Change the default device priority for non-CPU/GPU devices to be

lower than both CPU and GPU, to prevent accidentally placing
ops on devices.

Most ops are implemented for either CPU and GPU, and this default can
slow down models that do not carefully place all of their ops
explicitly.

RELNOTES: non-CPU/GPU devices are no longer automatically preferred
over CPU/GPU devices: they must be explicitly named in the nodes to be
used.
Change: 140748128

2 files changed, 19 insertions, 10 deletions

diff --git a/tensorflow/core/common_runtime/device_set.cc b/tensorflow/core/common_runtime/device_set.cc
index 8ff93760d4..84f435b022 100644
--- a/tensorflow/core/common_runtime/device_set.cc
+++ b/tensorflow/core/common_runtime/device_set.cc
@@ -54,10 +54,15 @@ Device* DeviceSet::FindDeviceByName(const string& name) const {
 int DeviceSet::DeviceTypeOrder(const DeviceType& d) {
   if (StringPiece(d.type()) == DEVICE_CPU) {
     return 3;
-  } else if (StringPiece(d.type()) == DEVICE_GPU) {
+  } else if (StringPiece(d.type()) == DEVICE_GPU ||
+             StringPiece(d.type()) == DEVICE_SYCL) {
     return 2;
   } else {
-    return 1;
+    // Non-CPU/GPU devices are never prioritized over CPU and GPU, and
+    // must be explicitly selected.  This is to prevent surprising
+    // placements that cause a lot of cross-device communication
+    // between the host CPU device and other devices.
+    return 10;
   }
 }
 
diff --git a/tensorflow/core/common_runtime/device_set_test.cc b/tensorflow/core/common_runtime/device_set_test.cc
index 8ca744c004..7c59b43489 100644
--- a/tensorflow/core/common_runtime/device_set_test.cc
+++ b/tensorflow/core/common_runtime/device_set_test.cc
@@ -69,18 +69,22 @@ TEST_F(DeviceSetTest, PrioritizedDeviceTypeList) {
       types());
 
   AddDevice("SYCL", "/job:a/replica:0/task:0/device:sycl:0");
-  EXPECT_EQ(
-      (std::vector<DeviceType>{DeviceType(DEVICE_SYCL), DeviceType(DEVICE_GPU),
-                               DeviceType(DEVICE_CPU)}),
-      types());
+  EXPECT_TRUE((types()[0] == DeviceType(DEVICE_SYCL) ||
+               types()[0] == DeviceType(DEVICE_GPU)));
+  EXPECT_TRUE((types()[1] == DeviceType(DEVICE_SYCL) ||
+               types()[1] == DeviceType(DEVICE_GPU)));
+  EXPECT_TRUE(types()[2] == DeviceType(DEVICE_CPU));
 
   AddDevice("T1", "/job:a/replica:0/task:0/device:T1:0");
   AddDevice("T1", "/job:a/replica:0/task:0/device:T1:1");
   AddDevice("T2", "/job:a/replica:0/task:0/device:T2:0");
-  EXPECT_EQ((std::vector<DeviceType>{DeviceType(DEVICE_SYCL), DeviceType("T1"),
-                                     DeviceType("T2"), DeviceType(DEVICE_GPU),
-                                     DeviceType(DEVICE_CPU)}),
-            types());
+  EXPECT_TRUE((types()[0] == DeviceType(DEVICE_SYCL) ||
+               types()[0] == DeviceType(DEVICE_GPU)));
+  EXPECT_TRUE((types()[1] == DeviceType(DEVICE_SYCL) ||
+               types()[1] == DeviceType(DEVICE_GPU)));
+  EXPECT_TRUE(types()[2] == DeviceType(DEVICE_CPU));
+  EXPECT_TRUE(types()[3] == DeviceType("T1"));
+  EXPECT_TRUE(types()[4] == DeviceType("T2"));
 }
 
 }  // namespace