3d grouped conv fwd with input/output fp16 and comp fp8 (#931)

* add f8 comp instance

* fixed

* fixed comments

* rename

* fixed dtype

* format

* fixed CI

* fixed ci

* add missing ComputeType

* fixed cit

* fixed

* Update cmake-ck-dev.sh

---------

Co-authored-by: Jing Zhang <jizha@amd.com>

include/ck/tensor_operation/gpu/device/device_grouped_conv_fwd_multiple_d.hpp

@@ -29,7 +29,8 @@ template <index_t NDimSpatial,
           typename EDataType,
           typename AElementwiseOperation,
           typename BElementwiseOperation,
-          typename CDEElementwiseOperation>
+          typename CDEElementwiseOperation,
+          typename ComputeType = ADataType>
 struct DeviceGroupedConvFwdMultipleD : public BaseOperator
 {
     static constexpr index_t NumDTensor = DsDataType::Size();

include/ck/tensor_operation/gpu/device/impl/device_grouped_conv_fwd_multiple_d_xdl_cshuffle.hpp

@@ -211,7 +211,8 @@ template <index_t NDimSpatial,
           index_t CShuffleNXdlPerWavePerShuffle,
           typename CDEBlockTransferClusterLengths_MBlock_MPerBlock_NBlock_NPerBlock,
           index_t CDEBlockTransferScalarPerVector_NPerBlock,
-          LoopScheduler LoopSched = make_default_loop_scheduler()>
+          typename ComputeDataType = ADataType,
+          LoopScheduler LoopSched  = make_default_loop_scheduler()>
 struct DeviceGroupedConvFwdMultipleD_Xdl_CShuffle
     : public DeviceGroupedConvFwdMultipleD<NDimSpatial,
                                            ALayout,
@@ -224,7 +225,8 @@ struct DeviceGroupedConvFwdMultipleD_Xdl_CShuffle
                                            EDataType,
                                            AElementwiseOperation,
                                            BElementwiseOperation,
-                                           CDEElementwiseOperation>
+                                           CDEElementwiseOperation,
+                                           ComputeDataType>
 {
     using DeviceOp = DeviceGroupedConvFwdMultipleD_Xdl_CShuffle;
 
@@ -323,8 +325,6 @@ struct DeviceGroupedConvFwdMultipleD_Xdl_CShuffle
     using DsGridDesc_M_N = remove_cvref_t<decltype(MakeDsGridDescriptor_M_N({}, {}))>;
     using EGridDesc_M_N  = remove_cvref_t<decltype(MakeEGridDescriptor_M_N<ELayout>({}, {}))>;
 
-    using ComputeDataType = ADataType;
-
     // GridwiseGemm
     using GridwiseGemm = GridwiseGemmMultipleD_xdl_cshuffle<
         ADataType, // TODO: distinguish A/B datatype