- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hello,
About two years ago I reported ifort F03 compilation issues with PSBLAS library (http://people.uniroma2.it/salvatore.filippone/psblas/). With Intel 2015, the compilation issues are solved but there are runtime problems. I have also tried the Ifort 2016 Beta and same problem occurs. PSBLAS is usable with several other compilers (such as GNU/IBM/CRAY) other than Intel. We are thoroughly using Intel Compilers for research and using PSBLAS will save us big time for not reinventing the wheel.
Is there any kind promise/timeline the Intel developers can give to finally fix these bugs?
Link Copied
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Please provide details we can use to reproduce the problem, and describe the "bugs" you have observed.
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
I am using psblas-3.3.3/test/pargen/ppde2d to test.
To Build in Ubuntu 14.04, Intel fortran 16.0.0 Beta:
PSBLAS Build
./configure --enable-serial F90COPT=-g
make
cd psblas-3.3.3/test/pargen/
make
cd runs/
./ppde2d < ppde.inp
--------------------------------------------------------------
Welcome to PSBLAS version: 3.3.3
This is the pde2d90 sample program
Solving matrix : ell1
Grid dimensions : 100 x 100
Number of processors : 1
Data distribution : BLOCK
Preconditioner : BJAC
Iterative method : BICGSTAB
Generating Matrix (size=10000)...
--------------------------------------------------------------
This is abnormal termination. To trace back the problem: In the subroutine "psb_d_csall" in file "serial/impl/psb_d_mat_impl.F90", near line 600, in the "call a%set_bld()", the compiler is not able choose "subroutine psb_d_set_bld(a)" to run.
- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Printer Friendly Page