I don't have enough of different types of hardware to make such a
comparison either but you'll probably see about 50% of the clock speed
difference in performance for AMBER here so the difference is around 9% or
so so you might see a 4% speedup. HOWEVER!!! The higher clocked card will
almost certainly run hotter and you run the risk of it not being as
reliable because it is closer to the bleeding edge but that said I have no
firm metrics to back that up.
These cards are basically vendor over clocked cards so really no different
to you over clocking them yourself to those speeds and I'm betting the
cheaper one will probably run at that core clock no problems (for games!)
- For AMBER that could be a different situation entirely.
This is the card I have run with considerably without issue:
http://www.amazon.com/gp/product/B008BGXXB8/ref=as_li_qf_sp_asin_tl?ie=UTF8
&camp=211189&creative=373489&creativeASIN=B008BGXXB8&link_code=as3&tag=free
lydownloa-20
So feel free to go with the over clocked ones but be prepared to report
back your findings to the rest of us please.
All the best
Ross
On 11/29/12 4:35 PM, "Jason Swails" <jason.swails.gmail.com> wrote:
>On Thu, Nov 29, 2012 at 7:13 PM, Marek Maly <marek.maly.ujep.cz> wrote:
>
>> Hi guys,
>>
>> when speaking about the EVGA GTX 680s what is your opinion about this
>> variant:
>>
>> EVGA GeForce GTX680 Classified
>>
>> 1536 CUDA cores
>> 4 GB 256-bit GDDR5
>> Core clock at 1111 MHz
>> Boost Clock at 1176 MHz
>> Effective Memory clock at 6008 MHz
>>
>> Seems to me as the most powerful from the actual 4 GB EVGA GTX 680 cards
>> (and maybe from the all 4 GB GTXs 680 at the moment).
>>
>
>Same comments apply (the diff b/w this card and jonathan's 2nd card is
>about the same as the difference between the two cards he asked about).
> Again, when comparing only memory speeds and processor speeds (all else
>being equal), I only expect memory speeds to matter.
>
>Of course, since each card is >$500 a pop, I'm a graduate student, and
>HPCs
>won't buy consumer gaming cards (and certainly not different models to run
>on otherwise identical hardware), this evaluation is based on no
>real-world
>testing whatsoever. This is, therefore, an invitation for anyone that
>knows the real-world answers of these questions to drop a hint ;). (Or
>someone else's hunch/suspicion, of course)
>
>All the best,
>Jason
>
>--
>Jason M. Swails
>Quantum Theory Project,
>University of Florida
>Ph.D. Candidate
>352-392-4032
>_______________________________________________
>AMBER mailing list
>AMBER.ambermd.org
>http://lists.ambermd.org/mailman/listinfo/amber
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Thu Nov 29 2012 - 17:00:03 PST
