Re: Looking for CUDA advice

Joseph Kaye <[email protected]>
Newsgroups gmane.comp.distributed-net.rc5
Message-ID <[email protected]>
Hey Bob,

What is the clock rate of the EVGA and the BFG Tech?  Another thing that 
I have found is setting the priority of the client (in the client 
config) to 2 really helps the client "be all that it can be".

--Joseph



Bob Steward wrote:
> I've run 3 different Nvidia cards with the CUDA drivers and Dnetc client, 
> and from the cheapo clone 8000 series I was getting really disappointing 
> key rates in the 20,000 per second range.  I nabbed an EVGA brand 9800GT 
> with 1 GB RAM and saw serious improvement to an average of around 300,000 
> keys per second.  Good stuff.  Had the chance to try a BFG Tech 9800GT with 
> same 1 GB RAM, and was disappointed to only see about 2/3 the speed of the 
> EVGA with the same RAM and chipset.  The BFG Tech card bounces around 
> 214,000 keys per second.
>
> All this is on a Windows XP SP3 box with a Phenom II processor (which is 
> simultaneously running the client with 4 processors enabled) at an average 
> of about 50,000 keys per second.  Turning off the Phenom client(s) seems to 
> make no difference to the Nvidia CUDA clients, and running only the Phenom 
> II with 4 processors enabled and no Nvidia CUDA doesn't seem to make any 
> real difference in the Phenom's keys per second, so I run them both.
>
> Tried the EVGA card on an XP SP3 box with an Athlon dual core 5600, and it 
> runs about the same speed as the Phenom platform.
>
> All this to say, I am going to stay away from the BFG Tech line, and look 
> for others that perform well.
>
> I have a new motherboard ordered with multiple 16x PCIe slots, and am 
> looking to populate it with very high performance cards, as a means to have 
> a "stats monster" box.  It will be running another Phenom II CPU.
>
> Looking for others real world numbers for the Quadra series and the various 
> GTX cards.  Would like to know your OS, CPU, brand of card, and keys per 
> second, etc. so I can compare apples to apples.
>
> Also the current CUDA beta is about to expire, is there going to be another 
> beta, or are they going to go live with an official public client?
>
> Thanks for your input,
>
> Bob Steward
> Birmingham, AL 
>
> _______________________________________________
> rc5 mailing list
> rc5-Ra3b/QYEcJ3d140v2zMXi0fjHoOT/h/[email protected]
> http://lists.distributed.net/mailman/listinfo/rc5
>
>   

_______________________________________________
rc5 mailing list
rc5-Ra3b/QYEcJ3d140v2zMXi0fjHoOT/h/[email protected]
http://lists.distributed.net/mailman/listinfo/rc5
lmpx.com only provides a reader for public news (NNTP) servers. It is not affiliated with the servers or forums shown here and is not responsible for the content of articles, which is written by their respective authors.