From b387ffd9f75951bce4acdafeb2d6252423b5bd48 Mon Sep 17 00:00:00 2001
From: Eugen Betke <betke@dkrz.de>
Date: Fri, 19 Oct 2018 15:33:06 +0200
Subject: [PATCH] Experiment for reduction of benchmark runtime

---
 micro_run.py | 21 +++++++++++++++++++++
 1 file changed, 21 insertions(+)
 create mode 100755 micro_run.py

diff --git a/micro_run.py b/micro_run.py
new file mode 100755
index 000000000..abf1ca9d3
--- /dev/null
+++ b/micro_run.py
@@ -0,0 +1,21 @@
+#!/bin/bash 
+
+export MODULEPATH=/gsfs/betke/software/modules:$MODULEPATH
+
+module purge
+module load betke/hdf5/1.8.20-ddn
+module load betke/ior/git-ddn
+module list
+
+T=$((1 * 1024 * 1024))
+B=$((128 * 1024 * 1024 * 1024))
+
+IOR="$(which ior)"
+LUSTRE_TESTFILE="/esfs/jtacquaviva/file2"
+OUTPUTDIR="./output"
+GENV="-genv IM_CLIENT_NET_BUFFERS 32 -genv MV2_NUM_HCAS 1 -genv MV2_CPU_BINDING_LEVEL core -genv MV2_CPU_BINDING_POLICY scatter -genv MV2_SHOW_CPU_BINDING 1"
+MPIEXEC="/opt/ddn/mvapich/bin/mpiexec -ppn 1 -np 1 -hosts isc17-c17 $GENV"
+
+$MPIEXEC $IOR -i 3  -s 1 -t $T  -b $B  -o $LUSTRE_TESTFILE -e -g -z -k -w
+$MPIEXEC $IOR -i 3  -s 1 -t $T  -b $B  -o $LUSTRE_TESTFILE -e -g -z -k -D 1 -r
+rm $LUSTRE_TESTFILE