32x4_add(vo1x1, wasm_f32x4_mul(vk22c2, wasm_v32x4_shuffle(vi4x3, vi4x3, 3, 3, 3, 3))… in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__wasmsimd_2x2() 578 const v128_t vk22c2 = wasm_v128_load(w + 108); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__wasmsimd_2x2() local 580 …vo0x0 = wasm_f32x4_add(vo0x0, wasm_f32x4_mul(vk22c2, wasm_v32x4_shuffle(vi2x2, vi2x2, 1, 1, 1, 1))… in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__wasmsimd_2x2() 581 …vo1x0 = wasm_f32x4_add(vo1x0, wasm_f32x4_mul(vk22c2, wasm_v32x4_shuffle(vi4x2, vi4x2, 1, 1, 1, 1))… in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__wasmsimd_2x2()
|
D | 3x3s2p1c3x4-sse-2x2.c | 312 const __m128 vk22c2 = _mm_load_ps(w + 108); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__sse_2x2() local 314 …vo0x0 = _mm_add_ps(vo0x0, _mm_mul_ps(vk22c2, _mm_shuffle_ps(vi2x2, vi2x2, _MM_SHUFFLE(1, 1, 1, 1))… in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__sse_2x2() 315 …vo1x0 = _mm_add_ps(vo1x0, _mm_mul_ps(vk22c2, _mm_shuffle_ps(vi4x2, vi4x2, _MM_SHUFFLE(1, 1, 1, 1))… in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__sse_2x2() 316 …vo0x1 = _mm_add_ps(vo0x1, _mm_mul_ps(vk22c2, _mm_shuffle_ps(vi2x3, vi2x3, _MM_SHUFFLE(3, 3, 3, 3))… in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__sse_2x2() 317 …vo1x1 = _mm_add_ps(vo1x1, _mm_mul_ps(vk22c2, _mm_shuffle_ps(vi4x3, vi4x3, _MM_SHUFFLE(3, 3, 3, 3))… in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__sse_2x2() 578 const __m128 vk22c2 = _mm_load_ps(w + 108); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__sse_2x2() local 580 …vo0x0 = _mm_add_ps(vo0x0, _mm_mul_ps(vk22c2, _mm_shuffle_ps(vi2x2, vi2x2, _MM_SHUFFLE(1, 1, 1, 1))… in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__sse_2x2() 581 …vo1x0 = _mm_add_ps(vo1x0, _mm_mul_ps(vk22c2, _mm_shuffle_ps(vi4x2, vi4x2, _MM_SHUFFLE(1, 1, 1, 1))… in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__sse_2x2()
|