diff --git a/templates/namd/2.14/data.namd.arm.gpu.opt.config b/templates/namd/2.14/data.namd.arm.gpu.opt.config
index b1f5ad4942dc597c59f369c3bc819f7e78522b64..bf9a1c14496eeb396de55d5ab60e2fb9e9b27acb 100644
--- a/templates/namd/2.14/data.namd.arm.gpu.opt.config
+++ b/templates/namd/2.14/data.namd.arm.gpu.opt.config
@@ -3,79 +3,118 @@
 
 [DOWNLOAD]
 #download NAMD_2.14 and charm-6.10.2 by hand
-https://www.ks.uiuc.edu/Development/Download/download.cgi?PackageName=NAMD
-http://charm.cs.illinois.edu/distrib/charm-6.10.2.tar.gz
+#https://www.ks.uiuc.edu/Development/Download/download.cgi?PackageName=NAMD
+NAMD/2.14 https://www.ks.uiuc.edu/Research/namd/2.14/download/946183/NAMD_2.14_Source.tar.gz
+charm/6.10.2 http://charm.cs.illinois.edu/distrib/charm-6.10.2.tar.gz
+cuda/11.4.0 https://developer.download.nvidia.com/compute/cuda/11.4.0/local_installers/cuda_11.4.0_470.42.01_linux_sbsa.run
+tcl/8.5.9 http://www.ks.uiuc.edu/Research/namd/libraries/tcl8.5.9-linux-arm64-threaded.tar.gz
 
 [DEPENDENCY]
-yum install -y fftw-devel tcl-devel
+yum install -y fftw-devel tcl-devel 
 module use ./software/modulefiles
 module purge
 ./jarvis -install bisheng/2.1.0 com
 module load bisheng/2.1.0
 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
+module use ./software/moduledeps/bisheng2.1.0
+module load hmpi/1.1.1
+./jarvis -install fftw/3.3.8 clang+mpi
+module use ./software/moduledeps/bisheng2.1.0-hmpi1.1.1
+module load fftw/3.3.8
+#tar xzf $JARVIS_DOWNLOAD/NAMD_2.14_Source.tar.gz
+#cd NAMD_2.14_Source
+rm -rf NAMD-2.13
+mkdir NAMD-2.14
+#tar xf charm-6.10.2.tar
+# 1. 添加执行权限
+sudo chmod +x ${JARVIS_DOWNLOAD}/cuda_11.4.0_470.42.01_linux_sbsa.run
+# 2. 静默安装
+sudo sh ${JARVIS_DOWNLOAD}/cuda_11.4.0_470.42.01_linux_sbsa.run \
+  --silent \
+  --toolkit \
+  --override \
+  --no-opengl-libs \
+  --no-man-page \
+  --no-drm \
+  --installpath=/usr/local/cuda \
+  --toolkitpath=/usr/local/cuda \
+  --samplespath=/usr/local/cuda/samples
 
 [ENV]
 module use ./software/modulefiles
 module purge
 module load bisheng/2.1.0
+module use ./software/moduledeps/bisheng2.1.0
 module load hmpi/1.1.1
+module use ./software/moduledeps/bisheng2.1.0-hmpi1.1.1
+module load fftw/3.3.8
 export CC=clang CXX=clang++ FC=flang
-export PATH=//usr/local/cuda/bin:$PATH
+export PATH=/usr/local/cuda/bin:$PATH
 
 [APP]
 app_name = NAMD
-build_dir = $JARVIS_ROOT/NAMD_2.14_Source/
-#binary_dir = $JARVIS_ROOT/NAMD_2.14_Source/Linux-ARM64-g++
-binary_dir = 
+build_dir = $JARVIS_ROOT/NAMD-2.14/
+binary_dir = $JARVIS_ROOT/NAMD-2.14/build/NAMD_2.14_Source/Linux-ARM64-g++
+binary_dir =
 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) $(TCLFLAGS)
+rm -rf build
+rm -rf install
+mkdir build install
+export BUILD_DIR=$JARVIS_ROOT/NAMD-2.14/build
+export INSTALL_DIR=$JARVIS_ROOT/NAMD-2.14/install
+cd $JARVIS_ROOT/NAMD-2.14/build
+tar -xvf $JARVIS_DOWNLOAD/charm-6.10.2.tar.gz
+cd charm-v6.10.2
+mkdir ${INSTALL_DIR}/charm690/
+cp src/arch/mpi-linux-x86_64 src/arch/mpi-linux-arm8 -rf
+grep -rl 'm64' src/arch/mpi-linux-arm8 | xargs sed -i 's/\-m64//g'
+./build charm++ mpi-linux-arm8 smp --with-production --destination=$JARVIS_ROOT/NAMD-2.14/install/charm690/mpi-linux-arm8
+
+cd $BUILD_DIR
+tar xzf $JARVIS_DOWNLOAD/NAMD_2.14_Source.tar.gz
+cd NAMD_2.14_Source
+cp arch/Linux-x86_64.fftw3 arch/Linux-ARM64.fftw3
+cd $JARVIS_ROOT/NAMD-2.14/
+tar -xzvf $JARVIS_DOWNLOAD/tcl8.5.9-linux-arm64-threaded.tar.gz
+cd $BUILD_DIR/NAMD_2.14_Source
+sed -i "s|^CHARMBASE[[:space:]]*=.*|CHARMBASE = $INSTALL_DIR/charm690|" Make.charm
+
+cd $BUILD_DIR/NAMD_2.14_Source
+sed -i '1,$d' arch/Linux-ARM64-g++.arch
+cat <<EOF >> "arch/Linux-ARM64-g++.arch"
+NAMD_ARCH = Linux-ARM64
+CHARMARCH = mpi-linux-arm8
+#FLOATOPTS = -O2 -ffast-math -funsafe-math-optimizations -fomit-frame-pointer -mcpu=cortex-a9 -mtune=cortex-a9
+FLOATOPTS = -O2 -ffast-math -funsafe-math-optimizations -fomit-frame-pointer
+CXX = clang++
+CXXOPTS = \$(FLOATOPTS)
+CXXNOALIASOPTS = \$(FLOATOPTS) -fno-strict-aliasing
+CC = clang
+COPTS = \$(FLOATOPTS)
 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
+./config Linux-ARM64-g++ --with-fftw3 --fftw-prefix $FFTW_PATH --with-tcl --tcl-prefix $JARVIS_ROOT/NAMD-2.14/tcl8.5.9-linux-arm64-threaded --charm-arch mpi-linux-arm8
 cd Linux-ARM64-g++
-make -j
+make
+
 
 [CLEAN]
 make clean
 
 [RUN]
 run = rm -rf namd_* && dsub -s dsub_namd_1n.sh
-binary = 
+binary =
 # binary = namd2 +p126 +setcpuaffinity +maffinity +isomalloc_sync +devices 0,1 stmv_nve_cuda.namd 2>&1 | tee namd.log && python2 $JARVIS_ROOT/templates/namd/2.14/ns_per_day.py namd.log
 nodes = 1
 
 [PERF]
 kperf = --spe --uncore
-perf = 
+perf =
 nsys = -y 5s -d 60s
 ncu = --target-processes all --launch-skip 16 --launch-count 1
 
diff --git a/templates/namd/2.14/namd.arm.gpu.config b/templates/namd/2.14/namd.arm.gpu.config
index 3b49fd8491364b3f8844f43dff9e9d1bce19a57d..d8e6e6784bce1468107488004a20f0c86f133488 100644
--- a/templates/namd/2.14/namd.arm.gpu.config
+++ b/templates/namd/2.14/namd.arm.gpu.config
@@ -2,55 +2,103 @@
 1.1.11.1
 
 [DOWNLOAD]
-https://www.ks.uiuc.edu/Development/Download/download.cgi?PackageName=NAMD
+NAMD/2.14 https://www.ks.uiuc.edu/Research/namd/2.14/download/946183/NAMD_2.14_Source.tar.gz
+charm/6.10.2 http://charm.cs.illinois.edu/distrib/charm-6.10.2.tar.gz
+cuda/11.4.0 https://developer.download.nvidia.com/compute/cuda/11.4.0/local_installers/cuda_11.4.0_470.42.01_linux_sbsa.run
+tcl/8.5.9 http://www.ks.uiuc.edu/Research/namd/libraries/tcl8.5.9-linux-arm64-threaded.tar.gz
 
 [DEPENDENCY]
+yum install -y fftw-devel tcl-devel 
 module use ./software/modulefiles
 module purge
 ./jarvis -install bisheng/2.1.0 com
 module load bisheng/2.1.0
 export CC=clang CXX=clang++ FC=flang
-./jarvis -install hmpi/1.2.0 clang
-tar -xzvf $JARVIS_DOWNLOAD/NAMD_2.14_Source.tar.gz
-cd NAMD_2.14_Source
-tar xvf charm-6.10.2.tar
+./jarvis -install hmpi/1.1.1 clang
+module use ./software/moduledeps/bisheng2.1.0
+module load hmpi/1.1.1
+./jarvis -install fftw/3.3.8 clang+mpi
+module use ./software/moduledeps/bisheng2.1.0-hmpi1.1.1
+module load fftw/3.3.8
+#tar xzf $JARVIS_DOWNLOAD/NAMD_2.14_Source.tar.gz
+#cd NAMD_2.14_Source
+rm -rf NAMD-2.13
+mkdir NAMD-2.14
+#tar xf charm-6.10.2.tar
+# 1. 添加执行权限
+sudo chmod +x ${JARVIS_DOWNLOAD}/cuda_11.4.0_470.42.01_linux_sbsa.run
+# 2. 静默安装
+sudo sh ${JARVIS_DOWNLOAD}/cuda_11.4.0_470.42.01_linux_sbsa.run \
+  --silent \
+  --toolkit \
+  --override \
+  --no-opengl-libs \
+  --no-man-page \
+  --no-drm \
+  --installpath=/usr/local/cuda \
+  --toolkitpath=/usr/local/cuda \
+  --samplespath=/usr/local/cuda/samples
 
 [ENV]
 module use ./software/modulefiles
 module purge
 module load bisheng/2.1.0
-module load hmpi/1.2.0
-# add cuda
-export CUDA_INCLUDE_DIRS=/usr/local/cuda/include
-export CUDA_CUDART_LIBRARY=/usr/local/cuda/lib64/libcudart.so
-export CUDA_HOME=/usr/local/cuda
-export PATH=$CUDA_HOME/bin:$PATH
-export LD_LIBRARY_PATH=$CUDA_HOME/lib64:$LD_LIBRARY_PATH
-export C_INCLUDE_PATH=$CUDA_HOME/include:$C_INCLUDE_PATH
-export LIBRARY_PATH=$CUDA_HOME/lib64:$LIBRARY_PATH
+module use ./software/moduledeps/bisheng2.1.0
+module load hmpi/1.1.1
+module use ./software/moduledeps/bisheng2.1.0-hmpi1.1.1
+module load fftw/3.3.8
+export CC=clang CXX=clang++ FC=flang
+export PATH=/usr/local/cuda/bin:$PATH
 
 [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}/stmv
+build_dir = $JARVIS_ROOT/NAMD-2.14/
+binary_dir = $JARVIS_ROOT/NAMD-2.14/build/NAMD_2.14_Source/Linux-ARM64-g++
+binary_dir =
+case_dir = ${JARVIS_ROOT}/workloads/namd/stmv
 
 [BUILD]
 set -x
 set -e
-cd charm-6.10.2
-#charm_arch='netlrts-linux-arm8-flang-smp-clang'
-#./build charm++ netlrts-linux-arm8 smp -j32 clang flang --with-production
-charm_arch='multicore-arm8-flang-clang'
-./build charm++ multicore-arm8 -j32 clang flang --with-production
-cd $charm_arch/tests/charm++/megatest
-make pgm
-./pgm +p4
-cd ../../../../..
-rm -rf Linux-ARM64-g++
-./config Linux-ARM64-g++ --charm-arch $charm_arch --with-cuda --cxx clang++ --cc clang
+
+rm -rf build
+rm -rf install
+mkdir build install
+export BUILD_DIR=$JARVIS_ROOT/NAMD-2.14/build
+export INSTALL_DIR=$JARVIS_ROOT/NAMD-2.14/install
+cd $JARVIS_ROOT/NAMD-2.14/build
+tar -xvf $JARVIS_DOWNLOAD/charm-6.10.2.tar.gz
+cd charm-v6.10.2
+mkdir ${INSTALL_DIR}/charm690/
+cp src/arch/mpi-linux-x86_64 src/arch/mpi-linux-arm8 -rf
+grep -rl 'm64' src/arch/mpi-linux-arm8 | xargs sed -i 's/\-m64//g'
+./build charm++ mpi-linux-arm8 smp --with-production --destination=$JARVIS_ROOT/NAMD-2.14/install/charm690/mpi-linux-arm8
+
+cd $BUILD_DIR
+tar xzf $JARVIS_DOWNLOAD/NAMD_2.14_Source.tar.gz
+cd NAMD_2.14_Source
+cp arch/Linux-x86_64.fftw3 arch/Linux-ARM64.fftw3
+cd $JARVIS_ROOT/NAMD-2.14/
+tar -xzvf $JARVIS_DOWNLOAD/tcl8.5.9-linux-arm64-threaded.tar.gz
+cd $BUILD_DIR/NAMD_2.14_Source
+sed -i "s|^CHARMBASE[[:space:]]*=.*|CHARMBASE = $INSTALL_DIR/charm690|" Make.charm
+
+cd $BUILD_DIR/NAMD_2.14_Source
+sed -i '1,$d' arch/Linux-ARM64-g++.arch
+cat <<EOF >> "arch/Linux-ARM64-g++.arch"
+NAMD_ARCH = Linux-ARM64
+CHARMARCH = mpi-linux-arm8
+#FLOATOPTS = -O2 -ffast-math -funsafe-math-optimizations -fomit-frame-pointer -mcpu=cortex-a9 -mtune=cortex-a9
+FLOATOPTS = -O2 -ffast-math -funsafe-math-optimizations -fomit-frame-pointer
+CXX = clang++
+CXXOPTS = \$(FLOATOPTS)
+CXXNOALIASOPTS = \$(FLOATOPTS) -fno-strict-aliasing
+CC = clang
+COPTS = \$(FLOATOPTS)
+EOF
+./config Linux-ARM64-g++ --with-fftw3 --fftw-prefix $FFTW_PATH --with-tcl --tcl-prefix $JARVIS_ROOT/NAMD-2.14/tcl8.5.9-linux-arm64-threaded --charm-arch mpi-linux-arm8
 cd Linux-ARM64-g++
-make -j
+make
 
 [CLEAN]
 make clean
@@ -64,4 +112,4 @@ nodes = 1
 kperf = --spe --uncore
 perf =
 nsys = -y 5s -d 60s
-ncu = --target-processes all --launch-skip 16 --launch-count 1
\ No newline at end of file
+ncu = --target-processes all --launch-skip 16 --launch-count 1