From 479689f6fb78f2c07755203ecd6b1d7cf47da400 Mon Sep 17 00:00:00 2001
From: Jacek Czaja <jacek.czaja@intel.com>
Date: Fri, 14 May 2021 11:16:16 +0200
Subject: [PATCH] [oneDNN] Refactoring of softmax grad onednn kernel to match
 common API (#32851)

---
 .../operators/mkldnn/softmax_mkldnn_op.cc     | 75 ++++++++-----------
 .../mkldnn/test_softmax_mkldnn_op.py          |  2 +
 2 files changed, 35 insertions(+), 42 deletions(-)

diff --git a/paddle/fluid/operators/mkldnn/softmax_mkldnn_op.cc b/paddle/fluid/operators/mkldnn/softmax_mkldnn_op.cc
index 1138d511392..4a55945936e 100644
--- a/paddle/fluid/operators/mkldnn/softmax_mkldnn_op.cc
+++ b/paddle/fluid/operators/mkldnn/softmax_mkldnn_op.cc
@@ -15,15 +15,6 @@ limitations under the License. */
 #include "paddle/fluid/operators/softmax_op.h"
 #include "paddle/fluid/platform/mkldnn_reuse.h"
 
-namespace paddle {
-namespace framework {
-class Tensor;
-}  // namespace framework
-namespace platform {
-class MKLDNNDeviceContext;
-}  // namespace platform
-}  // namespace paddle
-
 namespace paddle {
 namespace operators {
 
@@ -74,22 +65,34 @@ class SoftmaxMKLDNNHandler
     }
   }
 
-  SoftmaxMKLDNNHandler(const std::vector<int64_t>& dims,
-                       const MKLDNNMemoryFormat fmt,
-                       const MKLDNNMemoryFormat diff_fmt, const int& axis,
-                       const platform::MKLDNNDeviceContext& dev_ctx,
-                       platform::Place cpu_place, const std::string& uniq_name)
+  SoftmaxMKLDNNHandler(const framework::ExecutionContext& ctx,
+                       const MKLDNNDeviceContext& dev_ctx,
+                       platform::Place cpu_place, const Tensor* out,
+                       const Tensor* out_grad, Tensor* in_x_grad,
+                       const std::string& unique_name)
       : platform::MKLDNNHandlerT<T, mkldnn::softmax_forward,
                                  mkldnn::softmax_backward>(
             dev_ctx, dev_ctx.GetEngine(), cpu_place,
-            platform::CreateKey(dev_ctx, dims, uniq_name)) {
-    auto data_softmax_md =
-        mkldnn::memory::desc(dims, platform::MKLDNNGetDataType<T>(), fmt);
-    auto diff_softmax_md =
-        mkldnn::memory::desc(dims, platform::MKLDNNGetDataType<T>(), diff_fmt);
-
-    this->AcquireBackwardPrimitiveDescriptor(diff_softmax_md, data_softmax_md,
-                                             axis);
+            platform::CreateKey(dev_ctx, framework::vectorize(out->dims()),
+                                unique_name)) {
+    if (!this->isBwdCached()) {
+      PADDLE_ENFORCE_EQ(
+          out_grad->dims(), in_x_grad->dims(),
+          platform::errors::InvalidArgument("The shape of softmax_grad's input "
+                                            "and output must be identical."));
+
+      auto dims = out_grad->dims();  // input and output share the same shape
+      const int axis = CanonicalAxis(ctx.Attr<int>("axis"), dims.size());
+      auto softmax_tz = framework::vectorize<int64_t>(dims);
+
+      auto data_softmax_md = MKLDNNMemDesc(
+          softmax_tz, platform::MKLDNNGetDataType<T>(), out->format());
+      auto diff_softmax_md = MKLDNNMemDesc(
+          softmax_tz, platform::MKLDNNGetDataType<T>(), out_grad->format());
+
+      this->AcquireBackwardPrimitiveDescriptor(diff_softmax_md, data_softmax_md,
+                                               axis);
+    }
   }
 };
 
@@ -145,27 +148,15 @@ class SoftmaxMKLDNNGradKernel : public paddle::framework::OpKernel<T> {
                           "Operator DNNL SoftmaxGrad must use CPUPlace"));
     auto& dev_ctx = ctx.template device_context<MKLDNNDeviceContext>();
     const Tensor* output = ctx.Input<Tensor>("Out");
-    auto* dout = ctx.template Input<Tensor>(framework::GradVarName("Out"));
-    auto* dx =
-        ctx.template Output<framework::Tensor>(framework::GradVarName("X"));
-
-    PADDLE_ENFORCE_EQ(
-        dout->dims(), dx->dims(),
-        platform::errors::InvalidArgument(
-            "The shape of softmax_grad's input and output must be identical."));
-
-    auto dims = dout->dims();  // input and output share the same shape
-    const int axis = CanonicalAxis(ctx.Attr<int>("axis"), dims.size());
-
-    auto softmax_tz = paddle::framework::vectorize<int64_t>(dims);
+    auto* out_grad = ctx.template Input<Tensor>(framework::GradVarName("Out"));
+    auto* in_x_grad = ctx.template Output<Tensor>(framework::GradVarName("X"));
 
-    SoftmaxMKLDNNHandler<T> handler(softmax_tz, output->format(),
-                                    dout->format(), axis, dev_ctx,
-                                    ctx.GetPlace(), ctx.InputName("Out"));
+    SoftmaxMKLDNNHandler<T> handler(ctx, dev_ctx, ctx.GetPlace(), output,
+                                    out_grad, in_x_grad, ctx.InputName("Out"));
 
     auto dst_memory_p = handler.AcquireDstMemory(output);
-    auto diff_dst_memory_p = handler.AcquireDiffDstMemory(dout);
-    auto diff_src_memory_p = handler.AcquireDiffSrcMemory(dx);
+    auto diff_dst_memory_p = handler.AcquireDiffDstMemory(out_grad);
+    auto diff_src_memory_p = handler.AcquireDiffSrcMemory(in_x_grad);
 
     auto softmax_bwd_p = handler.AcquireBackwardPrimitive();
 
@@ -176,8 +167,8 @@ class SoftmaxMKLDNNGradKernel : public paddle::framework::OpKernel<T> {
                             {MKLDNN_ARG_DIFF_SRC, *diff_src_memory_p}});
     astream.wait();
 
-    dx->set_layout(framework::DataLayout::kMKLDNN);
-    dx->set_format(dout->format());
+    in_x_grad->set_layout(framework::DataLayout::kMKLDNN);
+    in_x_grad->set_format(platform::GetMKLDNNFormat(*diff_src_memory_p));
   }
 };
 }  // namespace operators
diff --git a/python/paddle/fluid/tests/unittests/mkldnn/test_softmax_mkldnn_op.py b/python/paddle/fluid/tests/unittests/mkldnn/test_softmax_mkldnn_op.py
index 9e2229cece7..13c1883af61 100644
--- a/python/paddle/fluid/tests/unittests/mkldnn/test_softmax_mkldnn_op.py
+++ b/python/paddle/fluid/tests/unittests/mkldnn/test_softmax_mkldnn_op.py
@@ -129,4 +129,6 @@ class TestSoftmaxMKLDNNPrimitivesAlreadyExist(unittest.TestCase):
 
 
 if __name__ == '__main__':
+    from paddle import enable_static
+    enable_static()
     unittest.main()
-- 
GitLab