Hello,Just wondering if this solver can be used for a tridiagonal matrix with periodic boundary conditions as it has elements in the far ends of the matrix in addition to the tridiagonal matrix.Thanks.
I'm trying Intel adaptive spike-based solver.
When I run the sparse.f90 example in parallel with my own matrix data, it gave an error -309.
n= 5844 nnz= 350064
UNSUCCESSFUL RUN FOR SPIKE - INFO EXIT -1
SPIKE_CORE ERROR CODE -309
If I run the program with only one processor (-np=1) it goes well.
n= 5844 nnz= 350064
I'm new to this software.When I tried to compile my program with "-mcmodel=medium" option, Intel compiler gives an error below.
src/spike_algo.f90:(.text+0x1c): relocation truncated to fit: R_X86_64_32 against `.bss'
Because my code includes large static array,"-mcmodel=medium" isnecessary.Adaptive Spike Solver is not compiled with"-mcmodel=medium" ?Can I getappropriate build from web site?regards,
I am currently experimenting some troubles using Adaptive-Spike with sparse preconditioners while it works perfectly with dense preconditioners.
I am using the library to solve general sparse linear systems with preconditioners almost always sparse within the band.
When I set my sparse preconditioner in the dense format (pre.format='D'), it works but leads to poor performances (due to comunication and computation overheads). When I set my preconditioner in the CSR format (pre.format='S'), the library seems to use an identity matrix as the preconditioner.
The libraries link ok with gcc, but I can't link it with g++ icpc when I rename files to *.cpp:
The sources are:
hlp|.c|.cpp|.h same as help|.c|.c|.h, and
ldb|.c|.cpp almost the same as exemple4.c; ldb.h the has functions prototipes
Have design in mind to use spikes . Buildson disk to model protein.
I have installed SPIKE on a system with Intel Xeon with 4 processors. I have openmpi also installed.
When trying to compile and run the test examples, it was complaining 'mpiifort was not found.'
I changed the makefile (in the SPIKE/1.0/examples/examples_f90 directory and replaced mpiifort by mpif90 and then the compilation ran ok.
However the test results show
cs8/_results> more toeplitz.txt
UNSUCCESSFUL RUN FOR SPIKE - INFO EXIT -3
MPI ERROR CODE 5
cs8/_results> more *.txt
I have just downloaded Spike 1.0 and encounted a link error when making the C examples on an AMD opteron and Linux platform. The spike library is automatically included in the compile command line:
cc cc -o _results/toeplitz toeplitz.o help.o -L/home/SPIKE/1.0/examples/examples_c/../../lib/em64t -lspike -lspike_adapt -lspike_adapt_de -lspike_adapt_grid_f -lspike_mpi_comm_intelmpi -lmkl_solver -lmkl_lapack -lmkl -lguide -lpthread -lm
The linker complains:
/usr/bin/ld: cannot find -lspike_adapt_de
It looks like the spike solver only supports real double precision. Are there plans for a complex double precision version?
I have a question about Intel Spike solver. I want to solve matrix ecology2 (http://www.cise.ufl.edu/research/sparse/matrices/McRae/ecology2.html) from florida matrix collection using SPIKE. What kind of solving strategy (RSS, DFS, OIS) should I choose? The matrix is solved on a 48-nodes cluster with 8 cores on each node.