Next NV Infomania

This is not a chop. It's a very reliable source.

 

As long as your new toolchain supports your designlanguage swaping is not a problem.

I know that 90nm is not a shrink because of this I am talked about funktionlibs. If you use this nice solution the developer of the lib take care about the process changes for you. All you need to do is to buy an upgrade for the new process.

 

I agree, and a poor chop at that. The perspective (or lack of it in the lettering) gives it away.

 

I'd get a better "reliable" source, that is most definately a chop.

 

making an analogy to things that are possible and things that are possible for the common man..., you can do 16xAA, I can do triple buffering, equal now?

 

Ummm my point wasn't that you were inferior because you can't do 16xAA. Where did you get that from. I simply stated that it is available now, in current drivers, on existing products to refute your claim that it is vapourware at the moment.

 

but enabling it would make your drivers non whql though, right?

Sure it's there, just like triple buffering, but I can't tag it, or set a slider for it in my control panel to make it work, I need third party tools or unsupported drivers..

 

I think you've forgotten where this all started. You drew an analogy between 16xAA support and Crossfire. No third party tool or unsupported driver can make Crossfire boards magically appear in your machine......

PS. WHQL is for wussies

 

Haha, probably. I've been through at least four drivers this week alone with all this BF2 madness.

 

[quote="trinibwoy"
Haha, probably. I've been through at least four drivers this week alone with all this BF2 madness.[/quote]

Lol.. and still everyone complains about Ati's drivers

anyway, my analogy should be 16aa / 3buf .. I think I wanted to rant about "if you can do 16aa, my x850xt phase changed could beat your 7800gtx"

Anyway, this quote from the Inquirer also sparks curiosity

 

You mean standard cell libraries. Which get you not much at all since things like RC don't scale linearly, you have inductance to worry about, new corner cases and critical paths. No to mention ram arrays, etc.

While you don't have to rewrite the RTL, porting from one process to another is non-trivial.

Aaron Spink
speaking for myself inc.

 

No it doesn't. Unless you're an idiot.

 

I realise that the progression of an ASIC design from a specific fabrication node and it's associated cost-reduction node (i.e. 250nm --> 220nm, 180nm --> 150nm, 130nm --> 110nm, etc) does not translate unchanged (has any ASIC -- non-GPU included -- ever bucked this trend/limitation?) to the next major node (e.g. 150nm --> 130nm) and that obviously these transitions are significantly more complex than suggested, but since this query was predicated by an editorial I have subsequently found to have been published a number of weeks ago and already discussed/dismissed and/or superceded by more recent information, the suggestion that was I looking for a "yay" or "nay" on is no longer relevant.

Apologies for the repost of Josh's G70 article, however Digi made a valid point about timestamping -- Josh, I know you read and post on the boards here -- please include the publication date in future to avoid this confusion . As much as I enjoy reading the B3D forums, it's difficult to keep track of every thread and comment, although in future I will make more effort to check the status (date, ammendments, previous postings, etc) of information before posting. We can't all be like Digi!

Cheers,


BrynS

 

One things for sure, with the 7800 and the FX-57 coming out next week ( maybe both on Tuesday ) I'm not going to be getting much work done on that morning whilst planning my new system

2 x 7800, new 24 pin 600w power supply, new PCie motherboard and FX-57 is all I would need for my new system, I already have the RAM ....

Actually, maybe I don't have enough RAM .

By 5pm that day I can go home from work after finally deciding I cannot afford my new system and putting it once again firmly on the back burner.

 

Assuming that the overclock is above 500MHz then this is rather interesting. The 6200 and 6600 chips based on nv43 use 110nm and they also overclock well, though of course they have a lot less transistors than G70.

Now it seems that the G70 on the cheap 110nm process is also overclocking well, so in that case it does seem that if some video card manufacturers can stick 1400MHz memory on there then these will definitely be the ones to go for.

Menawhile Mr SolidSnake at nvnews seems to have gone off the beaten track with his tidbits

http://www.nvnews.net/vbulletin/showpost.php?p=632102&postcount=58

Hmmm.

and then

http://www.nvnews.net/vbulletin/showpost.php?p=632118&postcount=64

700Mhz ok, but 200watts ?

 

Probably the lettering was adjusted for better reading afterwards because it wasn't readable at first. But still it's the real thing.

 

Well then that adjustment destroys the credibility of the image. It doesn't matter what it's a photo of, real thing or not. In photo-journalism image manipulation is beyond the pale.