Directory branches/budaev/HEDG2_06_gprof2/

Lines of Code

Developers

Most Recent Commits

sbu062 2019-01-31 20:51 Rev.: 7741

deleted work branch: fixed intel compiler bug workaround for FUCK_IFORT_WA1 block

0 lines of code changed in 3 files:

• branches/budaev/HEDG2_06_gprof2: Makefile (del), m_common.f90 (del), m_evolut.f90 (del)

sbu062 2019-01-31 20:11 Rev.: 7736

fix notification on Windows

0 lines of code changed in 2 files:

• branches/budaev/HEDG2_06_gprof2: Makefile (changed)

sbu062 2019-01-31 16:18 Rev.: 7735

more info describing intel compiler bug with segmentation fault as workaround code in r7734

9 lines of code changed in 1 file:

• branches/budaev/HEDG2_06_gprof2: m_evolut.f90 (+9 -1)

sbu062 2019-01-31 16:09 Rev.: 7734

added workarounds against intel fortran compiler bugs that apear in optimised binary builds with O3, components of derived types and getters from different modules become unaccessible and result in segmentation crashes

22 lines of code changed in 1 file:

• branches/budaev/HEDG2_06_gprof2: m_evolut.f90 (+22 -9)

sbu062 2019-01-31 15:55 Rev.: 7733

print sanitiser bounds to log at start

2 lines of code changed in 1 file:

• branches/budaev/HEDG2_06_gprof2: m_common.f90 (+2 -1)

sbu062 2019-01-31 15:50 Rev.: 7732

doc: description for sanitiser bounds

5 lines of code changed in 1 file:

• branches/budaev/HEDG2_06_gprof2: m_common.f90 (+5)

sbu062 2019-01-31 15:10 Rev.: 7731

global validation, low and high,parameters in (sanitise) and (do_sanitise)

7 lines of code changed in 1 file:

• branches/budaev/HEDG2_06_gprof2: m_common.f90 (+7 -4)

sbu062 2019-01-31 10:38 Rev.: 7730

refactored sanitise functions: (do_sanitise) and (sanitise), the former calling IEEE math modules

102 lines of code changed in 2 files:

• branches/budaev/HEDG2_06_gprof2: m_common.f90 (+101 -15), m_evolut.f90 (+1 -1)

sbu062 2019-01-30 22:46 Rev.: 7728

fix ifort crashes: got rid of FILE_IO object as it breaks ifort; use array sanitizer in (generation_stats_record_write) to avoid crashes in csv file

When using a very long history, some perception object history stack values are broken/uninitialised/invalid which breaks output in CSV_IO call, hence use sanitiser added in r7727. Switched to raw CSV_IO calls instead of the FILE_IO wrapper because intel ifort breaks public file handle object rendering file unit and name unaccessible. Debugging did not succeed, perhaps it's a bug in Intel compiler. So, raw CSV_IO calls are used, but they are not verbose. Actually, FILE_IO may be oblivious...

0 lines of code changed in 2 files:

• branches/budaev/HEDG2_06_gprof2: m_evolut.f90 (changed)

sbu062 2019-01-30 21:56 Rev.: 7727

fix ifort crashes: initial implementation of data sanitiser

0 lines of code changed in 2 files:

• branches/budaev/HEDG2_06_gprof2: m_common.f90 (changed)