fix spelling

example/01_gemm/run_gemm_example.inc

@@ -37,7 +37,7 @@ bool run_gemm(const ProblemType& problem_size, const ExecutionConfig& config)
         [](std::size_t row, std::size_t col, std::size_t stride, auto layout) {
             if(stride == 0)
             {
-                // give a chance if stride is zero, return a defalt packed stride
+                // give a chance if stride is zero, return a default packed stride
                 if constexpr(std::is_same_v<decltype(layout), ck::tensor_layout::gemm::RowMajor>)
                 {
                     return col;

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

@@ -686,7 +686,7 @@ struct BlockToCTileMap_GemmStreamK
         // one cu can hold one wg at one time, from the whole chip's point of view
         // if number of wg is same as num_cu, we call it 1 dispatch
         // if number of wg is 2x num_cu, we call it 2 dispatches.
-        // one dispatch can deliever wg same as num_cu (full dispatch), or less than num_cu (partial
+        // one dispatch can deliver wg same as num_cu (full dispatch), or less than num_cu (partial
         // dispatch)
         //
         uint32_t full_dispatches         = num_tiles / num_cu;