Thursday, January 7, 2016

How fast is Gordon? 2.6GHz 2x8 cores, compare with Byte Intel core i7 6 cores


@Byte, simulate aging for 1000 cells using 4 cores.
11:02am -11:05
time Rscript netwk_aging_sim.v0.1b.R -if1 merged_PPIGIN_Factorized2015Oct132Cols.csv -if2 essentialGeneLookupTable_20151013.csv -l1 0.0023 -l2 0 -dt 4 -p 0.9 -n 1000  -op __TEST -od tmp -iC 4 -d 1 -h /Users/hqin/github/mactower-network-failure-simulation/1.ori.ginppi.2015Oct

real 2m23.106s
user 8m25.295s
sys 0m37.280s

@gordon,  job with 16 cores
[hongqin@gordon-ln2 1.ori.ginppi.2015Oct]$ cat ori_aging_loopgordon.3.pbs
#!/bin/bash
#PBS -l nodes=1:ppn=16
#PBS -l walltime=07:00:00:00

module load R

echo hostname
cd /oasis/scratch/hongqin/temp_project/mactower-network-failure-simulation/1.ori.ginppi.2015Oct
pwd

#for p in 1.0 0.99 0.98 0.97 0.96 0.95 0.94 0.93 0.92 0.91 0.9 0.89 0.88 0.87 0.86 0.85 0.82 0.8 
for p in 1.0 0.99 0.98 0.97 0.96 0.95 0.94 0.93 0.92 0.91 0.9
do
   #for LOne in 0.002 0.0021 0.0022 0.0023 0.0024 0.0025 0.0026 0.0028 0.0028 0.0029 0.003 0.0035 0.004 0.0045
   for LOne in 0.001 0.0011 0.0012 0.0013 0.0014 0.0015 00016 0.0017 0.0018 0.0019 0.002
   do
      echo "$p $LOne"
 time Rscript netwk_aging_sim.v0.1b.R -if1 merged_PPIGIN_Factorized2015Oct132Cols.csv -if2 essentialGeneLookupTable_20151013.csv -l1 $LOne -l2 0 -dt 5 -p $p -n 5000 -op ginppi -od ginppi.ori.out -iC 16 -d 1 -h /oasis/scratch/hongqin/temp_project/mactower-network-failure-simulation/1.ori.ginppi.2015Oct
   done

done

real    3m35.381s
user    54m18.786s

sys     0m50.109s


[hongqin@gordon-ln1 1.ori.ginppi.2015Oct]$ cat ori_aging_loopgordon.3.pbs
#!/bin/bash
#PBS -l nodes=1:ppn=16
#PBS -l walltime=07:00:00:00

module load R

echo hostname
cd /oasis/scratch/hongqin/temp_project/mactower-network-failure-simulation/1.ori.ginppi.2015Oct
pwd

#for p in 1.0 0.99 0.98 0.97 0.96 0.95 0.94 0.93 0.92 0.91 0.9 0.89 0.88 0.87 0.86 0.85 0.82 0.8 
for p in 0.99 0.98 
do
   #for LOne in 0.002 0.0021 0.0022 0.0023 0.0024 0.0025 0.0026 0.0028 0.0028 0.0029 0.003 0.0035 0.004 0.0045
   for LOne in 0.001 0.0011
   do
      echo "$p $LOne"
 time Rscript netwk_aging_sim.v0.1b.R -if1 merged_PPIGIN_Factorized2015Oct132Cols.csv -if2 essentialGeneLookupTable_20151013.csv -l1 $LOne -l2 0 -dt 5 -p $p -n 1000 -op ginppi -od ginppi.ori.out -iC 16 -d 1 -h /oasis/scratch/hongqin/temp_project/mactower-network-failure-simulation/1.ori.ginppi.2015Oct
   done

done


real    0m45.861s
user    11m8.614s
sys     0m10.168s


Conclusions: Gordon single core seems to be similar or even slower than Byte. However, its numbers of cores saves time.

"Each compute node contains two 8-core 2.6 GHz Intel EM64T Xeon E5 (Sandy Bridge) processors and 64 GB of DDR3-1333 memory. " By is Intel 2.6 GHz Corei7 (6 cores?)


No comments:

Post a Comment