[RC5] 999 to 1000 blocks and P5: MMX vs Classic

Chris Jones fiddles at distributed.net
Mon Jan 12 18:59:30 EST 2004


On 11-Jan-2004, Nuno Louren?o wrote:
> Hi.
> 
> 999 to 1000 blocks
> I don't understand why the client can't go from 999 to 1000 blocks in the in
> file when fetching and in the out file when writing a worked block. It
> doesn't have that problem with the "-import" and "-update" command (using a
> alternate-buffer-directory). I am using the dnetc v2.9005-483-GTR-03033119
> for Win32 client and Win XP Pro.

If you have 999 blocks in your out buffer, why not simply flush them? Blocks aren't 
much good unless you send 'em back! :-P
 
> P5: MMX vs Classic
> The latest distribudet.net client's for RC5-64 had diferent cores for
> Pentium MMX and Pentium Classic, and the MMX core was about 50% faster than
> a Pentium Classic at the same MHz. Recently,  I decided do run the
> distributed.net client for RC5-72 on one Pentium MMX and two 2 Pentium
> Classic and found out that they had the same key rate per MHz (and were
> running the same core I think). Why did the Pentium MMX lose de +50% per MHz
> speed advantage on the shift from RC5-64 to RC5-72?
> 
> Thanks,
> Nuno Louren?o
> 
> _______________________________________________
> rc5 mailing list
> rc5 at lists.distributed.net
> http://lists.distributed.net/mailman/listinfo/rc5

-- 
Chris Jones
fiddles at distributed.net


More information about the rc5 mailing list