NVIDIA Fermi: Architecture discussion

I haven't been following this thread much the last few weeks, is the card out yet?

 

hey thats a pretty smooth PR man . Got give it to ATi.

 

You're real optimistic, don't think the chips are even out yet

 

Tesselation

 

Trinibwoy, why didn't you link the two articles you selectively quoted from? Or is context in a conversation irrelevant?

Jawed

 

+DisplayLists, +texture compression, +SM5.0. Its not as though DirectCompute isn't significant for gaming either (witness the DiCE lighting demo or the Dirt 2 lighting).

 

Has anyone tested it?

Jawed

 

I did link them, look again Btw those quotes were perfectly fine in context of my Fuddy comment and summed up the articles so I'm not really sure where your problem lies. Maybe if you tried reading the articles first.....but it's just more of the same:

Quite Fuddy, no?

 

There are no fair/usefull tests.

 

Yes, I have done that a while ago...perhaps it should be done again to see if things have budged at all. And I'm not sure what MfA has in mind, but IMHO and IME x264 is preferable to all GPU assisted video encoders currently out if one wants high-quality encodes, and manages to be equally fast if not faster under many circumstances. ATI's Stream assisted stuff produces crap output, nV's is better, but is still inferior to a properly configured x264 output.

 

Don't know about the second one, but this one doesn't sound fuddy at all. At the moment, Nv doesn't have a new lineup, and the meat of their old one, GTX260-285, are EOL. Edit: plus their Fermi presentation made no mention of gaming aspect of Fermi. What are game players supposed to think?

 

You linked two distinct articles in a single post which contains a single link?

Oh I did read the articles, after I went and found the second one.

The second comment definitely is, since NVidia clearly isn't forcing CUDA down anyone's throat. There's loads of people champing at the bit for Fermi, it seems. Presumably based on their NDA'd insight...

The funny thing is that these two articles, a week apart, have the look of being derived from a single discussion.

Jawed

 

Using the built-in encoding that's part of W7?

Jawed

 

It's just payback for that whole "ATI's given up" when they merged with AMD.

 

And the most obvious conclusion is that they're "sneering at gamers"?

When I was reading the first article the second was right there next to it in the "related articles" section so didn't think it was necessary to link both.....

Yeah, it probably was. I didn't understand that comment either, it seems to me that in the past few years Nvidia has been an enabler more than anything else, proprietary standards aside. Where did you see that stuff that folks are chomping at the bit for Fermi?

 

I saw someone purporting to have inside information about pre-orders already being taken. NVidia may well be planning to deliver early cards for such non-gaming customers. Can't remember where I saw it, now... Maybe it was XtremeSystems forum?

I think it's reasonable to presume that there are CUDA-specific customers using engineering samples right now. Larrabee should be competing for that same mindshare before long, too, I imagine...

Also, if CUDA 3.0 is required to make Fermi work, then the arrival of CUDA 3.0, able to support older cards, is prolly no coincidence.

Jawed

 

Fudo is saying something about pre-orders but I thought it was just his usual pissing into the wind.

 

For what a few HTPCs?

 

HP for one they have already stated this if I'm not mistaken.

 

Some pre-emptive "but you said thanksgiving" scrambling.