diff --git a/ci/gpu/build.sh b/ci/gpu/build.sh
index 8ea2c43658..9d8feebb14 100755
--- a/ci/gpu/build.sh
+++ b/ci/gpu/build.sh
@@ -53,7 +53,7 @@ gpuci_conda_retry install -c conda-forge -c rapidsai -c rapidsai-nightly -c nvid
       "libcumlprims=${MINOR_VERSION}" \
       "dask-cudf=${MINOR_VERSION}" \
       "dask-cuda=${MINOR_VERSION}" \
-      "ucx-py=0.20.*" \
+      "ucx-py=0.21" \
       "ucx-proc=*=gpu" \
       "xgboost=1.4.2dev.rapidsai${MINOR_VERSION}" \
       "rapids-build-env=${MINOR_VERSION}.*" \
@@ -195,8 +195,8 @@ else
 
     gpuci_logger "Install the main version of dask and distributed"
     set -x
-    pip install "git+https://github.com/dask/distributed.git@2021.05.1" --upgrade --no-deps
-    pip install "git+https://github.com/dask/dask.git@2021.05.1" --upgrade --no-deps
+    pip install "git+https://github.com/dask/distributed.git@main" --upgrade --no-deps
+    pip install "git+https://github.com/dask/dask.git@main" --upgrade --no-deps
     set +x
 
     gpuci_logger "Building cuml"
diff --git a/cpp/CMakeLists.txt b/cpp/CMakeLists.txt
index a69f9d17a3..6bd281ef08 100644
--- a/cpp/CMakeLists.txt
+++ b/cpp/CMakeLists.txt
@@ -58,12 +58,12 @@ option(BUILD_CUML_BENCH "Build cuML C++ benchmark tests" ON)
 option(BUILD_CUML_PRIMS_BENCH "Build ml-prims C++ benchmark tests" ON)
 option(BUILD_CUML_STD_COMMS "Build the standard NCCL+UCX Communicator" ON)
 option(BUILD_CUML_MPI_COMMS "Build the MPI+NCCL Communicator (used for testing)" OFF)
+option(CUDA_ENABLE_KERNEL_INFO "Enable kernel resource usage info" OFF)
+option(CUDA_ENABLE_LINE_INFO "Enable lineinfo in nvcc" OFF)
 option(DETECT_CONDA_ENV "Enable detection of conda environment for dependencies" ON)
 option(DISABLE_DEPRECATION_WARNINGS "Disable depreaction warnings " ON)
 option(DISABLE_OPENMP "Disable OpenMP" OFF)
 option(ENABLE_CUMLPRIMS_MG "Enable algorithms that use libcumlprims_mg" ON)
-option(KERNEL_INFO "Enable kernel resource usage info" OFF)
-option(LINE_INFO "Enable lineinfo in nvcc" OFF)
 option(NVTX "Enable nvtx markers" OFF)
 option(SINGLEGPU "Disable all mnmg components and comms libraries" OFF)
 option(USE_CCACHE "Cache build artifacts with ccache" OFF)
@@ -82,7 +82,8 @@ message(VERBOSE "CUML: Enabling detection of conda environment for dependencies:
 message(VERBOSE "CUML: Disabling OpenMP: ${DISABLE_OPENMP}")
 message(VERBOSE "CUML: Enabling algorithms that use libcumlprims_mg: ${ENABLE_CUMLPRIMS_MG}")
 message(VERBOSE "CUML: Enabling kernel resource usage info: ${KERNEL_INFO}")
-message(VERBOSE "CUML: Enabling lineinfo in nvcc: ${LINE_INFO}")
+message(VERBOSE "CUML: Enabling kernelinfo in nvcc: ${CUDA_ENABLE_KERNEL_INFO}")
+message(VERBOSE "CUML: Enabling lineinfo in nvcc: ${CUDA_ENABLE_LINE_INFO}")
 message(VERBOSE "CUML: Enabling nvtx markers: ${NVTX}")
 message(VERBOSE "CUML: Disabling all mnmg components and comms libraries: ${SINGLEGPU}")
 message(VERBOSE "CUML: Cache build artifacts with ccache: ${USE_CCACHE}")
diff --git a/cpp/cmake/modules/ConfigureCUDA.cmake b/cpp/cmake/modules/ConfigureCUDA.cmake
index b9f6795a67..b1c3ae83de 100644
--- a/cpp/cmake/modules/ConfigureCUDA.cmake
+++ b/cpp/cmake/modules/ConfigureCUDA.cmake
@@ -34,9 +34,13 @@ if(DISABLE_DEPRECATION_WARNING)
 endif()
 
 # Option to enable line info in CUDA device compilation to allow introspection when profiling / memchecking
-if(CUDA_ENABLE_LINEINFO)
-    list(APPEND CUML_CUDA_FLAGS -lineinfo)
-endif()
+if(CUDA_ENABLE_LINE_INFO)
+  list(APPEND CUML_CUDA_FLAGS -lineinfo)
+endif(LINE_INFO)
+
+if(CUDA_ENABLE_KERNEL_INFO)
+  list(APPEND CUML_CUDA_FLAGS -Xptxas=-v)
+endif(KERNEL_INFO)
 
 # Debug options
 if(CMAKE_BUILD_TYPE MATCHES Debug)