Practical: Working environment
------------------------------
1. Your working directory: cd ~/CGV/{nr} with {nr} = number of your group (01..19)
2. Choose your task: {task} (out of 01 02 04)
3. Fetch your skeleton: cp ~/CGV/skel/cgv_{task}.F90 .
4. Add your code, compile, run and test it (correct result?, same as serial result?)
5. If your task works:
extract your part (from /*== task_ii begin ==*/ to /*== task_ii end ==*/ )
into cgvp{task}.F90
6. When all groups have finished, everyone can check the total result with:
ls -l ../*/cgvp*.F90
cat ../00/cgvp00.F90 ../*/cgvp08.F90 ../*/cgvp01.F90 ../*/cgvp02.F90 ../*/cgvp03.F90
../*/cgvp04.F90 ../*/cgvp05.F90 ../*/cgvp06.F90 ../*/cgvp07.F90 > cg_all.F90
Caution: - duplicate parts must be selected by hand ({nr} instead of *)
- missing parts may be fetched also from ../source/parts/cgvp{task}.F90
7. Compile and run cgv_all.F90
- on T3E:
f90 -o cgv_all cgv_all.F90 -lm (compile)
fpart (to look whether there are free CPUs)
mpirun -np 2 ./cgv_all (run parallel)
- on many other platforms
f90 -o cgv_all cgv_all.F90 -lm -lmpi (compile)
or mpif90 -o cgv_all cgv_all.F90 -lm (compile)
mpirun -np 2 ./cgv_all (run parallel)
mpirun -np 4 ./cgv_all -n 100 -m 100 (run parallel with larger dataset)
- on non-MPI platforms:
f90 -Dserial -o cgv_all cgv_all.F90 -lm (compile)
./cgv_all (run serial)
Practical: Options
------------------
Compile-time options [default]:
-Dserial compile without MPI and without distribution [parallel]
-DWITH_MEMOPS use memcpy and memset functions instead of loops
for memory copy and set operations [loops]
-DFASTMEMXS run through the physical area during matrix-vector-mult
calculating two rows in one loop for better memory bandwidth usage
Run-time options [default]:
-m {m} vertical dimension of physical heat area [4]
-n {n} horizontal dimension [4]
-imax {iter_max} maximum number of iterations in the CG solver [500]
-eps {epsilon} abort criterion of the solver for residual vector [1e-6]
-twodims choose 2-dimensional domain decomposition [1-dim]
-mprocs {m_procs} choose number of processors, vertical, (-twodims needed)
-nprocs {n_procs} and horizontal [given by MPI_Dims_create]
-prtlev 0|1|2|3|4|5 printing and debug level [1]:
1 = only || result - exact solution || and partial result matrix
2 = and residual norm after each iteration
3 = and result of physical heat matrix
4 = and all vector and matrix information in 1st iteration
5 = and in all iterations
Example: mpirun -np 4 ./cgv_all -m 200 -n 200 -twodims