Merge commit '30480288' into conv_quant_int8

include/ck/tensor_operation/gpu/device/device_batched_gemm_multi_d_xdl.hpp → include/ck/tensor_operation/gpu/device/impl/device_batched_gemm_multi_d_xdl.hpp

@@ -38,9 +38,9 @@ namespace device {
  * \note Using \p ComputePtrOffsetOfBatch gives us the flexibility that 2 workgroups can compute 2
  * tiles from different matrices. Keep in mind that these 2 matrices can share the same grid
  * descriptor (like in BatchedGEMM), or use their own grid descriptors (in GroupedGemm). \link
- * device_conv3d_fwd_xdl_ndhwc_kzyxc_ndhwk.hpp kernel_gemm_xdlops_v2r3_for_conv3d \endlink for \link
- * DeviceConv3d \endlink uses the same concept, but currently does NOT encapsulate the computing of
- * pointer offset into \p ComputePtrOffsetOfStridedBatch.
+ * impl/device_conv3d_fwd_xdl_ndhwc_kzyxc_ndhwk.hpp kernel_gemm_xdlops_v2r3_for_conv3d \endlink for
+ * \link DeviceConv3d \endlink uses the same concept, but currently does NOT encapsulate the
+ * computing of pointer offset into \p ComputePtrOffsetOfStridedBatch.
  *
  * \note \p Block2ETileMap allows customized mapping between a workgroup and the C-tile it computes.
  * Together with \p ComputePtrOffsetOfBatch, we can reuse GridwiseGemm (and GridwiseGemm fusion ) to

include/ck/tensor_operation/gpu/device/device_batched_gemm_softmax_gemm_permute_xdl_cshuffle.hpp → include/ck/tensor_operation/gpu/device/impl/device_batched_gemm_softmax_gemm_permute_xdl_cshuffle.hpp

@@ -9,10 +9,10 @@
 #include "ck/utility/common_header.hpp"
 #include "ck/tensor_description/tensor_descriptor.hpp"
 #include "ck/tensor_description/tensor_descriptor_helper.hpp"
-#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
 #include "ck/tensor_operation/gpu/device/device_batched_gemm_softmax_gemm_permute.hpp"
 #include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
 #include "ck/tensor_operation/gpu/device/matrix_padder.hpp"
+#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
 #include "ck/tensor_operation/gpu/grid/gridwise_batched_gemm_softmax_gemm_xdl_cshuffle_v1.hpp"
 #include "ck/host_utility/device_prop.hpp"
 #include "ck/host_utility/kernel_launch.hpp"

include/ck/tensor_operation/gpu/device/device_conv3d_fwd_xdl_ndhwc_kzyxc_ndhwk.hpp → include/ck/tensor_operation/gpu/device/impl/device_conv3d_fwd_xdl_ndhwc_kzyxc_ndhwk.hpp

@@ -22,7 +22,7 @@ namespace tensor_operation {
 namespace device {
 
 /*
- * \see \link device_batched_gemm_xdl.hpp kernel_batched_gemm_xdlops_v2r3() \endlink.
+ * \see \link impl/device_batched_gemm_xdl.hpp kernel_batched_gemm_xdlops_v2r3() \endlink.
  */
 template <typename GridwiseGemm,
           typename FloatAB,