/external/XNNPACK/src/f32-conv-hwc2chw/ |
D | 3x3s2p1c3x4-neonfma-2x2.c | 305 const float32x4_t vk12c2 = vld1q_f32(w + 104); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__neonfma_2x2() local 307 vo0x0 = vfmaq_laneq_f32(vo0x0, vk12c2, vi1x2, 1); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__neonfma_2x2() 308 vo1x0 = vfmaq_laneq_f32(vo1x0, vk12c2, vi3x2, 1); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__neonfma_2x2() 309 vo0x1 = vfmaq_laneq_f32(vo0x1, vk12c2, vi1x3, 3); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__neonfma_2x2() 310 vo1x1 = vfmaq_laneq_f32(vo1x1, vk12c2, vi3x3, 3); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__neonfma_2x2() 573 const float32x4_t vk12c2 = vld1q_f32(w + 104); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__neonfma_2x2() local 575 vo0x0 = vfmaq_laneq_f32(vo0x0, vk12c2, vi1x2, 1); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__neonfma_2x2() 576 vo1x0 = vfmaq_laneq_f32(vo1x0, vk12c2, vi3x2, 1); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__neonfma_2x2()
|
D | 3x3s2p1c3x4-neon-2x2.c | 305 const float32x4_t vk12c2 = vld1q_f32(w + 104); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__neon_2x2() local 307 vo0x0 = vmlaq_lane_f32(vo0x0, vk12c2, vget_low_f32(vi1x2), 1); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__neon_2x2() 308 vo1x0 = vmlaq_lane_f32(vo1x0, vk12c2, vget_low_f32(vi3x2), 1); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__neon_2x2() 309 vo0x1 = vmlaq_lane_f32(vo0x1, vk12c2, vget_high_f32(vi1x3), 1); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__neon_2x2() 310 vo1x1 = vmlaq_lane_f32(vo1x1, vk12c2, vget_high_f32(vi3x3), 1); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__neon_2x2() 571 const float32x4_t vk12c2 = vld1q_f32(w + 104); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__neon_2x2() local 573 vo0x0 = vmlaq_lane_f32(vo0x0, vk12c2, vget_low_f32(vi1x2), 1); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__neon_2x2() 574 vo1x0 = vmlaq_lane_f32(vo1x0, vk12c2, vget_low_f32(vi3x2), 1); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__neon_2x2()
|
D | 3x3s2p1c3x4-wasmsimd-2x2.c | 306 const v128_t vk12c2 = wasm_v128_load(w + 104); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__wasmsimd_2x2() local 308 …vo0x0 = wasm_f32x4_add(vo0x0, wasm_f32x4_mul(vk12c2, wasm_v32x4_shuffle(vi1x2, vi1x2, 1, 1, 1, 1))… in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__wasmsimd_2x2() 309 …vo1x0 = wasm_f32x4_add(vo1x0, wasm_f32x4_mul(vk12c2, wasm_v32x4_shuffle(vi3x2, vi3x2, 1, 1, 1, 1))… in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__wasmsimd_2x2() 310 …vo0x1 = wasm_f32x4_add(vo0x1, wasm_f32x4_mul(vk12c2, wasm_v32x4_shuffle(vi1x3, vi1x3, 3, 3, 3, 3))… in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__wasmsimd_2x2() 311 …vo1x1 = wasm_f32x4_add(vo1x1, wasm_f32x4_mul(vk12c2, wasm_v32x4_shuffle(vi3x3, vi3x3, 3, 3, 3, 3))… in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__wasmsimd_2x2() 573 const v128_t vk12c2 = wasm_v128_load(w + 104); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__wasmsimd_2x2() local 575 …vo0x0 = wasm_f32x4_add(vo0x0, wasm_f32x4_mul(vk12c2, wasm_v32x4_shuffle(vi1x2, vi1x2, 1, 1, 1, 1))… in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__wasmsimd_2x2() 576 …vo1x0 = wasm_f32x4_add(vo1x0, wasm_f32x4_mul(vk12c2, wasm_v32x4_shuffle(vi3x2, vi3x2, 1, 1, 1, 1))… in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__wasmsimd_2x2()
|
D | 3x3s2p1c3x4-sse-2x2.c | 305 const __m128 vk12c2 = _mm_load_ps(w + 104); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__sse_2x2() local 307 …vo0x0 = _mm_add_ps(vo0x0, _mm_mul_ps(vk12c2, _mm_shuffle_ps(vi1x2, vi1x2, _MM_SHUFFLE(1, 1, 1, 1))… in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__sse_2x2() 308 …vo1x0 = _mm_add_ps(vo1x0, _mm_mul_ps(vk12c2, _mm_shuffle_ps(vi3x2, vi3x2, _MM_SHUFFLE(1, 1, 1, 1))… in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__sse_2x2() 309 …vo0x1 = _mm_add_ps(vo0x1, _mm_mul_ps(vk12c2, _mm_shuffle_ps(vi1x3, vi1x3, _MM_SHUFFLE(3, 3, 3, 3))… in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__sse_2x2() 310 …vo1x1 = _mm_add_ps(vo1x1, _mm_mul_ps(vk12c2, _mm_shuffle_ps(vi3x3, vi3x3, _MM_SHUFFLE(3, 3, 3, 3))… in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__sse_2x2() 573 const __m128 vk12c2 = _mm_load_ps(w + 104); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__sse_2x2() local 575 …vo0x0 = _mm_add_ps(vo0x0, _mm_mul_ps(vk12c2, _mm_shuffle_ps(vi1x2, vi1x2, _MM_SHUFFLE(1, 1, 1, 1))… in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__sse_2x2() 576 …vo1x0 = _mm_add_ps(vo1x0, _mm_mul_ps(vk12c2, _mm_shuffle_ps(vi3x2, vi3x2, _MM_SHUFFLE(1, 1, 1, 1))… in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__sse_2x2()
|