User Tools

Site Tools


namd

Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Both sides previous revision Previous revision
Next revision
Previous revision
namd [2017/07/05 21:11]
root
namd [2017/08/07 20:57]
root
Line 3: Line 3:
 The namd-verbs-smp binary [[http://​www.ks.uiuc.edu/​Development/​Download/​download.cgi?​PackageName=NAMD]] version 2.11 or 2.12 is installed in /​share/​apps/​NAMD on [[razor]] and [[trestles]]. ​ It does not use MPI. The namd-verbs-smp binary [[http://​www.ks.uiuc.edu/​Development/​Download/​download.cgi?​PackageName=NAMD]] version 2.11 or 2.12 is installed in /​share/​apps/​NAMD on [[razor]] and [[trestles]]. ​ It does not use MPI.
  
-This is for multiple-node runs. +This is for multiple-node runs with ''​charmrun''​ as the distributed component and ''​namd2''​ on each compute node. We have found most runs are faster with the ''​+setcpuaffinity +isomalloc_sync''​ options. charmrun ''​++ppn''​ should match PBS ''​ppn=''​.
 <​code>​ <​code>​
 module load namd/​2.12 ​ [or 2.11] module load namd/​2.12 ​ [or 2.11]
Line 13: Line 13:
   echo "host ${node}"​ >> nodelist   echo "host ${node}"​ >> nodelist
 done done
-charmrun ++remote-shell ssh ++ppn 16 `which namd2` +p $NP apoa1.namd >​apoa1.logfile+charmrun ++remote-shell ssh ++ppn 16 `which namd2` +p $NP +setcpuaffinity +isomalloc_sync ​apoa1.namd >​apoa1.logfile
 </​code>​ </​code>​
  
Line 21: Line 21:
 cd $PBS_O_WORKDIR cd $PBS_O_WORKDIR
 NP=$(wc -l <​$PBS_O_WORKDIR) NP=$(wc -l <​$PBS_O_WORKDIR)
-namd2 +p $NP apoa1.namd >​apoa1.logfile+namd2 +p $NP apoa1.namd ​+setcpuaffinity +isomalloc_sync ​>​apoa1.logfile
 </​code>​ </​code>​
  
 == Benchmarks == == Benchmarks ==
-The NAMD website has benchmarks run on Trestles while at UCSD [[http://​www.ks.uiuc.edu/​Research/​namd/​performance.html]],​ but they don't have any info on how the scores were obtained (namd2, charmrun, or mpi).  These are shown as benchmark time*cores. Best results for charmrun here were obtained with multiple nodes using ppn=cores/​node,​p=total cores (or ppn*nodes). Single nodes running namd2 were both p=cores and are comparable with the published benchmarks. Version 2.12 is substantially faster than 2.11.  The downloaded verbs-smp version is set by the module as it is faster than the ibverbs-smp version.+The NAMD website has benchmarks run on Trestles while at UCSD [[http://​www.ks.uiuc.edu/​Research/​namd/​performance.html]],​ but they don't have any info on how the scores were obtained (namd2, charmrun, or mpi).  These are shown as benchmark time*cores. Best results for charmrun here were obtained with multiple nodes using ppn=cores/​node,​p=total cores (or ppn*nodes). Single nodes running namd2 were both p=cores and are comparable with the published benchmarks. Version 2.12 is substantially faster than 2.11.  The downloaded verbs-smp version is set by the module as it is faster than the ibverbs-smp version. On this problem, the Intel version didn't show any useful scaling for more than 2 nodes, and AMD not very useful scaling for more than 3 nodes.
  
 <​code>​ <​code>​
-Node Type      ppn  version ​ p  Nodes Bench WallClock  ​Trestles+Node Type      ppn  version ​ p  Nodes Bench WallClock  ​UCSD Bench
 16-core Intel  16   ​2.11 ​   16    1   ​1.21 ​   383        n/a 16-core Intel  16   ​2.11 ​   16    1   ​1.21 ​   383        n/a
 16-core Intel  16   ​2.12 ​   16    1   ​0.76 ​   256        n/a 16-core Intel  16   ​2.12 ​   16    1   ​0.76 ​   256        n/a
-16-core Intel  16   ​2.12 ​   16    2   0.92    151        n/a +16-core Intel  16   ​2.12 ​   16    2   0.90    146        n/a 
-32-core AMD    32   ​2.12 ​   32    1   2.05    340        1.9 +16-core Intel  16   ​2.12 ​   16    3   ​1.32 ​   146        n/a 
-32-core AMD    32   ​2.12 ​   32    2   2.34    189        2.0+32-core AMD    32   ​2.12 ​   32    1   1.95    317        1.9 
 +32-core AMD    32   ​2.12 ​   32    2   2.22    185        2.0 
 +32-core AMD    32   ​2.12 ​   32    3   ​2.29 ​   127        n/a 
 +32-core AMD    32   ​2.12 ​   32    4   ​2.56 ​   104        n/a
 </​code>​ </​code>​
  
namd.txt · Last modified: 2017/08/07 20:57 by root