【Error Message BUAA No. 36】paddle/fluid/distributed/ps/service/communicator/communicator.cc (#66733)

tlxd · web-flow · commit 52d640ebf88e · 2024-07-31T10:03:01.000+08:00
* change CHECK to PADDLE_ENFORCE_xx paddle/fluid/distributed/ps/service/communicator/communicator.cc

* add #include 'paddle/fluid/platform/enforce.h' in communicator.cc + change a bug
diff --git a/paddle/fluid/distributed/ps/service/communicator/communicator.cc b/paddle/fluid/distributed/ps/service/communicator/communicator.cc
@@ -19,6 +19,7 @@ limitations under the License. */
 #include "paddle/common/flags.h"
 #include "paddle/fluid/distributed/ps/service/brpc_ps_client.h"
 #include "paddle/fluid/distributed/ps/wrapper/fleet.h"
+#include "paddle/fluid/platform/enforce.h"
 #include "paddle/fluid/platform/profiler.h"
 #include "paddle/utils/string/string_helper.h"
 
@@ -161,7 +162,10 @@ void Communicator::RpcSendDenseParam(const std::vector<std::string> &varnames,
   std::vector<::paddle::distributed::Region> regions;
   for (auto &t : varnames) {
     Variable *var = scope.FindVar(t);
-    CHECK(var != nullptr) << "var[" << t << "] not found";
+    PADDLE_ENFORCE_NE(
+        var,
+        nullptr,
+        phi::errors::InvalidArgument("Param var should not be NULL!"));
     phi::DenseTensor *tensor = var->GetMutable<phi::DenseTensor>();
     if (phi::is_gpu_place(tensor->place())) {
 #ifdef PADDLE_WITH_CUDA
@@ -212,9 +216,13 @@ void Communicator::RpcSendDense(const CommContext &ctx,
         scope.FindVar(var_name)->Get<phi::DenseTensor>();
     size_t count = static_cast<size_t>(tensor.numel());
     const float *g = tensor.data<float>();
-    CHECK(pos + count <= dense_data->size())
-        << "invalid dense size, cur pos[" << pos << "]"
-        << " data_num[" << count << "] size[" << dense_data->size() << "]";
+    PADDLE_ENFORCE_LE(
+        pos + count,
+        dense_data->size(),
+        phi ::errors::InvalidArgument("Param pos add count should be less than "
+                                      "or equal to %d, but got %d.",
+                                      dense_data->size(),
+                                      pos + count));
     memcpy(data + pos, g, count * sizeof(float));
     pos += count;
   }
@@ -614,13 +622,28 @@ void AsyncCommunicator::PullSparseToTensorSync(
     for (size_t i = 0; i < len; ++i, output_len += fea_dim) {
       if (!output || output_len == size_t(output->numel())) {
         ++output_index;
-        CHECK(output_index < outputs->size());  // NOLINT
+        PADDLE_ENFORCE_LT(
+            output_index,
+            outputs->size(),
+            phi::errors::InvalidArgument(
+                "The output index should be less than %d, but got %d.",
+                outputs->size(),
+                output_index));  // NOLINT
         output = outputs->at(output_index);
         output->set_lod(tensor->lod());
         output_data = output->mutable_data<float>(place);
         output_len = 0;
-        CHECK(output->numel() % fea_dim == 0);  // NOLINT
-        CHECK(output_data != nullptr);          // NOLINT
+        PADDLE_ENFORCE_EQ(
+            output->numel() % fea_dim,
+            0,
+            phi::errors::InvalidArgument("The 'output->numel() % fea_dim' "
+                                         "should be equal to 0, but got %d.",
+                                         output->numel() % fea_dim));  // NOLINT
+        PADDLE_ENFORCE_NE(
+            output_data,
+            nullptr,
+            phi::errors::InvalidArgument(
+                "The output data should not be NULL!"));  // NOLINT
       }
       uint64_t real_id = static_cast<uint64_t>(ids[i]);
       if (real_id == padding_id) {
@@ -663,21 +686,41 @@ void AsyncCommunicator::PushSparseFromTensorAsync(
     if (batch_size == -1) {
       batch_size = cur_batch_size;
     } else if (batch_size != cur_batch_size) {
-      // CHECK(batch_size == cur_batch_size);  // NOLINT
       batch_size_consist = false;
       break;
     }
   }
-  CHECK(batch_size > 0);  // NOLINT
+  PADDLE_ENFORCE_GT(batch_size,
+                    0,
+                    phi::errors::InvalidArgument(
+                        "The batch size should be greater than 0, but got %d.",
+                        batch_size));  // NOLINT
 
   int show_size =
       !shows->lod().empty() ? shows->lod()[0].size() - 1 : shows->dims()[0];
-  CHECK(show_size == batch_size || show_size == 1);
+  PADDLE_ENFORCE_EQ(
+      show_size == batch_size || show_size == 1,
+      true,
+      phi::errors::InvalidArgument("The show size should be equal to batch "
+                                   "size or equal to 1, but got %d.",
+                                   show_size));
   int clk_size =
       !clks->lod().empty() ? clks->lod()[0].size() - 1 : clks->dims()[0];
-  CHECK(clk_size == batch_size || clk_size == 1);
+  PADDLE_ENFORCE_EQ(
+      clk_size == batch_size || clk_size == 1,
+      true,
+      phi::errors::InvalidArgument("The clk size should be equal to batch size "
+                                   "or equal to 1, but got %d.",
+                                   clk_size));
 
-  CHECK(outputs->size() == inputs->size());
+  PADDLE_ENFORCE_EQ(
+      outputs->size(),
+      inputs->size(),
+      phi::errors::InvalidArgument(
+          "The size of outputs should be equal to inputs, but the size of "
+          "outputs is %d, the size of inputs is %d",
+          outputs->size(),
+          inputs->size()));
   std::vector<uint64_t> push_keys;
   push_keys.reserve(MAX_FEASIGN_NUM / 100);
   std::vector<std::vector<float>> push_values;
@@ -758,7 +801,13 @@ void AsyncCommunicator::PushSparseFromTensorAsync(
         ++input_idx;
       }
     }
-    CHECK(static_cast<int64_t>(output_len) == g_tensor->numel());
+    PADDLE_ENFORCE_EQ(
+        static_cast<int64_t>(output_len),
+        g_tensor->numel(),
+        phi::errors::InvalidArgument(
+            "The output length should be equal to %d, but got %d.",
+            g_tensor->numel(),
+            static_cast<int64_t>(output_len)));
   }
 
   std::vector<float *> push_g_vec(input_idx, nullptr);
