[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Help-gsl] another test release for gsl 1.16
From: |
Rhys Ulerich |
Subject: |
Re: [Help-gsl] another test release for gsl 1.16 |
Date: |
Fri, 19 Jul 2013 06:34:44 -0500 |
> I have uploaded another test release for gsl 1.16...
On an x86_64 Ubuntu system with just 'configure', I see 'make check'
pass for all of gcc 4.5, gcc 4.6, and gcc 4.7.
Intel 11.1 succeeds on 'make all' but fails in sys/ during 'make check' with
make[2]: Entering directory `/net/setun/workspace/rhys/tmp/intel-11.1/sys'
icc -DHAVE_CONFIG_H -I. -I../../gsl-1.15.91/sys -I..
-I../../gsl-1.15.91 -g -O2 -MT test.o -MD -MP -MF .deps/test.Tpo -c
-o test.o ../../gsl-1.15.91/sys/test.c
(0): internal error: backend signals
Intel 12.1 and 13.1 both succeed on 'make all' but dies in linalg/
with 'grep FAIL linalg/test-suite.log' showing:
FAIL: QR_solve hilbert(4) [32]
FAIL: QR Solve [38]
FAIL: QR_QRsolve hilbert(4) [122]
FAIL: QR QR Solve [128]
FAIL: QR_lssolve hilbert(4) [132]
FAIL: QR LS Solve [138]
FAIL: SV_decomp bigsparse [212]
FAIL: Singular Value Decomposition [66886]
FAIL: SV_solve hilbert(4) [233789]
FAIL: SVD Solve [233795]
These are likely just https://savannah.gnu.org/bugs/?39152 and/or
https://savannah.gnu.org/bugs/?39171 that we've discussed in the past.
I cataloged them here for completeness/archive searchability.
For lack of time, I have not tried backing the Intel builds down to
-O1 to see if they pass there. On older releases, this has tended to
work.
Thanks for shepherding the release Patrick.
- Rhys