View Single Post
  #9  
Old July 9th 04, 08:59 PM
J. Clarke
external usenet poster
 
Posts: n/a
Default

CapFusion wrote:

[snip]
"John Russell" wrote in message
...
Nvidia have the right to the acronym SLI but don't get confused into
thinking this is Scan Line Interleaving,as it isn't.

[/snip]
According to NVIDIA for SLI = Scalable Link Interface

[snip]

From what i've seen only one card is responsible for generating the
image, the other is just a second GPU which performs tasks allocated by
the first
GPU. This is far more efficeint than the Voodoo solution as no graphics
calculations are done twice.

[/snip]
Each GPU will start half and will change depending on the load. Software
will determine which load to distribute to which GPU.

[snip]
It also means that cards don't have to match so another upgrade path
becomes available. You can add a new faster card as the master and keep

the
old one as the slave GPU.

[/snip]
You will need both identical card. Best is to have that same card from the
same manufacturer.


Since there are no boards on the market today or coming in the near future
with two PCI-Express 16x slots (read the fine print on the Alienware very
carefully then read the spec sheets on the components they use and you'll
find that it has one 16x and one 8x), if it depends on the boards being
identical then it's not going to work at all for a good long time.

CapFusion,...


--
--John
Reply to jclarke at ae tee tee global dot net
(was jclarke at eye bee em dot net)