3.6.0 update (#2005)

* 3.6.0 update

* doc and swap stuff

---------

Co-authored-by: yuzhai <yuzhai@nvidia.com>
Co-authored-by: Haicheng Wu <haichengw@nvidia.com>

tools/library/include/cutlass/library/handle.h

@@ -72,6 +72,8 @@ private:
   /// Pointer to the most recently executed operation
   Operation const *last_operation_;
 
+  int device_idx_;
+
 public:
 
   /// Constructor

tools/library/include/cutlass/library/library.h

@@ -103,7 +103,6 @@ public:
     void *device_workspace = nullptr,
     cudaStream_t stream = nullptr) const = 0;
 
-  // Originally designed for metadata, but should be useful for FP8/6/4 too.
   virtual Status initialize_with_profiler_workspace(
     void const *configuration,
     void *host_workspace,
@@ -118,7 +117,8 @@ public:
     void const *arguments,
     void *host_workspace,
     void *device_workspace = nullptr,
-    cudaStream_t stream = nullptr) const = 0;
+    cudaStream_t stream = nullptr,
+    bool launch_with_pdl = false) const = 0;
 
 };
 
@@ -272,6 +272,8 @@ struct GemmUniversalConfiguration {
   int64_t ldb{0};
   int64_t ldc{0};
   int64_t ldd{0};
+
+  int device_count{1};
 };
 
 struct GemmUniversalArguments {
@@ -303,6 +305,8 @@ struct GemmUniversalArguments {
   int sm_count{0};
   library::RasterOrder raster_order{};
   int swizzle_size{1};
+
+  int device_index{0};
 };
 
 /////////////////////////////////////////////////////////////////////////////////////////////////