[SciPy-dev] random module not working on opteron

Travis Oliphant oliphant at ee.byu.edu
Mon Oct 24 09:48:19 CDT 2005


Arnd Baecker wrote:

>On Sun, 23 Oct 2005, Robert Kern wrote:
>
>  
>
>>I think the issue is that the state is an array of 624 unsigned longs.
>>rk_seed() essentially just does 32-bit arithmetic and so fills those
>>longs with numbers up to 2**32-1. rk_randomseed(), however, takes raw
>>bits from /dev/urandom and drops them into the array. I believe that the
>>algorithm is expecting 32-bit integers.
>>
>>I've applied a mask to the state after it gets seeded by /dev/urandom.
>>Please checkout the latest version of randomkit.c from SVN and give it a
>>try. We may need to make this more robust later, but for now it might work.
>>    
>>
>
>I went straight to a full build (because that is done automatically
>via a script ;-):
>
>It looks good - scipy.test(10,10) runs through on the opteron
>the first time after quite a while!
>
>Ran 569 tests in 31.891s
>
>FAILED (failures=14, errors=30)
>Out[3]: <unittest.TextTestRunner object at 0x2aaab5cfbad0>
>  
>
The buld logs are still useful (I recently added the remaining packages 
to the build -- but they need 64-bit cleaning).

Also, I'm not getting that many errors and failures.  So, the details 
would be helpful.

-Travis




More information about the Scipy-dev mailing list