Bug#655586: libatlas3gf-base and "Illegal instruction"

Stefano Lattarini stefano.lattarini at gmail.com
Thu Jan 12 15:14:40 UTC 2012


On 01/12/2012 04:07 PM, Sylvestre Ledru wrote:
> Hello,
> 
> Next time, please provide the information on the top if the bug report
>
Sorry, will do.

> Le jeudi 12 janvier 2012 à 15:57 +0100, Stefano Lattarini a écrit :
>> Programs linked with libatlas3gf-base unexpectedly die with an
>> "Illegal instruction" signal:
>>
>>   $ cat > foo.c <<'END'
>>   #include <atlas/cblas.h>
>>   int main (void)
>>   {
>>       static double a[2*2];
>>       static double b[2*2];
>>       static double c[2*2];
>>       double alpha = 1.0, beta = 0.0;
>>       cblas_dgemm (CblasRowMajor, CblasNoTrans, CblasNoTrans, 2, 2, 2,
>>                    alpha, a, 2, b, 2, beta, c, 2);
>>       return 0;
>>   }
>>   END
>>   $ gcc -lcblas foo.c
>>   $ ./a.out; echo status = $?
>>   Illegal instruction
>>   status = 132
> 
> What is your CPU ? 
> $ cat /proc/cpuinfo
> 
processor       : 0
vendor_id       : AuthenticAMD
cpu family      : 6
model           : 6
model name      : AMD Athlon(tm) XP 1800+
stepping        : 2
cpu MHz         : 1532.918
cache size      : 256 KB
fdiv_bug        : no
hlt_bug         : no
f00f_bug        : no
coma_bug        : no
fpu             : yes
fpu_exception   : yes
cpuid level     : 1
wp              : yes
flags           : fpu vme de pse tsc msr pae mce cx8 sep mtrr pge mca cmov pat pse36 mmx fxsr sse syscall mmxext 3dnowext 3dnow up
bogomips        : 3065.83
clflush size    : 32
power management: ts


> Did you try with an optimized version of atlas?
>
No.  Should I?

> (see its README.Debian file).
> 

Thanks,
  Stefano





More information about the debian-science-maintainers mailing list