[SE] Platforms return Device values

[jhen]

Summary

Platforms were returning Device pointers, but a Device is now basically
just a pointer to an underlying PlatformDevice, so we will now just pass
it around as a value.

[jlebar]

Perhaps it's worth a comment on the Device class that Device objects should be passed by value, and are cheap to copy?

[jhen]

• Device doc says device is small

[jhen]

In D24537#541929, @jlebar wrote:

Perhaps it's worth a comment on the Device class that Device objects should be passed by value, and are cheap to copy?

Good idea! Done

streamexecutor/examples/CUDASaxpy.cpp

int main() {
   size_t ArraySize = HostX.size();
 
   // Get a device object.
   se::Platform *Platform =
       getOrDie(se::PlatformManager::getPlatformByName("CUDA"));
   if (Platform->getDeviceCount() == 0) {
     return EXIT_FAILURE;
   }
-  se::Device *Device = getOrDie(Platform->getDevice(0));
+  se::Device Device = getOrDie(Platform->getDevice(0));
 
   // Load the kernel onto the device.
   cg::SaxpyKernel Kernel =
-      getOrDie(Device->createKernel<cg::SaxpyKernel>(cg::SaxpyLoaderSpec));
+      getOrDie(Device.createKernel<cg::SaxpyKernel>(cg::SaxpyLoaderSpec));
 
   se::RegisteredHostMemory<float> RegisteredX =
-      getOrDie(Device->registerHostMemory<float>(HostX));
+      getOrDie(Device.registerHostMemory<float>(HostX));
   se::RegisteredHostMemory<float> RegisteredY =
-      getOrDie(Device->registerHostMemory<float>(HostY));
+      getOrDie(Device.registerHostMemory<float>(HostY));
 
   // Allocate memory on the device.
   se::GlobalDeviceMemory<float> X =
-      getOrDie(Device->allocateDeviceMemory<float>(ArraySize));
+      getOrDie(Device.allocateDeviceMemory<float>(ArraySize));
   se::GlobalDeviceMemory<float> Y =
-      getOrDie(Device->allocateDeviceMemory<float>(ArraySize));
+      getOrDie(Device.allocateDeviceMemory<float>(ArraySize));
 
   // Run operations on a stream.
-  se::Stream Stream = getOrDie(Device->createStream());
+  se::Stream Stream = getOrDie(Device.createStream());
   Stream.thenCopyH2D(RegisteredX, X)
       .thenCopyH2D(RegisteredY, Y)
       .thenLaunch(ArraySize, 1, Kernel, A, X, Y)
       .thenCopyD2H(X, RegisteredX);
   // Wait for the stream to complete.
   se::dieIfError(Stream.blockHostUntilDone());
 
   // Process output data in HostX.
   std::vector<float> ExpectedX = {4, 47, 90, 133};
   assert(std::equal(ExpectedX.begin(), ExpectedX.end(), HostX.begin()));
   /// [Example saxpy host main]
 }

streamexecutor/examples/HostSaxpy.cpp

int main() {
   size_t ArraySize = HostX.size();
 
   // Get a device object.
   se::Platform *Platform =
       getOrDie(se::PlatformManager::getPlatformByName("host"));
   if (Platform->getDeviceCount() == 0) {
     return EXIT_FAILURE;
   }
-  se::Device *Device = getOrDie(Platform->getDevice(0));
+  se::Device Device = getOrDie(Platform->getDevice(0));
 
   // Load the kernel onto the device.
   cg::SaxpyKernel Kernel =
-      getOrDie(Device->createKernel<cg::SaxpyKernel>(cg::SaxpyLoaderSpec));
+      getOrDie(Device.createKernel<cg::SaxpyKernel>(cg::SaxpyLoaderSpec));
 
   se::RegisteredHostMemory<float> RegisteredX =
-      getOrDie(Device->registerHostMemory<float>(HostX));
+      getOrDie(Device.registerHostMemory<float>(HostX));
   se::RegisteredHostMemory<float> RegisteredY =
-      getOrDie(Device->registerHostMemory<float>(HostY));
+      getOrDie(Device.registerHostMemory<float>(HostY));
 
   // Allocate memory on the device.
   se::GlobalDeviceMemory<float> X =
-      getOrDie(Device->allocateDeviceMemory<float>(ArraySize));
+      getOrDie(Device.allocateDeviceMemory<float>(ArraySize));
   se::GlobalDeviceMemory<float> Y =
-      getOrDie(Device->allocateDeviceMemory<float>(ArraySize));
+      getOrDie(Device.allocateDeviceMemory<float>(ArraySize));
 
   // Run operations on a stream.
-  se::Stream Stream = getOrDie(Device->createStream());
+  se::Stream Stream = getOrDie(Device.createStream());
   Stream.thenCopyH2D(RegisteredX, X)
       .thenCopyH2D(RegisteredY, Y)
       .thenLaunch(1, 1, Kernel, A, X, Y, ArraySize)
       .thenCopyD2H(X, RegisteredX);
   // Wait for the stream to complete.
   se::dieIfError(Stream.blockHostUntilDone());
 
   // Process output data in HostX.
   std::vector<float> ExpectedX = {4, 47, 90, 133};
   assert(std::equal(ExpectedX.begin(), ExpectedX.end(), HostX.begin()));
 }

streamexecutor/include/streamexecutor/Platform.h

 
 class Platform {
 public:
   virtual ~Platform();
 
   /// Gets the number of devices available for this platform.
   virtual size_t getDeviceCount() const = 0;
 
-  /// Gets a pointer to a Device with the given index for this platform.
-  ///
-  /// Ownership of the Device instance is NOT transferred to the caller.
-  virtual Expected<Device *> getDevice(size_t DeviceIndex) = 0;
+  /// Gets a Device with the given index for this platform.
+  virtual Expected<Device> getDevice(size_t DeviceIndex) = 0;
 };
 
 } // namespace streamexecutor
 
 #endif // STREAMEXECUTOR_PLATFORM_H

streamexecutor/include/streamexecutor/platforms/host/HostPlatform.h

 namespace host {
 
 /// Platform that performs work on the host rather than offloading to an
 /// accelerator.
 class HostPlatform : public Platform {
 public:
   size_t getDeviceCount() const override { return 1; }
 
-  Expected<Device *> getDevice(size_t DeviceIndex) override {
+  Expected<Device> getDevice(size_t DeviceIndex) override {
     if (DeviceIndex != 0) {
       return make_error(
           "Requested device index " + llvm::Twine(DeviceIndex) +
           " from host platform which only supports device index 0");
     }
     llvm::sys::ScopedLock Lock(Mutex);
-    if (!TheDevice) {
+    if (!ThePlatformDevice)
       ThePlatformDevice = llvm::make_unique<HostPlatformDevice>();
-      TheDevice = llvm::make_unique<Device>(ThePlatformDevice.get());
-    }
-    return TheDevice.get();
+    return Device(ThePlatformDevice.get());
   }
 
 private:
   llvm::sys::Mutex Mutex;
   std::unique_ptr<HostPlatformDevice> ThePlatformDevice;
-  std::unique_ptr<Device> TheDevice;
 };
 
 } // namespace host
 } // namespace streamexecutor
 
 #endif // STREAMEXECUTOR_PLATFORMS_HOST_HOSTPLATFORM_H