switch the launch argument order

cuda_core/cuda/core/experimental/_launcher.py

@@ -107,17 +107,17 @@ def _cast_to_3_tuple(self, cfg):
             raise ValueError
 
 
-def launch(kernel, config, *kernel_args):
+def launch(config, kernel, *kernel_args):
     """Launches a :obj:`~_module.Kernel`
     object with launch-time configuration.
 
     Parameters
     ----------
-    kernel : :obj:`~_module.Kernel`
-        Kernel to launch.
     config : :obj:`~_launcher.LaunchConfig`
         Launch configurations inline with options provided by
         :obj:`~_launcher.LaunchConfig` dataclass.
+    kernel : :obj:`~_module.Kernel`
+        Kernel to launch.
     *kernel_args : Any
         Variable length argument list that is provided to the
         launching kernel.

cuda_core/examples/saxpy.py

@@ -59,7 +59,7 @@
 ker_args = (a, x.data.ptr, y.data.ptr, out.data.ptr, size)
 
 # launch kernel on stream s
-launch(ker, config, *ker_args)
+launch(config, ker, *ker_args)
 s.sync()
 
 # check result
@@ -90,7 +90,7 @@
 ker_args = (a, x.data.ptr, y.data.ptr, buf, size)
 
 # launch kernel on stream s
-launch(ker, config, *ker_args)
+launch(config, ker, *ker_args)
 s.sync()
 
 # check result

cuda_core/examples/strided_memory_view.py

@@ -127,7 +127,7 @@ def my_func(arr, work_stream):
         block = 256
         grid = (size + block - 1) // block
         config = LaunchConfig(grid=grid, block=block, stream=work_stream)
-        launch(gpu_ker, config, view.ptr, np.uint64(size))
+        launch(config, gpu_ker, view.ptr, np.uint64(size))
         # Here we're being conservative and synchronize over our work stream,
         # assuming we do not know the data stream; if we know then we could
         # just order the data stream after the work stream here, e.g.

cuda_core/examples/thread_block_cluster.py

@@ -59,7 +59,7 @@
 config = LaunchConfig(grid=grid, cluster=cluster, block=block, stream=dev.default_stream)
 
 # launch kernel on the default stream
-launch(ker, config)
+launch(config, ker)
 dev.sync()
 
 print("done!")

cuda_core/examples/vector_add.py

@@ -55,7 +55,7 @@
 config = LaunchConfig(grid=grid, block=block, stream=s)
 
 # launch kernel on stream s
-launch(ker, config, a.data.ptr, b.data.ptr, c.data.ptr, cp.uint64(size))
+launch(config, ker, a.data.ptr, b.data.ptr, c.data.ptr, cp.uint64(size))
 s.sync()
 
 # check result