diff --git a/package/bisheng/2.3.0/install.sh b/package/bisheng/2.3.0/install.sh
index d5a5f8c92ba6b32ac7cf6ffe75104cac61590207..d8fa5803ba5c76aba9b26bc1fd0cc7c9923f7ff2 100644
--- a/package/bisheng/2.3.0/install.sh
+++ b/package/bisheng/2.3.0/install.sh
@@ -1,11 +1,3 @@
 #!/bin/bash
-set -e
-bisheng_version='2.3.0'
-arch='x86'
-if [ x$(arch) = xaarch64 ];then
-	arch='aarch64'
-fi
-. ${DOWNLOAD_TOOL} -u https://mirrors.huaweicloud.com/kunpeng/archive/compiler/bisheng_compiler/bisheng-compiler-${bisheng_version}-${arch}-linux.tar.gz
-cd ${JARVIS_TMP}
-. $CHECK_ROOT && yum -y install libatomic libstdc++ libstdc++-devel
-tar xzvf ${JARVIS_DOWNLOAD}/bisheng-compiler-${bisheng_version}-${arch}-linux.tar.gz -C $1 --strip-components=1
+bisheng_ver='bisheng-compiler-2.3.0'
+../meta.sh $1
diff --git a/package/bisheng/2.4.0/install.sh b/package/bisheng/2.4.0/install.sh
new file mode 100644
index 0000000000000000000000000000000000000000..7b1108079b3d7bf0dc070b891dff8b0868575316
--- /dev/null
+++ b/package/bisheng/2.4.0/install.sh
@@ -0,0 +1,4 @@
+#!/bin/bash
+set -e
+export bisheng_ver='BiSheng-compiler-2.4.0'
+../meta.sh $1
\ No newline at end of file
diff --git a/package/bisheng/meta.sh b/package/bisheng/meta.sh
new file mode 100644
index 0000000000000000000000000000000000000000..894b029f9aa44d420418f39029ff70df0f11485f
--- /dev/null
+++ b/package/bisheng/meta.sh
@@ -0,0 +1,10 @@
+#!/bin/bash
+set -e
+arch='x86'
+if [ x$(arch) = xaarch64 ];then
+	arch='aarch64'
+fi
+. ${DOWNLOAD_TOOL} -u https://mirrors.huaweicloud.com/kunpeng/archive/compiler/bisheng_compiler/${bisheng_ver}-${arch}-linux.tar.gz
+cd ${JARVIS_TMP}
+. $CHECK_ROOT && yum -y install libatomic libstdc++ libstdc++-devel
+tar xzvf ${JARVIS_DOWNLOAD}/${bisheng_ver}-${arch}-linux.tar.gz -C $1 --strip-components=1
diff --git a/package/cuda/FAQ b/package/cuda/FAQ
new file mode 100644
index 0000000000000000000000000000000000000000..69b2f537830087b6efcea15191529de1034cfeec
--- /dev/null
+++ b/package/cuda/FAQ
@@ -0,0 +1,11 @@
+Q：The file '/tmp/.X0-lock' exists and appears to contain the process ID '1062' of a >runnning X server.
+ERROR: You appear to be running an X server; please exit X before installing.  For >further details, please see the section INSTALLING THE NVIDIA DRIVER in the README >available on the Linux driver download page at www.nvidia.com.
+ERROR: Installation has failed.  Please see the file '/var/log/nvidia-installer.log' >for details.  You may find suggestions on fixing installation problems in the README >available on the Linux driver download page at www.nvidia.com.
+
+A：service lightdm stop 
+
+Q: 升级CUDA及驱动报错 ERROR: An NVIDIA kernel module 'nvidia-uvm' appears to already be loaded in your kernel. 
+A: 
+nvidia-uninstall
+lsof -n -w /dev/nvidia*
+kill -9 PID
\ No newline at end of file
diff --git a/package/kfft/1.0.0/install.sh b/package/kfft/1.0.0/install.sh
new file mode 100644
index 0000000000000000000000000000000000000000..5d44996fe4e689a62abd6c4c57797180bca6c051
--- /dev/null
+++ b/package/kfft/1.0.0/install.sh
@@ -0,0 +1,15 @@
+#!/bin/bash
+set -x
+set -e
+lib_name='kml_adapter'
+. ${DOWNLOAD_TOOL} -u https://gitee.com/openeuler/kml_adapter.git -t git
+cd $JARVIS_DOWNLOAD
+cd ./$lib_name/kml_fft_adapter
+export C_INCLUDE_PATH=/usr/local/kml/include/
+export CPLUS_INCLUDE_PATH=/usr/local/kml/include/
+chmod +x build.sh
+dos2unix build.sh
+rm -rf build
+./build.sh
+mkdir -p $1/lib
+cp -rf build/* $1/lib
\ No newline at end of file
diff --git a/templates/namd/2.14/data.namd.arm.gpu.opt.config b/templates/namd/2.14/data.namd.arm.gpu.opt.config
new file mode 100644
index 0000000000000000000000000000000000000000..07f9f5135b58eea7668e2fae13c4954e3a157dd2
--- /dev/null
+++ b/templates/namd/2.14/data.namd.arm.gpu.opt.config
@@ -0,0 +1,76 @@
+[SERVER]
+1.1.1.1
+
+[DOWNLOAD]
+https://www.ks.uiuc.edu/Development/Download/download.cgi?PackageName=NAMD
+
+[DEPENDENCY]
+yum install -y fftw-devel tcl-devel
+module use ./software/modulefiles
+module purge
+./jarvis -install bisheng/2.4.0 com
+module load bisheng/2.4.0
+export CC=clang CXX=clang++ FC=flang
+./jarvis -install hmpi/1.1.1 clang
+tar xzf $JARVIS_DOWNLOAD/NAMD_2.14_Source.tar.gz
+cd NAMD_2.14_Source
+tar xf $JARVIS_DOWNLOAD/charm-6.10.2.tar
+
+[ENV]
+module use ./software/modulefiles
+module purge
+module load bisheng/2.4.0
+module load hmpi/1.1.1
+export CC=clang CXX=clang++ FC=flang
+
+[APP]
+app_name = NAMD
+build_dir = $JARVIS_ROOT/NAMD_2.14_Source/
+binary_dir = $JARVIS_ROOT/NAMD_2.14_Source/Linux-ARM64-g++/
+case_dir = ${JARVIS_ROOT}/workloads/namd/stmv
+
+[BUILD]
+set -x
+set -e
+cat > ./arch/Linux-ARM64.fftw <<\EOF
+FFTDIR=/usr
+FFTINCL=-I$(FFTDIR)/include
+FFTLIB=-L$(FFTDIR)/lib -lfftw3f
+FFTFLAGS=-DNAMD_FFTW -DNAMD_FFTW_3
+FFT=$(FFTINCL) $(FFTFLAGS)
+EOF
+
+cat > ./arch/Linux-ARM64.tcl <<\EOF
+TCLDIR=/usr
+TCLINCL=-I$(TCLDIR)/include
+TCLLIB=-L$(TCLDIR)/lib64 -ltcl8.6 -ldl -lpthread
+TCLFLAGS=-DNAMD_TCL
+TCL=$(TCLINCL) $(TCLFLAGS)
+EOF
+#compile charm
+cd charm-6.10.2
+./build charm++ multicore-arm8 -j32 clang flang --with-production
+cd multicore-arm8-flang-clang/tests/charm++/megatest
+make pgm
+./pgm +p4
+cd ../../../../..
+#compile NAMD
+rm -rf Linux-ARM64-g++
+./config Linux-ARM64-g++ --charm-arch multicore-arm8-flang-clang --with-cuda --cxx clang++ --cc clang
+cd Linux-ARM64-g++
+make -j
+
+[CLEAN]
+make clean
+
+[RUN]
+run = 
+binary = namd2 +p126 +setcpuaffinity +maffinity +isomalloc_sync +devices 0,2 stmv_nve_cuda.namd 2>&1 | tee namd.log && python $JARVIS_ROOT/templates/namd/2.14/ns_per_day.py namd.log
+nodes = 1
+
+[PERF]
+kperf = --spe --uncore
+perf = 
+nsys = -y 5s -d 60s
+ncu = --target-processes all --launch-skip 16 --launch-count 1
+
diff --git a/templates/qe/6.8/data.qe.arm.gpu.opt.config b/templates/qe/6.8/data.qe.arm.gpu.opt.config
new file mode 100644
index 0000000000000000000000000000000000000000..2b46f14924646abe47a999f9cc25b64b893d1677
--- /dev/null
+++ b/templates/qe/6.8/data.qe.arm.gpu.opt.config
@@ -0,0 +1,79 @@
+[SERVER]
+1.1.1.1
+
+[DOWNLOAD]
+qe/6.8 https://github.com/QEF/q-e/archive/refs/tags/qe-6.8.tar.gz
+
+[DEPENDENCY]
+set -x
+set -e
+./jarvis -install kgcc/9.3.1 com
+module purge
+module use ./software/modulefiles
+module load kgcc/9.3.1
+export CC=`which gcc`
+export CXX=`which g++`
+export FC=`which gfortran`
+./jarvis -install kml/1.5.0/gcc gcc
+# use openmpi4 instead of default openmpi3
+cd /opt/nvidia/hpc_sdk/Linux_aarch64/21.9/comm_libs/
+rm -rf mpi
+ln -s openmpi4/openmpi-4.0.5 mpi
+
+[ENV]
+module purge
+./jarvis -install fftw/3.3.10 gcc
+export PATH=/usr/local/cuda-11.4/bin/:$PATH
+module use /opt/nvidia/hpc_sdk/modulefiles/
+module load nvhpc/21.9
+
+module use ./software/modulefiles
+module load kgcc/9.3.1
+source /etc/profile
+module load kfft/1.0.0
+# add BLAS\Lapack
+export BLAS_LIBS="-I/usr/local/kml/include -L/usr/local/kml/lib/kblas/omp -lkblas"
+export LAPACK_LIBS="-L/usr/local/kml/lib -lklapack_full"
+export FFT_LIBS="-I${FFTW_PATH}/include -L${KFFT_PATH}/lib -lfftw3 -lfftw3f"
+# add MPI
+export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/opt/nvidia/hpc_sdk/Linux_aarch64/21.9/comm_libs/openmpi4/openmpi-4.0.5/lib/
+export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/opt/nvidia/hpc_sdk/Linux_aarch64/21.9/compilers/lib/
+export LD_LIBRARY_PATH=${JARVIS_ROOT}/software/libs/gcc9/kml/1.4.0/kfft/:$LD_LIBRARY_PATH
+if ! nvidia-cuda-mps-control -d; then
+    echo "ERROR: Failed to start MPS daemon. Please resolve issue or set GPU_WORKERS to 1"
+    exit 1
+fi
+echo "INFO: MPS server daemon started"
+trap "echo quit | nvidia-cuda-mps-control" EXIT
+
+[APP]
+app_name = QE
+build_dir = ${JARVIS_ROOT}/q-e-qe-6.8-opt/
+binary_dir = ${JARVIS_ROOT}/q-e-qe-6.8-opt/bin/
+case_dir = ${JARVIS_ROOT}/workloads/QE/qe-large/
+
+[BUILD]
+# install devicexlib
+cd external
+rm -rf devxlib/*
+tar xzf devicexlib-master.tar.gz -C devxlib --strip-components=1
+cd devxlib
+./configure FC=pgf90 CC=pgcc --with-cuda=yes --with-cuda-cc=80 --with-cuda-runtime=11.4 --disable-parallel --enable-cuda-env-check=no
+make all
+cd ../../
+#add patch here
+# patch -fp0 < xx.patch
+./configure --with-cuda=yes --with-cuda-runtime=11.4 --with-cuda-cc=80  --enable-openmp --with-scalapack=no
+echo DFLAGS += -D__FFTW3 -D__GPU_MPI >> make.inc
+echo LD_LIBS += -lcurand -lcufft >> make.inc
+echo CUDA_F90FLAGS += -Mipa -Munroll -Mvect -fma -use_fast_math -O4 >> make.inc
+echo FFLAGS += -O3 >> make.inc
+make pw -j 64
+
+[CLEAN]
+make clean
+
+[RUN]
+run = mpirun --allow-run-as-root -np 8 -x CUDA_VISIBLE_DEVICES=0,2 -x OMP_NUM_THREADS=1
+binary = pw.x -nk 8 -input scf.in
+nodes = 1
\ No newline at end of file