NVIDIA NV30 & NV40 Scoopage: Details

[boi]

The hardware keeps getting better and better while the software is at least 2 years behind

What! I can't even get 75fps in GTA3 or WC3 w/ a p4 2.4 and a g4 4600

 

[daywalker]

You americans will always fall for every stupid new video card that Nvidia throws in the market every 6 months.

Or does anyone think that a Geforce4 Ti4600 wont be enough or "outdated" for gaming in 2003? Or for playing Doom3?

 

[FatAlbo]

What! I can't even get 75fps in GTA3 or WC3 w/ a p4 2.4 and a g4 4600

I doubt the GTA3 problems are due to a very sophisticated engine. IMHO, an unoptimized port is the likely culprit.


Now relevant to multichip solutions, how would the introduction of multiple chips affect PCB sizes? My elementary understanding would say that more chips would mean more traces and a larger PCB.

 

[ElFenix]

Originally posted by: daywalker
You americans will always fall for every stupid new video card that Nvidia throws in the market every 6 months.

Or does anyone think that a Geforce4 Ti4600 wont be enough or "outdated" for gaming in 2003? Or for playing Doom3?

what does being american have to do with any of this?

 

[BrunoPuntzJones]

Originally posted by: daywalker
You americans will always fall for every stupid new video card that Nvidia throws in the market every 6 months.

Or does anyone think that a Geforce4 Ti4600 wont be enough or "outdated" for gaming in 2003? Or for playing Doom3?

Yeah because Nvidia only sells to the US

 

[nortexoid]

SMP is ugly.

i'm not buying any card using more than one GPU - too large, too costly, too inefficient (usually), too hot, too loud, too too.

 

[nortexoid]

also, we coudl assume that each GPU would require it's own memroy...

going with the lowest odd number between 1 and 8, that would be 3...say a min. of 128MB per GPU (which is min. on the GF4s now), it would require the card support 384MB of DDR 900MHz memory on it...:Q

of course, this assumption could be wrong.

SMP blows!

 

[grant2]

You americans will always fall for every stupid new video card that Nvidia throws in the market every 6 months.
Or does anyone think that a Geforce4 Ti4600 wont be enough or "outdated" for gaming in 2003? Or for playing Doom3?

You (insert irrelevant adjective here) will always fall for the stupid idea that anyone who buy this card own a Ti4600.

Or do you think that people who own geforce256s, g400s, Rage Furys are allowed to upgrade their "outdated" hardware for 2003?

 

[paralazarguer]

going with the lowest odd number between 1 and 8, that would be 3

Buddy, the lowest odd number between 1 and 8 is 1.

 

[Krk3561]

Buddy, the lowest odd number between 1 and 8 is 1.

Since when has 1 been between 1 and 8. I though that that range included 2,3,4,5,6, and 7. The odds in that range are 3,5,7. Therefore the lowest would be 3.

 

[Leo V]

Originally posted by: 7757524

going with the lowest odd number between 1 and 8, that would be 3

Buddy, the lowest odd number between 1 and 8 is 1.

You're both misinterpreting the post, which said: "The number of chips is an odd number. That's all I can say without giving out the exact number. I'll say this though. It's under 8.". No need to get excited about this!

 

[grant2]

going with the lowest odd number between 1 and 8, that would be 3
Buddy, the lowest odd number between 1 and 8 is 1.

a 1-chip design wouldn't be "multichip" now would it?

 

[paralazarguer]

The number of chips is an odd number. That's all I can say without giving out the exact number. I'll say this though. It's under 8.

Who said it has to be multi? Just because it is multi capable (supposedly) doesn't mean that it's going to ship multi at first or ever.

The number of chips is an odd number. That's all I can say without giving out the exact number. I'll say this though. It's under 8.

The lowest number would be 1. Not 3.

 

[nortexoid]

first he calls me buddy, then "he dumb".

but anyway....even assuming all 3 (or more) GPUs could share the same memory, how efficient would that be?...i imagine that would be quite difficult to accomplish as well.

i doubt this thing will have more than 2 GPUs...i could see 2 w/ 256MB onboard...that would cost a small fortune...no market for a gaming card that pricey, i'd imagine.

besides, why upgrade when i still have this cutting-edge 3 year old GF 256?

 

[paralazarguer]

Take a look at a video card. Try to imaging 3 GPUs on it. That's insane. Then ammount of power, heat, noise would be unmanagable not to mention price. When he says that it will be an odd number lower than 8 he's letting us know that there will "only" be 1GPU. 3 is just impractical. Common sense. since nvidia is going to .13 and ATI is sticking with .15, nvidia has no reason to go up to two gpu. They can easily hit higher clock frequencies.

 

[BD2003]

The explanation for odd chips is simple enough to me. No reason to have more than one T&L unit, since the scene is exactly the same for all the rendering units. But the rendering chips need to be in an even configuration due to SLI, since its a lot easier and faster to split the screen into halves or quarters instead of thirds.

So that leaves us with 1 chip or 2 chips as the baseline, depending on whether or not the GPU is fully integrated, and 3, 5, and 7 thereafter.

 

[MadRat]

Just ignore 7757524. He's trippled his post count in the last month with similar garbage.

 

[ElFenix]

Originally posted by: BD2003
The explanation for odd chips is simple enough to me. No reason to have more than one T&L unit, since the scene is exactly the same for all the rendering units. But the rendering chips need to be in an even configuration due to SLI, since its a lot easier and faster to split the screen into halves or quarters instead of thirds.

So that leaves us with 1 chip or 2 chips as the baseline, depending on whether or not the GPU is fully integrated, and 3, 5, and 7 thereafter.

sounds like rampage to me.


anyhow, why does 100 million transistors on 1 chip require less power than 50 million on 2 with the same transistor size? i don't see how one is more/less efficient than the other in terms of energy required/heat production. what i do see is that yields on the 50 million transistor chip are going to be higher, which leads to a lower costing part assuming the board isn't complex enough to offset it.

 

[paralazarguer]

Just ignore 7757524. He's trippled his post count in the last month with similar garbage.

Just ignore madratz. He's got a personal vendetta and the only things he's posted lately are personal attacks. Thanks for the post with relevent info on this thread!!!! In fact, I spend a lot of time providing tech support and answering questions. For these threads which are just speculation, if you don't like my opionion, tough!!! Feel free to PM a mod and please keep the personal attacks out of here!

 

[Snoop]

Posts: 578
Joined: Jun 2002
You joined 1 month ago and have 578 posts... thats gotta be a freaking record

 

[paralazarguer]

I like to help as much as possible. Contrary to what madratz says, I think if you look you'll see that I do help a lot of people out. I have opinions that not everyone agrees with but really, that's not appropriate at all. I really don't appreciate the slander.

 

[MadRat]

In most cases, 7757524, you go around and tell people how dumb are their arguments. This isn't exactly helpful nor appreciated by other forum members. Its not unusual to be overzealous when you first join the forum, but I suspect in your case its nothing to do with zealotry. You seem to have a chip on your shoulder where you are the only expert here. Its also annoying when you come in as a newbie and jack up your post count with meaningless statements.

 

[Sunner]

MadRat, the posts I've seen from 73485745(whatever, change bnick if you expect me to spell it right) have so far been quite mature, and definately above the standard of this forum(which doesn't always say much these days, unfortunately).

Anyways, like someone said, nVidia's NV30 is gonna be .13 while the R300 is gonna be .15, nVidia has all the advantage they need right there, I see no reason for them to go multichip, even if their chip is capable of it.

 

[paralazarguer]

Thanks Sunner
Anyway, what someone mentioned is having different chips do different functions and basically be 1 GPU spread out. That's one thing that I haven't thought of. I was thinking of three full fledged GPUs (similar to voodoo 5000) which would be crazy. So I assumed he meant only 1. If you're talking about spreading out a single GPU across many chips I guess that would work. I don't know why, though, it makes it harder to design the board, there would be more overhead...just doesn't make any sense TO ME. My $0.02

 

[Nick5324]

makes me wish i was rich.....