Installation on different environments

Instructions for installation in OLCF Titan cluster

The instructions below refer to release v0.4

#!bin/bash

# Requirements for GPU compilation:
# - CUDA
# - OpenCL kernels
# - OpenCL support for CUDA

module load cudatoolkit/7.5.18-1.0502.10743.2.1
#
# Create subdirectory build/include/
#
# Edit file "fast_protein_cluster.cpp" and in lines 67 and 68 replace
# string "/usr/local/cuda/include" with
# string "/ccs/proj/bip141/sw/titan/fast_protein_cluster/v0.4/build/include"  # Notice "v0.4"
#
# Edit file "Makefile" and:
# - Lines 51 and 52 should be:
#    OPENCLINCLUDE=/opt/nvidia/cudatoolkit7.5/7.5.18-1.0502.10743.2.1/include
#    OPENCLINCLUDELOCAL=/ccs/proj/bip141/sw/titan/fast_protein_cluster/v0.4/build/include  # Notice "v0.4"
# - Line 56 should be:
#    OPENCLLIB=/opt/nvidia/cudatoolkit7.5/7.5.18-1.0502.10743.2.1/lib64
# - Lines 74 and 75 should be:
#    /bin/cp tmscore.cl $(OPENCLINCLUDELOCAL)/tmscore.cl
#    /bin/cp rmsd.cl $(OPENCLINCLUDELOCAL)/rmsd.cl
#
make clean
make SSE3=1 GPU=1 NVIDIA=1

A modules file can help submissions:

#%Module 1.0
#
#  fast_protein_cluster v0.4 module
#

module load cudatoolkit/7.5.18-1.0502.10743.2.1

set	installhome	/ccs/proj/bip141/sw/titan/fast_protein_cluster/v0.4
setenv		PROTCLUSTERHOME $installhome
prepend-path    PATH            $installhome

We test with the following script:

#!/bin/bash

#PBS -q debug
#PBS -A BIP141
#PBS -N protCluster
#PBS -j eo
#PBS -l walltime=0:09:00
#PBS -l nodes=1
#PBS -l gres=atlas1%atlas2

# Load module for fast_protein_cluster
source $MODULESHOME/init/bash
module load fast_protein_cluster/v0.4

# Prepare the scratch directory
swd=$MEMBERWORK/bip141/fast_protein_cluster/test
mkdir -p $swd
/bin/rm -rf $swd/*  # in case it was previously used
cd $PROTCLUSTERHOME/test/
/bin/cp -rf traj.* $swd/

# Run the test
cd $swd/
setenv OMP_NUM_THREADS 8
aprun -n 1 -d 8 -j 1 fast_protein_cluster -i traj -o output_binpos --binary_coords --rmsd --nclusters 10 --nthreads 8 --hcomplete --gpu

#####################################################################
####### Bath Queue Walltimes as a Function of Number of Nodes #######
#####################################################################
#Note: 1 node = 16 cores
#Bin Min Nodes Max Nodes Max Walltime (Hours) Aging Boost (Days)
# 1   11,250        –             24.0             15
# 2    3,750     11,249           24.0              5
# 3      313      3,749           12.0              0
# 4      125        312            6.0              0
# 5        1        124            2.0              0
##################################