Grid/gcc-bug-report at bc09d7c3bd4146e914eb5114d8cb404a9ca84260 - Grid

mirror of https://github.com/paboyle/Grid.git synced 2024-09-20 09:15:38 +01:00

History

Peter Boyle 28bdc90908 Sizable improvement in multigrid for unsquared. 6000 matmuls CG unprec 2000 matmuls CG prec (4000 eo muls) 1050 matmuls PGCR on 16^3 x 32 x 8 m=.01 Substantial effort on timing and logging infrastructure		2015-07-24 01:31:13 +09:00
..
broken.cc	Sizable improvement in multigrid for unsquared.	2015-07-24 01:31:13 +09:00
README	Log the bug report code into the git repo.	2015-05-15 12:39:53 +01:00

README

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66153

Grid code breaks on GCC4.8, 4.9, 5.0 due to the 
peekIndex<n> operating on lattice.

It erroneously recurses back into the Lattice<obj> variant, even though
the lattice container is dropped.

Work around is possible; if the Lattice routine is given a disambiguating 
name prefix, such as 

latPeekIndex<n> 

GCC5 works. 

However this is ugly and for now I have submitted a bug report to see the reaction and
speed of fixing. 

The simple testcase in this directory is the submitted bug report that encapsulates the
problem. The test case works with icpc and with clang++, but fails consistently on g++
current variants.

Peter