FMHA, add support to gfx1101/gfx1102

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

@@ -588,7 +588,7 @@ struct DeviceBatchedGemmSoftmaxGemmPermute_Wmma_CShuffle
 
     static bool IsSupportedArgument(const Argument& arg)
     {
-        if(ck::get_device_name() == "gfx1100")
+        if(ck::get_device_name() == "gfx1100" || ck::get_device_name() == "gfx1101" || ck::get_device_name() == "gfx1102")
         {
             if constexpr(!(is_same_v<Acc0DataType, float> || is_same_v<Acc0DataType, int32_t>))
             {

include/ck/tensor_operation/gpu/grid/gridwise_batched_gemm_softmax_gemm_wmma_cshuffle.hpp

@@ -60,7 +60,7 @@ __global__ void
             const C0MatrixMask c0_matrix_mask,
             const Block2CTileMap block_2_ctile_map)
 {
-#if(!defined(__HIP_DEVICE_COMPILE__) || defined(__gfx1100__))
+#if(!defined(__HIP_DEVICE_COMPILE__) || defined(__gfx1100__) || defined(__gfx1101__) || defined(__gfx1102__))
     __shared__ char p_shared[GridwiseOp::GetSharedMemoryNumberOfByte()];
 
     const index_t num_blocks_per_batch =