diff --git a/cmake/cuda.cmake b/cmake/cuda.cmake
index 6f4671c13a9e3dccb9be0a06f4bc2453af94bd55..7a94bda0f5f73e48081f68d7b2730e3df1e46232 100644
--- a/cmake/cuda.cmake
+++ b/cmake/cuda.cmake
@@ -16,7 +16,7 @@ else()
   set(paddle_known_gpu_archs8 "30 35 50 52 60 61")
   set(paddle_known_gpu_archs9 "30 35 50 52 60 61 70")
   set(paddle_known_gpu_archs10 "30 35 50 52 60 61 70 75")
-  set(paddle_known_gpu_archs11 "35 50 52 60 61 70 75 80")
+  set(paddle_known_gpu_archs11 "52 60 61 70 75 80")
 endif()
 
 ######################################################################################
diff --git a/cmake/external/warpctc.cmake b/cmake/external/warpctc.cmake
index 0f6b1c182d5590354c8a970eea339a3e23846f39..ac6cf624e82c0a346fea42fa29fe9bab6ace8d47 100644
--- a/cmake/external/warpctc.cmake
+++ b/cmake/external/warpctc.cmake
@@ -18,7 +18,7 @@ SET(WARPCTC_PREFIX_DIR  ${THIRD_PARTY_PATH}/warpctc)
 SET(WARPCTC_SOURCE_DIR  ${THIRD_PARTY_PATH}/warpctc/src/extern_warpctc)
 SET(WARPCTC_INSTALL_DIR ${THIRD_PARTY_PATH}/install/warpctc)
 set(WARPCTC_REPOSITORY  https://github.com/baidu-research/warp-ctc.git)
-set(WARPCTC_TAG         bc29dcfff07ced1c7a19a4ecee48e5ad583cef8e)
+set(WARPCTC_TAG         fc7f226b93758216a03b1be9d24593a12819b984)
 
 SET(WARPCTC_INCLUDE_DIR "${WARPCTC_INSTALL_DIR}/include"
     CACHE PATH "Warp-ctc Directory" FORCE)
diff --git a/cmake/flags.cmake b/cmake/flags.cmake
index 9d07a0979d9392c9b2ab78562f8e0ceb8fc5d722..415e07c75425345f5f1ad29a8544e02a5bfb12e4 100644
--- a/cmake/flags.cmake
+++ b/cmake/flags.cmake
@@ -28,7 +28,15 @@ function(CheckCompilerCXX11Flag)
 endfunction()
 
 CheckCompilerCXX11Flag()
-set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -std=c++11")
+if (WITH_GPU)
+    if (${CMAKE_CUDA_COMPILER_VERSION} GREATER_EQUAL 11.0)
+       set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -std=c++14")
+    else()
+      set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -std=c++11")
+    endif()
+else()
+    set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -std=c++11")
+endif()
 # safe_set_flag
 #
 # Set a compile flag only if compiler is support
diff --git a/cmake/third_party.cmake b/cmake/third_party.cmake
index c9442e8f843ac152cac02908799a8d24f5951e58..9edfcb967abc26a25a94d368298c1c475295019f 100644
--- a/cmake/third_party.cmake
+++ b/cmake/third_party.cmake
@@ -243,9 +243,10 @@ IF(WITH_TESTING OR (WITH_DISTRIBUTE AND NOT WITH_GRPC))
 ENDIF()
 
 if(WITH_GPU)
-    include(external/cub)       # download cub
-    list(APPEND third_party_deps extern_cub)
-  
+    if (${CMAKE_CUDA_COMPILER_VERSION} LESS 11.0)
+        include(external/cub)       # download cub
+        list(APPEND third_party_deps extern_cub)
+    endif()
     set(CUDAERROR_URL  "http://paddlepaddledeps.bj.bcebos.com/cudaErrorMessage.tar.gz" CACHE STRING "" FORCE)
     file_download_and_uncompress(${CUDAERROR_URL} "cudaerror") # download file cudaErrorMessage
 endif(WITH_GPU)
diff --git a/paddle/fluid/operators/CMakeLists.txt b/paddle/fluid/operators/CMakeLists.txt
index 6e8ff52ed4a8846f5f6060e10cfd9bec22308e9e..f0a04d850dff01e0776e96bbe518cde2ce8bb88b 100644
--- a/paddle/fluid/operators/CMakeLists.txt
+++ b/paddle/fluid/operators/CMakeLists.txt
@@ -45,7 +45,9 @@ endif()
 SET(OP_HEADER_DEPS xxhash executor)
 
 if (WITH_GPU)
-    SET(OP_HEADER_DEPS ${OP_HEADER_DEPS} cub)
+    if (${CMAKE_CUDA_COMPILER_VERSION} LESS 11.0)
+        SET(OP_HEADER_DEPS ${OP_HEADER_DEPS} cub)
+    endif()
 endif()
 
 SET(OP_PREFETCH_DEPS "")
diff --git a/paddle/fluid/operators/detection/CMakeLists.txt b/paddle/fluid/operators/detection/CMakeLists.txt
index 537063640e4ef6e49f7b991482f0f3122ecef02f..c2b7c27ab4adb5282ad7aa5f7a16c15f81ba5f5e 100644
--- a/paddle/fluid/operators/detection/CMakeLists.txt
+++ b/paddle/fluid/operators/detection/CMakeLists.txt
@@ -41,9 +41,13 @@ detection_library(sigmoid_focal_loss_op SRCS sigmoid_focal_loss_op.cc sigmoid_fo
 detection_library(retinanet_detection_output_op SRCS retinanet_detection_output_op.cc)
 
 if(WITH_GPU)
-  detection_library(generate_proposals_op SRCS generate_proposals_op.cc generate_proposals_op.cu DEPS memory cub)
-  detection_library(distribute_fpn_proposals_op SRCS distribute_fpn_proposals_op.cc distribute_fpn_proposals_op.cu DEPS memory cub)
-  detection_library(collect_fpn_proposals_op SRCS collect_fpn_proposals_op.cc collect_fpn_proposals_op.cu DEPS memory cub)
+  set(TMPDEPS memory)
+  if (${CMAKE_CUDA_COMPILER_VERSION} LESS 11.0)
+      set(TMPDEPS memory cub)
+  endif()
+  detection_library(generate_proposals_op SRCS generate_proposals_op.cc generate_proposals_op.cu DEPS ${TMPDEPS})
+  detection_library(distribute_fpn_proposals_op SRCS distribute_fpn_proposals_op.cc distribute_fpn_proposals_op.cu DEPS ${TMPDEPS})
+  detection_library(collect_fpn_proposals_op SRCS collect_fpn_proposals_op.cc collect_fpn_proposals_op.cu DEPS ${TMPDEPS})
 else()
   detection_library(generate_proposals_op SRCS generate_proposals_op.cc)
   detection_library(distribute_fpn_proposals_op SRCS distribute_fpn_proposals_op.cc)
diff --git a/paddle/fluid/operators/math/CMakeLists.txt b/paddle/fluid/operators/math/CMakeLists.txt
index 3a19c7edff3569d503480fd060a6432dc59d2108..10d335b828b516fe08871f314ba4667c06f04714 100644
--- a/paddle/fluid/operators/math/CMakeLists.txt
+++ b/paddle/fluid/operators/math/CMakeLists.txt
@@ -9,7 +9,11 @@ function(math_library TARGET)
     set(hip_srcs)
     set(math_common_deps device_context framework_proto enforce)
     if (WITH_GPU)
-        list(APPEND math_common_deps cub)
+        if (${CMAKE_CUDA_COMPILER_VERSION} LESS 11.0)	
+            list(APPEND math_common_deps cub)
+	else()
+            list(APPEND math_common_deps)
+	endif()
     endif()
     set(multiValueArgs DEPS)
     cmake_parse_arguments(math_library "${options}" "${oneValueArgs}"
diff --git a/paddle/fluid/operators/reduce_ops/CMakeLists.txt b/paddle/fluid/operators/reduce_ops/CMakeLists.txt
index 3da481a142aa2282aade661de7679cf4edf597a0..a68666b100cb52c722c4fefc849e94947130010f 100644
--- a/paddle/fluid/operators/reduce_ops/CMakeLists.txt
+++ b/paddle/fluid/operators/reduce_ops/CMakeLists.txt
@@ -1,6 +1,10 @@
 include(operators)
 if(WITH_GPU)
-    register_operators(DEPS cub)
+    if(${CMAKE_CUDA_COMPILER_VERSION} LESS 11.0)
+        register_operators(DEPS cub)
+    else()
+        register_operators()
+    endif()
 else()
     register_operators()
 endif()
@@ -24,5 +28,9 @@ if(WITH_GPU)
 endif()
 
 if(WITH_GPU)
-    nv_test(check_reduce_rank_test SRCS check_reduce_rank_test.cu DEPS tensor cub)
+    if (${CMAKE_CUDA_COMPILER_VERSION} LESS 11.0)
+	nv_test(check_reduce_rank_test SRCS check_reduce_rank_test.cu DEPS tensor cub)
+    else()
+	nv_test(check_reduce_rank_test SRCS check_reduce_rank_test.cu DEPS tensor)
+    endif()
 endif()