This is the mail archive of the gcc@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]

RE: [Fortran] Lapack test suite results for i686-pc-linux-gnu


These results may not be too bad:
*	the _XV failures are a known LAPACK problem
*	the IZAMAX failure is due to excess precision  
*	some of the others may be a "near miss" and OK
*	the _gd failures look suspicious

Could you have a look at the *.out files that contain test failures.  

For example, on irix6.5, csep.out has a couple of failures.  Near the top of
the output file we see the relative precision required

   Routines pass computational tests if test ratio is less than   50.00

and later on we see the failure reports

   Matrix order=    5, type= 9, seed= 894,3587, 994, 217, result  36 is
60.73
   CST:    1 out of  4662 tests failed to pass the threshold

This is probably OK, as we haven't failed by much.  In some cases we can
fail by several orders of magnitude, and this often indicates a compiler
bug.

> -----Original Message-----
> From:	Peter Schmid [SMTP:schmid@snake.iap.physik.tu-darmstadt.de]
> Sent:	Monday, 14 May 2001 6:41
> To:	gcc@gcc.gnu.org
> Subject:	[Fortran] Lapack test suite results for i686-pc-linux-gnu 
> 
> Appended are the test suite results for three different g77 compiler
> versions on the i686-pc-linux-gnu target. My system setup is: 
> SuSE 7.1, glibc 2.2, binutils version 2.11.90.0.4, linux 2.4.4.
> I followed the instructions in the lapack-guide.html exactly.
> 
> Are all of these failures because of genuine compiler or runtime bugs,
> or are at least some related to the peculiarities of the intel
> implementation of the floating point unit (excess precision when a
> double is kept in a register in contrast to a double that is stored in
> memory, etc)?  
>   
> I cannot run the timing test suite since xlintimd crashes for all
> tested compilers on startup in the function f_setarg located in the file
> libf2c/libF77/setarg.c.  
> 
> 
> Hope this helps,
> Peter Schmid
> 
> g77 version 2.95.2 19991024 (release)
> cgd.out: CGV drivers:      4 out of   1092 tests failed to pass the
> threshold
> csep.out: CST:    1 out of  4662 tests failed to pass the threshold
> csep.out: CST drivers:      2 out of  11664 tests failed to pass the
> threshold
> csvd.out: CBD:      1 out of   4085 tests failed to pass the threshold
> dgd.out: DXV drivers:    200 out of   5000 tests failed to pass the
> threshold
> sgd.out: SXV drivers:     37 out of   5000 tests failed to pass the
> threshold
> ssep.out: SST:    1 out of  4662 tests failed to pass the threshold
> ssep.out: SST drivers:      1 out of  14256 tests failed to pass the
> threshold
> zgd.out: ZGV drivers:      4 out of   1092 tests failed to pass the
> threshold
> zgd.out: ZXV drivers:     24 out of   5000 tests failed to pass the
> threshold
> 
> 275 failures in total
> 
> 
> g77 version 3.0 20010503 (prerelease)
> cgd.out: CGV drivers:     66 out of   1092 tests failed to pass the
> threshold
> csvd.out: CBD:      1 out of   4085 tests failed to pass the threshold
> dgd.out: DXV drivers:    200 out of   5000 tests failed to pass the
> threshold
> sgd.out: SXV drivers:     37 out of   5000 tests failed to pass the
> threshold
> ssep.out: SST drivers:      1 out of  14256 tests failed to pass the
> threshold
> zgd.out: ZGV drivers:     61 out of   1092 tests failed to pass the
> threshold
> zgd.out: ZXV drivers:     24 out of   5000 tests failed to pass the
> threshold
> 
> 390 failures in total
> 
> 
> g77 version 3.1 20010512 (experimental)
> cgd.out: CGV drivers:     66 out of   1092 tests failed to pass the
> threshold
> csep.out: CST:    1 out of  4662 tests failed to pass the threshold
> csep.out: CST drivers:      1 out of  11664 tests failed to pass the
> threshold
> csvd.out: CBD:      1 out of   4085 tests failed to pass the threshold
> dgd.out: DXV drivers:    200 out of   5000 tests failed to pass the
> threshold
> sgd.out: SGV drivers:     10 out of   1092 tests failed to pass the
> threshold
> sgd.out: SXV drivers:     37 out of   5000 tests failed to pass the
> threshold
> ssep.out: SST drivers:      1 out of  14256 tests failed to pass the
> threshold
> zgd.out: ZGV drivers:     63 out of   1092 tests failed to pass the
> threshold
> zgd.out: ZXV drivers:     24 out of   5000 tests failed to pass the
> threshold
> 
> 404 failures in total
> 
> Additionally, there is a failure in the "Test of subprogram number 10
> IZAMA" from the Complex BLAS Test Program Results for all g77 compiler
> mentioned above.
> 
> Debugging xlintimd compiled by gcc-3.0
> (gdb) set args < dtime.in > dtime.out 2>&1
> (gdb) r
> Starting program: /mnt/LAPACK/TIMING/xlintimd < dtime.in > dtime.out 2>&1
> 
> Program received signal SIGSEGV, Segmentation fault.
> 0x080db2f6 in f_setarg (argc=1, argv=0xbffff83c) at setarg.c:27
> (gdb) bt
> #0  0x080db2f6 in f_setarg (argc=1, argv=0xbffff83c) at setarg.c:27
> #1  0x080db1e3 in main (argc=1, argv=0xbffff83c) at main.c:55
> #2  0x40057baf in ?? ()
> (gdb) l
> 22	f_setarg(argc, argv) int argc; char **argv;
> 23	#else
> 24	f_setarg(int argc, char **argv)
> 25	#endif
> 26	{
> 27	f__xargc = argc;
> 28	f__xargv = argv;
> 29	}
> (gdb) info locals
> argc = 1
> argv = (char **) 0x1
> (gdb) p *argv
> Cannot access memory at address 0x1
> (gdb) up
> #1  0x080db1e3 in main (argc=1, argv=0xbffff83c) at main.c:55
> (gdb) l
> 50	main(argc, argv) int argc; char **argv;
> 51	#else
> 52	main(int argc, char **argv)
> 53	#endif
> 54	{
> 55	f_setarg(argc, argv);
> 56	f_setsig();
> 57	f_init();
> 58	#ifndef NO_ONEXIT
> 59	ONEXIT(f_exit);
> (gdb) p argc
> $1 = 1
> (gdb) p argv
> $2 = (char **) 0xbffff83c
> (gdb) p *argv
> $3 = 0xbffff9ab "/mnt/LAPACK/TIMING/xlintimd"
> 


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]