a fix

example/02_gemm_bilinear/gemm_bilinear_wmma_fp16.cpp

@@ -87,8 +87,8 @@ using DeviceOpInstance =
                                                                     8,
                                                                     16,
                                                                     16,
-                                                                    8,
-                                                                    1,
+                                                                    4,
+                                                                    2,
                                                                     S<4, 64, 1>,
                                                                     S<1, 0, 2>,
                                                                     S<1, 0, 2>,
@@ -105,7 +105,7 @@ using DeviceOpInstance =
                                                                     true,
                                                                     1,
                                                                     1,
-                                                                    S<1, 16, 1, 16>,
+                                                                    S<1, 32, 1, 8>,
                                                                     8>;
 
 int main(int argc, char* argv[])

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

@@ -151,7 +151,7 @@ __global__ void
 #if(!defined(__HIP_DEVICE_COMPILE__) || defined(__gfx1100__) || defined(__gfx1101__) || \
     defined(__gfx1102__))
     // printf("entry kernel launch");
-    __shared__ char p_shared[GridwiseOp::GetSharedMemoryNumberOfByte()];
+    __shared__ char p_shared[GridwiseOp::SharedMemTrait::lds_size];
 
     const index_t num_blocks_per_batch =
         __builtin_amdgcn_readfirstlane(get_grid_size() / batch_count);
@@ -241,7 +241,7 @@ __global__ void
 {
 #if(!defined(__HIP_DEVICE_COMPILE__) || defined(__gfx1100__) || defined(__gfx1101__) || \
     defined(__gfx1102__))
-    __shared__ char p_shared[GridwiseOp::GetSharedMemoryNumberOfByte()];
+    __shared__ char p_shared[GridwiseOp::SharedMemTrait::lds_size];
 
     GridwiseOp::template Run<HasMainKBlockLoop>(p_a_grid,
                                                 p_b_grid,