Changes between Version 11 and Version 12 of ParallelismPerformances
- Timestamp:
- 2012-10-16T08:38:14+02:00 (12 years ago)
Legend:
- Unmodified
- Added
- Removed
- Modified
-
ParallelismPerformances
v11 v12 136 136 === ''' ANSWERS FROM YANN (08/10/2012) ''': === 137 137 138 The problem is a hardware problem : the lustre disksystem used by TGCC is configured for "big" files. When some data are read into a netcdf file, 2Mo of memory138 The problem is a hardware problem : the lustre system used by TGCC is configured for "big" files. When some data are read into a netcdf file, 2Mo of memory 139 139 is used by the system. Moreover, the array vec_tmp contains data which are not contiguous in the netcdf file. 140 140 Solutions proposed : … … 143 143 * forcing file : use LIMITED instead of LIMITED 144 144 145 === ''' PATCH :''' === 146 147 * Josefine suggested to modify the call to fliopen : 148 {{{ 149 150 * comment line 84 : call flinclo 151 152 * Modify the calling to flinopen line 105 (and allocation of itau) : 153 154 ttm_part=2 ! Here only read the 2 first values of time axis 155 ALLOCATE(itau(ttm_part)) 156 157 CALL flinopen & 158 & (filename, .FALSE., iim_full, jjm_full, llm_full, lon_full, lat_full, & 159 & lev_full, ttm_part, itau, date0, dt_force, force_id) 160 }}} 161 162 After first tests, the results are OK, we have the same performance when we set 2 in flincom module. 145 163 146