public inbox for gcc-bugs@sourceware.org
help / color / mirror / Atom feed
* [Bug target/51751] New: COMPLEX16 tests fail in Lapack
@ 2012-01-04 12:07 tkoenig at gcc dot gnu.org
2012-01-04 23:18 ` [Bug target/51751] " tkoenig at gcc dot gnu.org
` (4 more replies)
0 siblings, 5 replies; 6+ messages in thread
From: tkoenig at gcc dot gnu.org @ 2012-01-04 12:07 UTC (permalink / raw)
To: gcc-bugs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=51751
Bug #: 51751
Summary: COMPLEX16 tests fail in Lapack
Classification: Unclassified
Product: gcc
Version: 4.6.2
Status: UNCONFIRMED
Keywords: wrong-code
Severity: normal
Priority: P3
Component: target
AssignedTo: unassigned@gcc.gnu.org
ReportedBy: tkoenig@gcc.gnu.org
Target: powerpc64-unknown-linux-gnu
Running the Lapack 3.4.0 testsuite exposes the following errors:
ZGB: 281 out of 28488 tests failed to pass the threshold
*** On entry to ZGBSVX parameter number 13 had an illegal value ***
*** On entry to ZGBSVX parameter number 13 had an illegal value ***
*** On entry to ZGBSVX parameter number 13 had an illegal value ***
*** On entry to ZGBSVX parameter number 13 had an illegal value ***
...
*** On entry to ZGBSVX parameter number 14 had an illegal value ***
*** On entry to ZGBSVX parameter number 14 had an illegal value ***
*** On entry to ZGBSVX parameter number 14 had an illegal value ***
*** On entry to ZGBSVX parameter number 14 had an illegal value ***
and a few more.
The same occurs with current trunk.
This can be reproduced on gcc110 on the compile farm.
^ permalink raw reply [flat|nested] 6+ messages in thread
* [Bug target/51751] COMPLEX16 tests fail in Lapack
2012-01-04 12:07 [Bug target/51751] New: COMPLEX16 tests fail in Lapack tkoenig at gcc dot gnu.org
@ 2012-01-04 23:18 ` tkoenig at gcc dot gnu.org
2012-01-06 2:14 ` tkoenig at gcc dot gnu.org
` (3 subsequent siblings)
4 siblings, 0 replies; 6+ messages in thread
From: tkoenig at gcc dot gnu.org @ 2012-01-04 23:18 UTC (permalink / raw)
To: gcc-bugs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=51751
--- Comment #1 from Thomas Koenig <tkoenig at gcc dot gnu.org> 2012-01-04 23:18:15 UTC ---
Also fails with 4.5.3 on powerpc64-unknown-linux-gnu at -O3, and with current
trunk at -O0.
Does not fail on x86_64-unknown-linux-gnu with current trunk.
^ permalink raw reply [flat|nested] 6+ messages in thread
* [Bug target/51751] COMPLEX16 tests fail in Lapack
2012-01-04 12:07 [Bug target/51751] New: COMPLEX16 tests fail in Lapack tkoenig at gcc dot gnu.org
2012-01-04 23:18 ` [Bug target/51751] " tkoenig at gcc dot gnu.org
@ 2012-01-06 2:14 ` tkoenig at gcc dot gnu.org
2012-01-06 13:10 ` dominiq at lps dot ens.fr
` (2 subsequent siblings)
4 siblings, 0 replies; 6+ messages in thread
From: tkoenig at gcc dot gnu.org @ 2012-01-06 2:14 UTC (permalink / raw)
To: gcc-bugs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=51751
--- Comment #2 from Thomas Koenig <tkoenig at gcc dot gnu.org> 2012-01-06 02:14:13 UTC ---
The problem appears to be in the test routine zdrvgb, at line
639 (and following) of zdrvgb.f, where the contents of the array
s contain zero for powerpc64 and very low values for x86_64:
On the powerpc machine:
(gdb) up
#1 0x00000000100c88dc in zdrvgb (dotype=..., nn=7, nval=..., nrhs=1,
thresh=30, tsterr=.TRUE., a=..., la=43692,
afb=..., lafb=65472, asav=..., b=..., bsav=..., x=..., xact=..., s=...,
work=..., rwork=..., iwork=..., nout=6)
at zdrvgb.f:639
639 $ WORK, RWORK( 2*NRHS+1 ), INFO )
(gdb) p s
$1 = ()
(gdb) p s(1)
$2 = 0
(gdb) p s(2)
$3 = 0
On the x86_64 machine:
(gdb) p s(1)
$2 = 8.8030696956720451e-293
(gdb) p s(0)
$3 = 30
(gdb) p s(2)
$4 = 7.9525381577453941e-293
(gdb) p s(3)
$5 = 7.627547548164537e-293
^ permalink raw reply [flat|nested] 6+ messages in thread
* [Bug target/51751] COMPLEX16 tests fail in Lapack
2012-01-04 12:07 [Bug target/51751] New: COMPLEX16 tests fail in Lapack tkoenig at gcc dot gnu.org
2012-01-04 23:18 ` [Bug target/51751] " tkoenig at gcc dot gnu.org
2012-01-06 2:14 ` tkoenig at gcc dot gnu.org
@ 2012-01-06 13:10 ` dominiq at lps dot ens.fr
2012-01-08 10:13 ` tkoenig at gcc dot gnu.org
2012-01-08 12:29 ` dominiq at lps dot ens.fr
4 siblings, 0 replies; 6+ messages in thread
From: dominiq at lps dot ens.fr @ 2012-01-06 13:10 UTC (permalink / raw)
To: gcc-bugs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=51751
--- Comment #3 from Dominique d'Humieres <dominiq at lps dot ens.fr> 2012-01-06 13:10:24 UTC ---
It works for me with trunk r182883 on powerpc-apple-darwin9:
with -O0
--> LAPACK TESTING SUMMARY <--
Processing LAPACK Testing output found in the TESTING direcory
SUMMARY nb test run numerical error other error
================ =========== ================= ================
REAL 1070421 38 (0.004%) 0 (0.000%)
DOUBLE PRECISION 1052315 203 (0.019%) 0 (0.000%)
COMPLEX 508588 2 (0.000%) 0 (0.000%)
COMPLEX16 523860 28 (0.005%) 0 (0.000%)
--> ALL PRECISIONS 3155184 271 (0.009%) 0 (0.000%)
or with -O3
--> LAPACK TESTING SUMMARY <--
Processing LAPACK Testing output found in the TESTING direcory
SUMMARY nb test run numerical error other error
================ =========== =================
================
REAL 1033283 197 (0.019%) 0
(0.000%)
DOUBLE PRECISION 1047653 205 (0.020%) 0
(0.000%)
COMPLEX 503926 3 (0.001%) 0
(0.000%)
COMPLEX16 521520 27 (0.005%) 0
(0.000%)
--> ALL PRECISIONS 3106382 432 (0.014%) 0
(0.000%)
The main difference with -O3 is
--> Testing REAL Singular Value Decomposition [ ssvd.out ]
SBD drivers: 32 out of 8360 tests failed to pass the threshold
SBD drivers: 32 out of 8360 tests failed to pass the threshold
SBD drivers: 32 out of 8360 tests failed to pass the threshold
SBD drivers: 32 out of 8360 tests failed to pass the threshold
SBD drivers: 32 out of 8360 tests failed to pass the threshold
--> Tests passed: 27550
--> Tests failing to pass the threshold: 160
On x86_64-apple-darwin10 with trunk r182460, I got for -O2 or -O3
--> LAPACK TESTING SUMMARY <--
Processing LAPACK Testing output found in the TESTING direcory
SUMMARY nb test run numerical error other error
================ =========== ================= ================
REAL 1050655 40 (0.004%) 0 (0.000%)
DOUBLE PRECISION 1056977 202 (0.019%) 0 (0.000%)
COMPLEX 508588 2 (0.000%) 0 (0.000%)
COMPLEX16 530862 28 (0.005%) 0 (0.000%)
--> ALL PRECISIONS 3147082 272 (0.009%) 0 (0.000%)
^ permalink raw reply [flat|nested] 6+ messages in thread
* [Bug target/51751] COMPLEX16 tests fail in Lapack
2012-01-04 12:07 [Bug target/51751] New: COMPLEX16 tests fail in Lapack tkoenig at gcc dot gnu.org
` (2 preceding siblings ...)
2012-01-06 13:10 ` dominiq at lps dot ens.fr
@ 2012-01-08 10:13 ` tkoenig at gcc dot gnu.org
2012-01-08 12:29 ` dominiq at lps dot ens.fr
4 siblings, 0 replies; 6+ messages in thread
From: tkoenig at gcc dot gnu.org @ 2012-01-08 10:13 UTC (permalink / raw)
To: gcc-bugs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=51751
--- Comment #4 from Thomas Koenig <tkoenig at gcc dot gnu.org> 2012-01-08 10:13:33 UTC ---
Also reported upstream:
http://icl.cs.utk.edu/lapack-forum/viewtopic.php?f=13&t=3030&p=7974#p7974
^ permalink raw reply [flat|nested] 6+ messages in thread
* [Bug target/51751] COMPLEX16 tests fail in Lapack
2012-01-04 12:07 [Bug target/51751] New: COMPLEX16 tests fail in Lapack tkoenig at gcc dot gnu.org
` (3 preceding siblings ...)
2012-01-08 10:13 ` tkoenig at gcc dot gnu.org
@ 2012-01-08 12:29 ` dominiq at lps dot ens.fr
4 siblings, 0 replies; 6+ messages in thread
From: dominiq at lps dot ens.fr @ 2012-01-08 12:29 UTC (permalink / raw)
To: gcc-bugs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=51751
--- Comment #5 from Dominique d'Humieres <dominiq at lps dot ens.fr> 2012-01-08 12:29:24 UTC ---
For the first results in comment #3, I have used
FORTRAN = gfc -fimplicit-none -g
OPTS = -O0
DRVOPTS = $(OPTS)
NOOPT = -g -O0
LOADER = gfc -g
LOADOPTS =
and gfc was
Using built-in specs.
COLLECT_GCC=gfc
COLLECT_LTO_WRAPPER=/opt/gcc/gcc4.7w/libexec/gcc/powerpc-apple-darwin9.8.0/4.7.0/lto-wrapper
Target: powerpc-apple-darwin9.8.0
Configured with: ../work/configure --prefix=/opt/gcc/gcc4.7w
--enable-languages=c,c++,fortran,objc,obj-c++,java,ada,lto --with-gmp=/opt/mp
--with-system-zlib --with-cloog=/opt/mp --enable-lto --enable-cloog-backend=isl
Thread model: posix
gcc version 4.7.0 20120104 (experimental) [trunk revision 182883p3] (GCC)
^ permalink raw reply [flat|nested] 6+ messages in thread
end of thread, other threads:[~2012-01-08 12:29 UTC | newest]
Thread overview: 6+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2012-01-04 12:07 [Bug target/51751] New: COMPLEX16 tests fail in Lapack tkoenig at gcc dot gnu.org
2012-01-04 23:18 ` [Bug target/51751] " tkoenig at gcc dot gnu.org
2012-01-06 2:14 ` tkoenig at gcc dot gnu.org
2012-01-06 13:10 ` dominiq at lps dot ens.fr
2012-01-08 10:13 ` tkoenig at gcc dot gnu.org
2012-01-08 12:29 ` dominiq at lps dot ens.fr
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).