Merge branch 'optimize_general_convolution' into 'master'

optimize general convolution

See merge request !443

mace/kernels/conv_2d.h

@@ -98,33 +98,168 @@ struct Conv2dFunctor<DeviceType::CPU, float> : Conv2dFunctorBase {
                      const int dilation_h,
                      const int dilation_w,
                      float *output) {
+    const index_t in_image_size = in_height * in_width;
+    const index_t out_image_size = out_height * out_width;
+    const index_t in_batch_size = in_channels * in_image_size;
+    const index_t out_batch_size = out_channels * out_image_size;
+    const index_t filter_size = filter_height * filter_width;
+    const index_t in_tile_size =
+        3 * stride_w + (filter_width - 1) * dilation_w + 1;
+
 #pragma omp parallel for collapse(2)
     for (index_t b = 0; b < batch; ++b) {
-      for (index_t m = 0; m < out_channels; ++m) {
-        for (index_t h = 0; h < out_height; ++h) {
-          for (index_t w = 0; w < out_width; ++w) {
-            index_t out_offset =
-              ((b * out_channels + m) * out_height + h) * out_width + w;
-            for (index_t c = 0; c < in_channels; ++c) {
-              for (index_t kh = 0; kh < filter_height; ++kh) {
-                for (index_t kw = 0; kw < filter_width; ++kw) {
-                  index_t ih = h * stride_h + kh * dilation_h;
-                  index_t iw = w * stride_w + kw * dilation_w;
-                  index_t in_offset =
-                    ((b * in_channels + c) * in_height + ih) * in_width + iw;
-                  index_t filter_offset =
-                    (((m * in_channels) + c) * filter_height + kh)
-                      * filter_width
-                      + kw;
-                  output[out_offset] +=
-                    input[in_offset] * filter[filter_offset];
+      for (index_t m = 0; m < out_channels; m += 4) {
+        if (m + 3 < out_channels) {
+          float *out_ptr0_base =
+              output + b * out_batch_size + m * out_image_size;
+          float *out_ptr1_base =
+              output + b * out_batch_size + (m + 1) * out_image_size;
+          float *out_ptr2_base =
+              output + b * out_batch_size + (m + 2) * out_image_size;
+          float *out_ptr3_base =
+              output + b * out_batch_size + (m + 3) * out_image_size;
+          for (index_t c = 0; c < in_channels; ++c) {
+            const float *in_ptr_base =
+                input + b * in_batch_size + c * in_image_size;
+            const float *filter_ptr0 =
+                filter + m * in_channels * filter_size + c * filter_size;
+            const float *filter_ptr1 =
+                filter + (m + 1) * in_channels * filter_size + c * filter_size;
+            const float *filter_ptr2 =
+                filter + (m + 2) * in_channels * filter_size + c * filter_size;
+            const float *filter_ptr3 =
+                filter + (m + 3) * in_channels * filter_size + c * filter_size;
+            for (index_t h = 0; h < out_height; ++h) {
+              for (index_t w = 0; w + 3 < out_width; w += 4) {
+                 // input offset
+                index_t ih = h * stride_h;
+                index_t iw = w * stride_w;
+                index_t in_offset = ih * in_width + iw;
+                // output (4 outch x 1 height x 4 width): vo_outch_height
+                float vo0[4], vo1[4], vo2[4], vo3[4];
+                // load output
+                index_t out_offset = h * out_width + w;
+                for (index_t ow = 0; ow < 4; ++ow) {
+                   vo0[ow] = out_ptr0_base[out_offset + ow];
+                   vo1[ow] = out_ptr1_base[out_offset + ow];
+                   vo2[ow] = out_ptr2_base[out_offset + ow];
+                   vo3[ow] = out_ptr3_base[out_offset + ow];
                 }
-              }
-            }
-          }
-        }
-      }
-    }
+                // calc by row
+                for (index_t kh = 0; kh < filter_height; ++kh) {
+                  for (index_t kw = 0; kw < filter_width; ++kw) {
+                    // outch 0
+                    vo0[0] += in_ptr_base[in_offset
+                        + kw * dilation_w] * filter_ptr0[kw];
+                    vo0[1] += in_ptr_base[in_offset + stride_w
+                        + kw * dilation_w] * filter_ptr0[kw];
+                    vo0[2] += in_ptr_base[in_offset + 2 * stride_w
+                        + kw * dilation_w] * filter_ptr0[kw];
+                    vo0[3] += in_ptr_base[in_offset + 3 * stride_w
+                        + kw * dilation_w] * filter_ptr0[kw];
+                    // outch 1
+                    vo1[0] += in_ptr_base[in_offset
+                        + kw * dilation_w] * filter_ptr1[kw];
+                    vo1[1] += in_ptr_base[in_offset + stride_w
+                        + kw * dilation_w] * filter_ptr1[kw];
+                    vo1[2] += in_ptr_base[in_offset + 2 * stride_w
+                        + kw * dilation_w] * filter_ptr1[kw];
+                    vo1[3] += in_ptr_base[in_offset + 3 * stride_w
+                        + kw * dilation_w] * filter_ptr1[kw];
+                    // outch 2
+                    vo2[0] += in_ptr_base[in_offset
+                        + kw * dilation_w] * filter_ptr2[kw];
+                    vo2[1] += in_ptr_base[in_offset + stride_w
+                        + kw * dilation_w] * filter_ptr2[kw];
+                    vo2[2] += in_ptr_base[in_offset + 2 * stride_w
+                        + kw * dilation_w] * filter_ptr2[kw];
+                    vo2[3] += in_ptr_base[in_offset + 3 * stride_w
+                        + kw * dilation_w] * filter_ptr2[kw];
+                    // outch 3
+                    vo3[0] += in_ptr_base[in_offset
+                        + kw * dilation_w] * filter_ptr3[kw];
+                    vo3[1] += in_ptr_base[in_offset + stride_w
+                        + kw * dilation_w] * filter_ptr3[kw];
+                    vo3[2] += in_ptr_base[in_offset + 2 * stride_w
+                        + kw * dilation_w] * filter_ptr3[kw];
+                    vo3[3] += in_ptr_base[in_offset + 3 * stride_w
+                        + kw * dilation_w] * filter_ptr3[kw];
+                  }  // kw
+
+                  in_offset += dilation_h * in_width;
+                  filter_ptr0 += filter_width;
+                  filter_ptr1 += filter_width;
+                  filter_ptr2 += filter_width;
+                  filter_ptr3 += filter_width;
+                }  // kh
+
+                for (index_t ow = 0; ow < 4; ++ow) {
+                  out_ptr0_base[out_offset + ow] = vo0[ow];
+                  out_ptr1_base[out_offset + ow] = vo1[ow];
+                  out_ptr2_base[out_offset + ow] = vo2[ow];
+                  out_ptr3_base[out_offset + ow] = vo3[ow];
+                }
+
+                filter_ptr0 -= filter_size;
+                filter_ptr1 -= filter_size;
+                filter_ptr2 -= filter_size;
+                filter_ptr3 -= filter_size;
+              }  // w
+            }  // h
+          }  // c
+        } else {
+          for (index_t mm = m; mm < out_channels; ++mm) {
+            float *out_ptr0_base =
+                output + b * out_batch_size + mm * out_image_size;
+            for (index_t c = 0; c < in_channels; ++c) {
+              const float *in_ptr_base =
+                  input + b * in_batch_size + c * in_image_size;
+              const float *filter_ptr0 =
+                  filter + mm * in_channels * filter_size + c * filter_size;
+
+              for (index_t h = 0; h < out_height; ++h) {
+                for (index_t w = 0; w + 3 < out_width; w += 4) {
+                  // input offset
+                  index_t ih = h * stride_h;
+                  index_t iw = w * stride_w;
+                  index_t in_offset = ih * in_width + iw;
+                  // output (1 outch x 1 height x 4 width): vo_outch_height
+                  float vo0[4];
+                  // load output
+                  index_t out_offset = h * out_width + w;
+                  for (index_t ow = 0; ow < 4; ++ow) {
+                     vo0[ow] = out_ptr0_base[out_offset + ow];
+                  }
+
+                  // calc by row
+                  for (index_t kh = 0; kh < filter_height; ++kh) {
+                    for (index_t kw = 0; kw < filter_width; ++kw) {
+                      // outch 0
+                      vo0[0] += in_ptr_base[in_offset
+                          + kw * dilation_w] * filter_ptr0[kw];
+                      vo0[1] += in_ptr_base[in_offset + stride_w
+                          + kw * dilation_w] * filter_ptr0[kw];
+                      vo0[2] += in_ptr_base[in_offset + 2 * stride_w
+                          + kw * dilation_w] * filter_ptr0[kw];
+                      vo0[3] += in_ptr_base[in_offset + 3 * stride_w
+                          + kw * dilation_w] * filter_ptr0[kw];
+                    }  // kw
+
+                    in_offset += dilation_h * in_width;
+                    filter_ptr0 += filter_width;
+                  }  // kh
+
+                  for (index_t ow = 0; ow < 4; ++ow) {
+                    out_ptr0_base[out_offset + ow] = vo0[ow];
+                  }
+                  filter_ptr0 -= filter_size;
+                }  // w
+              }  // h
+            }  // c
+          }  // mm
+        }  // if
+      }  // m
+    }  // b
   }
 
   void operator()(const Tensor *input,
@@ -286,63 +421,15 @@ struct Conv2dFunctor<DeviceType::CPU, float> : Conv2dFunctorBase {
       if (extra_input_width != padded_input_width) {
         pad_right += (extra_input_width - padded_input_width);
       }
-    } else if (use_neon_3x3_s2) {
-      extra_output_height = height;
-      extra_input_height =
-        std::max(padded_input_height, (extra_output_height - 1) * 2 + 3);
-      extra_output_width = RoundUp<index_t>(width, 4);
-      extra_input_width =
-        std::max(padded_input_width, (extra_output_width - 1) * 2 + 3);
-      if (extra_input_height != padded_input_height) {
-        pad_bottom += (extra_input_height - padded_input_height);
-      }
-      if (extra_input_width != padded_input_width) {
-        pad_right += (extra_input_width - padded_input_width);
-      }
-    } else if (use_neon_5x5_s1) {
-      extra_output_height = height;
-      extra_input_height =
-          std::max(padded_input_height, extra_output_height + 4);
-      extra_output_width = RoundUp<index_t>(width, 4);
-      extra_input_width = std::max(padded_input_width, extra_output_width + 4);
-      if (extra_input_height != padded_input_height) {
-        pad_bottom += (extra_input_height - padded_input_height);
-      }
-      if (extra_input_width != padded_input_width) {
-        pad_right += (extra_input_width - padded_input_width);
-      }
-    } else if (use_neon_7x7_s1) {
-      extra_output_height = height;
-      extra_input_height =
-          std::max(padded_input_height, extra_output_height + 6);
-      extra_output_width = RoundUp<index_t>(width, 4);
-      extra_input_width = std::max(padded_input_width, extra_output_width + 6);
-      if (extra_input_height != padded_input_height) {
-        pad_bottom += (extra_input_height - padded_input_height);
-      }
-      if (extra_input_width != padded_input_width) {
-        pad_right += (extra_input_width - padded_input_width);
-      }
-    } else if (use_neon_7x7_s2) {
-      extra_output_height = height;
-      extra_input_height =
-          std::max(padded_input_height, (extra_output_height - 1) * 2 + 7);
-      extra_output_width = RoundUp<index_t>(width, 4);
-      extra_input_width =
-          std::max(padded_input_width, (extra_output_width - 1) * 2 + 7);
-      if (extra_input_height != padded_input_height) {
-        pad_bottom += (extra_input_height - padded_input_height);
-      }
-      if (extra_input_width != padded_input_width) {
-        pad_right += (extra_input_width - padded_input_width);
-      }
-    } else if (use_neon_7x7_s3) {
+    } else {
       extra_output_height = height;
       extra_input_height =
-          std::max(padded_input_height, (extra_output_height - 1) * 3 + 7);
+          std::max(padded_input_height, (extra_output_height - 1) * stride_h
+              + (filter_h - 1) * dilation_h + 1);
       extra_output_width = RoundUp<index_t>(width, 4);
       extra_input_width =
-          std::max(padded_input_width, (extra_output_width - 1) * 3 + 7);
+          std::max(padded_input_width, (extra_output_width - 1) * stride_w
+              + (filter_w - 1) * dilation_w + 1);
       if (extra_input_height != padded_input_height) {
         pad_bottom += (extra_input_height - padded_input_height);
       }