Fixes #651, #653:

* Can now (easily) access the default stream of a context
* Can now launch a kernel with `context_t::launch()`, implicitly using the context's default stream

src/cuda/api/context.hpp

@@ -404,6 +404,14 @@ class context_t {
 
 public: // other non-mutator methods
 
+	stream_t default_stream() const;
+
+	template <typename Kernel, typename ... KernelParameters>
+	void launch(
+		Kernel                  kernel,
+		launch_configuration_t  launch_configuration,
+		KernelParameters...     parameters) const;
+
 	/**
 	 * Determines the balance between L1 space and shared memory space set
 	 * for kernels executing within this context.

src/cuda/api/multi_wrapper_impls/context.hpp

@@ -313,6 +313,20 @@ inline event_t context_t::create_event(
 		uses_blocking_sync, records_timing, interprocess);
 }
 
+inline stream_t context_t::default_stream() const
+{
+	return stream::wrap(device_id_, handle_, stream::default_stream_handle, do_not_take_ownership);
+}
+
+template <typename Kernel, typename ... KernelParameters>
+void context_t::launch(
+	Kernel                  kernel,
+	launch_configuration_t  launch_configuration,
+	KernelParameters...     parameters) const
+{
+	default_stream().enqueue.kernel_launch(kernel, launch_configuration, parameters...);
+}
+
 } // namespace cuda
 
 #endif // MULTI_WRAPPER_IMPLS_CONTEXT_HPP_

src/cuda/api/multi_wrapper_impls/device.hpp

@@ -177,7 +177,7 @@ void device_t::launch(
 	KernelParameters...     parameters) const
 {
 	auto pc = primary_context();
-	pc.default_stream().enqueue.kernel_launch(kernel, launch_configuration, parameters...);
+	pc.launch(kernel, launch_configuration, parameters...);
 }
 
 inline context_t device_t::create_context(