repair npu matmul_grad and comm_init_hccl (#33719)

paddle/fluid/operators/collective/c_comm_init_hccl_op.cc

@@ -22,7 +22,11 @@ class Scope;
 }  // namespace framework
 }  // namespace paddle
 #if defined(PADDLE_WITH_ASCEND_CL)
+#include "acl/acl.h"
+#include "hccl/hccl.h"
+#include "hccl/hccl_types.h"
 #include "paddle/fluid/platform/collective_helper.h"
+#include "paddle/fluid/platform/hccl_helper.h"
 #endif
 
 namespace paddle {
@@ -57,6 +61,31 @@ class CCommInitOpAscend : public framework::OperatorBase {
     }
     platform::HCCLCommContext::Instance().CreateHCCLComm(
         hccl_id, rank_ids, rank_id, device_id, rid);
+
+    //  Build comm
+    float* buff;
+    int32_t size = 20;
+    std::vector<float> input(size, 0);
+    for (int32_t idx = 0; idx < size; idx++) {
+      input[idx] = 1.0;
+    }
+    aclrtMalloc(reinterpret_cast<void**>(&buff), size * sizeof(float),
+                ACL_MEM_MALLOC_HUGE_FIRST);
+    aclrtMemcpy(reinterpret_cast<void*>(buff), size * sizeof(float),
+                input.data(), size * sizeof(float), ACL_MEMCPY_HOST_TO_DEVICE);
+    VLOG(3) << "Build buff data successful.";
+
+    aclrtStream stream = nullptr;
+    auto comm = paddle::platform::HCCLCommContext::Instance().Get(rid, place);
+    if (rank_id == 0) {
+      stream = comm->stream();
+    } else {
+      auto dev_ctx = platform::DeviceContextPool::Instance().Get(place);
+      stream = static_cast<platform::NPUDeviceContext*>(dev_ctx)->stream();
+    }
+    platform::dynload::HcclBroadcast(buff, size, HCCL_DATA_TYPE_FP32, 0,
+                                     comm->comm(), stream);
+    VLOG(3) << "Build connection successful.";
 #else
     PADDLE_THROW(platform::errors::PreconditionNotMet(
         "PaddlePaddle should compile with NPU."));

paddle/fluid/operators/matmul_v2_op_npu.cc

@@ -138,6 +138,29 @@ class MatMulV2GradNPUKernel : public framework::OpKernel<T> {
         }
         if (dy) {
           dy->mutable_data<T>(ctx.GetPlace());
+          if ((x->dims().size() == 3) && (dout->dims().size() == 3) &&
+              (dy->dims().size() == 2)) {
+            framework::Tensor dout_;
+            TensorCopy(*dout, ctx.GetPlace(), &dout_);
+            ctx.template device_context<paddle::platform::NPUDeviceContext>()
+                .Wait();
+            std::vector<int> vec_dim = framework::vectorize<int>(dout_.dims());
+            std::vector<int> vec_dim_v{vec_dim[0] * vec_dim[1], vec_dim[2]};
+            dout_.Resize(framework::make_ddim(vec_dim_v));
+
+            framework::Tensor x_;
+            TensorCopy(*x, ctx.GetPlace(), &x_);
+            ctx.template device_context<paddle::platform::NPUDeviceContext>()
+                .Wait();
+            std::vector<int> vec_dim_x = framework::vectorize<int>(x_.dims());
+            std::vector<int> vec_dim_x_v{vec_dim_x[0] * vec_dim_x[1],
+                                         vec_dim_x[2]};
+            x_.Resize(framework::make_ddim(vec_dim_x_v));
+            const auto& runner_dy =
+                NpuOpRunner("MatMul", {x_, dout_}, {*dy},
+                            {{"transpose_x1", true}, {"transpose_x2", false}});
+            runner_dy.Run(stream);
+          } else {
             const auto& runner_dy =
                 NpuOpRunner("BatchMatMul", {*x, *dout}, {*dy},
                             {{"adj_x1", true}, {"adj_x2", false}});
@@ -146,6 +169,7 @@ class MatMulV2GradNPUKernel : public framework::OpKernel<T> {
         }
       }
     }
+  }
 };
 }  // namespace operators
 }  // namespace paddle

python/paddle/fluid/tests/unittests/npu/test_matmulv2_op_npu.py

@@ -206,5 +206,85 @@ class TestMatMulNet(unittest.TestCase):
         self.assertTrue(np.allclose(npu_loss, cpu_loss))
 
 
+# The precision is aligned in NPU and GPU separately, which is only used for the usage method.
+@unittest.skipIf(not paddle.is_compiled_with_npu(),
+                 "core is not compiled with NPU")
+class TestMatMulNet3_2(unittest.TestCase):
+    def _test(self, run_npu=True):
+        main_prog = paddle.static.Program()
+        startup_prog = paddle.static.Program()
+        main_prog.random_seed = SEED
+        startup_prog.random_seed = SEED
+        np.random.seed(SEED)
+        self._dtype = "float32"
+
+        a_np = np.random.random(size=(2, 1, 3)).astype(self._dtype)
+        b_np = np.random.random(size=(2, 1, 3)).astype(self._dtype)
+        c_np = np.random.random(size=(3, 2)).astype(self._dtype)
+        d_np = np.random.random(size=(3, 2)).astype(self._dtype)
+        label_np = np.random.randint(2, size=(2, 1)).astype('int64')
+
+        with paddle.static.program_guard(main_prog, startup_prog):
+            a = paddle.static.data(name="a", shape=[2, 1, 3], dtype=self._dtype)
+            b = paddle.static.data(name="b", shape=[2, 1, 3], dtype=self._dtype)
+            c = paddle.static.data(name="c", shape=[3, 2], dtype=self._dtype)
+            d = paddle.static.data(name="d", shape=[3, 2], dtype=self._dtype)
+            label = paddle.static.data(
+                name="label", shape=[2, 1], dtype='int64')
+
+            sum_1 = paddle.add(a, b)
+            sum_2 = paddle.add(c, d)
+            sum_1 = paddle.cast(sum_1, 'float16')
+            sum_2 = paddle.cast(sum_2, 'float16')
+            if not run_npu:
+                sum_1 = paddle.cast(sum_1, 'float32')
+                sum_2 = paddle.cast(sum_2, 'float32')
+
+            result = paddle.matmul(sum_1, sum_2)
+            if run_npu:
+                result = paddle.cast(result, 'float32')
+
+            result = paddle.reshape(result, shape=[2, 2])
+            fc_1 = fluid.layers.fc(input=result, size=8)
+            prediction = fluid.layers.fc(input=fc_1, size=2, act='softmax')
+
+            cost = fluid.layers.cross_entropy(input=prediction, label=label)
+            loss = fluid.layers.reduce_mean(cost)
+            sgd = fluid.optimizer.SGD(learning_rate=0.01)
+            sgd.minimize(loss)
+
+        if run_npu:
+            place = paddle.NPUPlace(0)
+        else:
+            place = paddle.CPUPlace()
+        exe = paddle.static.Executor(place)
+        exe.run(startup_prog)
+
+        print("Start run on {}".format(place))
+        for epoch in range(100):
+
+            pred_res, loss_res = exe.run(main_prog,
+                                         feed={
+                                             "a": a_np,
+                                             "b": b_np,
+                                             "c": c_np,
+                                             "d": d_np,
+                                             "label": label_np
+                                         },
+                                         fetch_list=[prediction, loss])
+            if epoch % 10 == 0:
+                print("Epoch {} | Prediction[0]: {}, Loss: {}".format(
+                    epoch, pred_res[0], loss_res))
+
+        return pred_res, loss_res
+
+    def test_npu(self):
+        cpu_pred, cpu_loss = self._test(False)
+        npu_pred, npu_loss = self._test(True)
+
+        self.assertTrue(np.allclose(npu_pred, cpu_pred, atol=1e-4))
+        self.assertTrue(np.allclose(npu_loss, cpu_loss, atol=1e-4))
+
+
 if __name__ == '__main__':
     unittest.main()