ICODE-ADC/lammps
4
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

git-svn-id: svn://svn.icms.temple.edu/lammps-ro/trunk@12132 f3b2605a-c512-4ea7-a41b-209d697bcdaa

Browse Source
This commit is contained in:
sjplimp
2014-06-14 19:34:19 +00:00
parent 7b5846a386
commit 2ec1442e97
5 changed files with 8 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
lib/gpu/lal_device.cpp
View File

@ -492,6 +492,7 @@ void DeviceT::output_times(UCL_Timer &time_pair, Answer<numtyp,acctyp> &ans,
const double driver_overhead,
const int threads_per_atom, FILE *screen) {
double single[9], times[9];
int post_final=0;
single[0]=atom.transfer_time()+ans.transfer_time();
single[1]=nbor.time_nbor.total_seconds()+nbor.time_hybrid1.total_seconds()+
@ -504,6 +505,9 @@ void DeviceT::output_times(UCL_Timer &time_pair, Answer<numtyp,acctyp> &ans,
single[7]=ans.cpu_idle_time();
single[8]=nbor.bin_time();
MPI_Finalized(&post_final);
if (post_final) return;
MPI_Reduce(single,times,9,MPI_DOUBLE,MPI_SUM,0,_comm_replica);
double my_max_bytes=max_bytes+atom.max_gpu_bytes();