/external/XNNPACK/src/f32-conv-hwc2chw/ |
D | 3x3s2p1c3x4-neonfma-2x2.c | 263 const float32x4_t vk12c0 = vld1q_f32(w + 80); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__neonfma_2x2() local 265 vo0x0 = vfmaq_laneq_f32(vo0x0, vk12c0, vi1x1, 3); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__neonfma_2x2() 266 vo1x0 = vfmaq_laneq_f32(vo1x0, vk12c0, vi3x1, 3); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__neonfma_2x2() 267 vo0x1 = vfmaq_laneq_f32(vo0x1, vk12c0, vi1x3, 1); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__neonfma_2x2() 268 vo1x1 = vfmaq_laneq_f32(vo1x1, vk12c0, vi3x3, 1); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__neonfma_2x2() 543 const float32x4_t vk12c0 = vld1q_f32(w + 80); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__neonfma_2x2() local 545 vo0x0 = vfmaq_laneq_f32(vo0x0, vk12c0, vi1x1, 3); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__neonfma_2x2() 546 vo1x0 = vfmaq_laneq_f32(vo1x0, vk12c0, vi3x1, 3); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__neonfma_2x2()
|
D | 3x3s2p1c3x4-neon-2x2.c | 263 const float32x4_t vk12c0 = vld1q_f32(w + 80); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__neon_2x2() local 265 vo0x0 = vmlaq_lane_f32(vo0x0, vk12c0, vget_high_f32(vi1x1), 1); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__neon_2x2() 266 vo1x0 = vmlaq_lane_f32(vo1x0, vk12c0, vget_high_f32(vi3x1), 1); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__neon_2x2() 267 vo0x1 = vmlaq_lane_f32(vo0x1, vk12c0, vget_low_f32(vi1x3), 1); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__neon_2x2() 268 vo1x1 = vmlaq_lane_f32(vo1x1, vk12c0, vget_low_f32(vi3x3), 1); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__neon_2x2() 541 const float32x4_t vk12c0 = vld1q_f32(w + 80); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__neon_2x2() local 543 vo0x0 = vmlaq_lane_f32(vo0x0, vk12c0, vget_high_f32(vi1x1), 1); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__neon_2x2() 544 vo1x0 = vmlaq_lane_f32(vo1x0, vk12c0, vget_high_f32(vi3x1), 1); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__neon_2x2()
|
D | 3x3s2p1c3x4-wasmsimd-2x2.c | 264 const v128_t vk12c0 = wasm_v128_load(w + 80); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__wasmsimd_2x2() local 266 …vo0x0 = wasm_f32x4_add(vo0x0, wasm_f32x4_mul(vk12c0, wasm_v32x4_shuffle(vi1x1, vi1x1, 3, 3, 3, 3))… in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__wasmsimd_2x2() 267 …vo1x0 = wasm_f32x4_add(vo1x0, wasm_f32x4_mul(vk12c0, wasm_v32x4_shuffle(vi3x1, vi3x1, 3, 3, 3, 3))… in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__wasmsimd_2x2() 268 …vo0x1 = wasm_f32x4_add(vo0x1, wasm_f32x4_mul(vk12c0, wasm_v32x4_shuffle(vi1x3, vi1x3, 1, 1, 1, 1))… in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__wasmsimd_2x2() 269 …vo1x1 = wasm_f32x4_add(vo1x1, wasm_f32x4_mul(vk12c0, wasm_v32x4_shuffle(vi3x3, vi3x3, 1, 1, 1, 1))… in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__wasmsimd_2x2() 543 const v128_t vk12c0 = wasm_v128_load(w + 80); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__wasmsimd_2x2() local 545 …vo0x0 = wasm_f32x4_add(vo0x0, wasm_f32x4_mul(vk12c0, wasm_v32x4_shuffle(vi1x1, vi1x1, 3, 3, 3, 3))… in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__wasmsimd_2x2() 546 …vo1x0 = wasm_f32x4_add(vo1x0, wasm_f32x4_mul(vk12c0, wasm_v32x4_shuffle(vi3x1, vi3x1, 3, 3, 3, 3))… in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__wasmsimd_2x2()
|
D | 3x3s2p1c3x4-sse-2x2.c | 263 const __m128 vk12c0 = _mm_load_ps(w + 80); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__sse_2x2() local 265 …vo0x0 = _mm_add_ps(vo0x0, _mm_mul_ps(vk12c0, _mm_shuffle_ps(vi1x1, vi1x1, _MM_SHUFFLE(3, 3, 3, 3))… in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__sse_2x2() 266 …vo1x0 = _mm_add_ps(vo1x0, _mm_mul_ps(vk12c0, _mm_shuffle_ps(vi3x1, vi3x1, _MM_SHUFFLE(3, 3, 3, 3))… in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__sse_2x2() 267 …vo0x1 = _mm_add_ps(vo0x1, _mm_mul_ps(vk12c0, _mm_shuffle_ps(vi1x3, vi1x3, _MM_SHUFFLE(1, 1, 1, 1))… in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__sse_2x2() 268 …vo1x1 = _mm_add_ps(vo1x1, _mm_mul_ps(vk12c0, _mm_shuffle_ps(vi3x3, vi3x3, _MM_SHUFFLE(1, 1, 1, 1))… in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__sse_2x2() 543 const __m128 vk12c0 = _mm_load_ps(w + 80); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__sse_2x2() local 545 …vo0x0 = _mm_add_ps(vo0x0, _mm_mul_ps(vk12c0, _mm_shuffle_ps(vi1x1, vi1x1, _MM_SHUFFLE(3, 3, 3, 3))… in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__sse_2x2() 546 …vo1x0 = _mm_add_ps(vo1x0, _mm_mul_ps(vk12c0, _mm_shuffle_ps(vi3x1, vi3x1, _MM_SHUFFLE(3, 3, 3, 3))… in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__sse_2x2()
|