diff --git a/include/ck_tile/ops/gemm_quant/block/block_universal_gemm_ar_aquant_flatbr_bquant_cr.hpp b/include/ck_tile/ops/gemm_quant/block/block_universal_gemm_ar_aquant_flatbr_bquant_cr.hpp
index 1d9512b7f7..cd5a98de35 100644
--- a/include/ck_tile/ops/gemm_quant/block/block_universal_gemm_ar_aquant_flatbr_bquant_cr.hpp
+++ b/include/ck_tile/ops/gemm_quant/block/block_universal_gemm_ar_aquant_flatbr_bquant_cr.hpp
@@ -253,7 +253,7 @@ struct BlockGemmWeightPreshuffleABQuantARegBRegCReg
                         constexpr auto AmIter = (mIter + m_preload) % MIterPerWarp;
                         constexpr auto AkIter = (kIter + (mIter + m_preload) / MIterPerWarp);
 
-                        load_int4_tile<ADataType, ComputeDataType, UnaryOpSize>(
+                        load_and_convert_tile<UnaryOpSize>(
                             a_warp_tensor(number<AwarpIter>{}),
                             a_warp_windows(number<AmIter>{})(number<AkIter>{}));
                     }
diff --git a/include/ck_tile/ops/gemm_quant/pipeline/gemm_wp_abquant_pipeline_ag_bg_cr_v2.hpp b/include/ck_tile/ops/gemm_quant/pipeline/gemm_wp_abquant_pipeline_ag_bg_cr_v2.hpp
index 03b2cf4cf2..1a77b6a0b4 100644
--- a/include/ck_tile/ops/gemm_quant/pipeline/gemm_wp_abquant_pipeline_ag_bg_cr_v2.hpp
+++ b/include/ck_tile/ops/gemm_quant/pipeline/gemm_wp_abquant_pipeline_ag_bg_cr_v2.hpp
@@ -400,8 +400,6 @@ struct WPABQuantBPipelineAgBgCrV2 : public WeightPreshufflePipelineAGmemBGmemCRe
         block_sync_lds();
 
         // preload A00,A10 from lds
-        using ATypeToUse =
-            mixed_prec_compute_type_from_input_t<ADataType, BDataType, ComputeDataType>;
         using ATileType =
             decltype(make_static_distributed_tensor<BTypeToUse>(a_warp_tile_distribution));
         statically_indexed_array<ATileType, m_preload> a_warp_tensor;
@@ -409,7 +407,7 @@ struct WPABQuantBPipelineAgBgCrV2 : public WeightPreshufflePipelineAGmemBGmemCRe
         static_for<0, m_preload, 1>{}([&](auto loadIter) {
             constexpr auto mIter = loadIter % MIterPerWarp;
             constexpr auto kIter = loadIter / MIterPerWarp;
-            load_int4_tile<ADataType, ATypeToUse, UnaryOpSize_>(
+            load_and_convert_tile<UnaryOpSize_>(
                 a_warp_tensor(loadIter), a_warp_windows_ping(number<mIter>{})(number<kIter>{}));
         });
         __builtin_amdgcn_sched_barrier(0);
@@ -459,7 +457,7 @@ struct WPABQuantBPipelineAgBgCrV2 : public WeightPreshufflePipelineAGmemBGmemCRe
             static_for<0, m_preload, 1>{}([&](auto loadIter) {
                 constexpr auto mIter = loadIter % MIterPerWarp;
                 constexpr auto kIter = loadIter / MIterPerWarp;
-                load_int4_tile<ADataType, ATypeToUse, UnaryOpSize_>(
+                load_and_convert_tile<UnaryOpSize_>(
                     a_warp_tensor(loadIter), a_warp_windows_pong(number<mIter>{})(number<kIter>{}));
             });
 
@@ -504,7 +502,7 @@ struct WPABQuantBPipelineAgBgCrV2 : public WeightPreshufflePipelineAGmemBGmemCRe
             static_for<0, m_preload, 1>{}([&](auto loadIter) {
                 constexpr auto mIter = loadIter % MIterPerWarp;
                 constexpr auto kIter = loadIter / MIterPerWarp;
-                load_int4_tile<ADataType, ATypeToUse, UnaryOpSize_>(
+                load_and_convert_tile<UnaryOpSize_>(
                     a_warp_tensor(loadIter), a_warp_windows_ping(number<mIter>{})(number<kIter>{}));
             });
             iCounter--;
@@ -544,7 +542,7 @@ struct WPABQuantBPipelineAgBgCrV2 : public WeightPreshufflePipelineAGmemBGmemCRe
             static_for<0, m_preload, 1>{}([&](auto loadIter) {
                 constexpr auto mIter = loadIter % MIterPerWarp;
                 constexpr auto kIter = loadIter / MIterPerWarp;
-                load_int4_tile<ADataType, ATypeToUse, UnaryOpSize_>(
+                load_and_convert_tile<UnaryOpSize_>(
                     a_warp_tensor(loadIter), a_warp_windows_pong(number<mIter>{})(number<kIter>{}));
             });