Moinak Ghosh wrote:
> On Tue, Aug 18, 2009 at 11:35 PM, Kaya Saman<SamanKaya at netscape.net> wrote:
>   
>>>   Also instead of splurging on so much RAM consider getting a cool Nvidia
>>>   display card (9x series or GTX if you can appord it) with 1GB GDDR3
>>> VRAM.
>>>   Then play on with CUDA, OpenCL, Vadapav ... er um ... VDPAU (The Mplayer
>>>   in BeleniX has VDPAU support) and other cool stuff.
>>>
>>> Regards,
>>> Moinak.
>>>
>>>       
>> Was thinking about ATI X-fire cards as I can grab 2 then use them with
>> triple or quad head displays.....
>>
>>     
>
>    ATI will give you trouble on Linux and BeleniX. ATI does not provide
>    drivers for OpenSolaris so nada. Linux drivers are shaky. Use Nvidia
>    in dual-SLI mode.
>    But believe me, most of you 25GB RAM will be wasted investment.
>
> Regards,
> Moinak.
>   
I will start with about 8GB, probably or 12GB and then see how much I 
need later if I need any at all.... I mean that is the real advantage of 
having modular memory slots!

Like servers too, I will be going for x86 systems rather then SPARCs as 
SPARCs are not only more expensive but the added computational abilities 
I wouldn't be using at home like multiple concurrent floating point 
arithmetic or the need for a cisc big endian system.

I plan to use one server as a Ray server meaning that I will probably 
need about 8 - 12GB RAM in there, as I wana hook my whole family upto it 
round the world and with new VDSL2 internet speeds that should be possible.

I mean currently I have an idea of what I want to do but will need to 
plan it closer to the time as now I don't even have enough money to do 
anything.

Anyway it's like in networking industry when creating baselines, you 
start with something you think will be ok, then adjust according to what 
you need.

But then also, I have many friends @ belenix.org to kick me up the butt 
and/or give me advice and tell me when I'm being silly :-) - and I 
really do value and appreciate that input and relationship!!

Am currently on the Nvidia site checking out the multi-head GPU's 
actually, things have come along way since I last looked which was many 
years ago!

K

Reply via email to