new Info about RV250 & R300

[jpeter]

in http://www.reactorcritical.com/
Janita Willems:
"New solutions powered by ATI, such as the RV250 and the R300 will be announced during Computex 2002. Mass production is expected in August 2002, R300 will be announced a week after RV250. The RV250 GPU features integrated Hardware MPEG in the chip and also VIVO functionality will be implemented. R300 is expected to have three times higher performance compared to the competitors."

This is starting to get very interesting

 

[Humus]

Three times higher sounds a little too much. Unless they are talking about certain special cases of course. When's computex btw?

 

[Dave Baumann]

FYI - see this thread.

 

[CoolAsAMoose]

R300 is expected to have three times higher performance compared to the competitors

Is that 3x or 4x the performance of the competitors? Remember that "3 times faster" is the same as "4 times as fast", "1 time (100%) faster" is the same as "twice as fast"...........

 

[Nappe1]

I am very interested to see, what kind of gimmicks they have to get more memory bandwidth.

to get three times higher performance than competitors (IMO means 4xperformance of competitors.), I am expecting about four times of competitors memory bandwidth or technology that drops need of bandwidth to 1/4 what competitors have.

sounds logical?
at least for me.

 

[Entropy]

I am very interested to see, what kind of gimmicks they have to get more memory bandwidth.

to get three times higher performance than competitors (IMO means 4xperformance of competitors.), I am expecting about four times of competitors memory bandwidth or technology that drops need of bandwidth to 1/4 what competitors have.

sounds logical?
at least for me.

lol!
Get real Nappe1, can't you speak Marketing?
This means that there is some obscure instance where the R300 is 3xGF3 performance. And that would only be true if it was official marketing. Since this is pre-release hype, you might be able to assume that there is some obscure instance where the R300 is twice as fast as an original GF3.

Of course, your own guesses are probably just as good as any "info" you gain from rumours such as these.

Entropy

 

[Nappe1]

I am very interested to see, what kind of gimmicks they have to get more memory bandwidth.

to get three times higher performance than competitors (IMO means 4xperformance of competitors.), I am expecting about four times of competitors memory bandwidth or technology that drops need of bandwidth to 1/4 what competitors have.

sounds logical?
at least for me.

lol!
Get real Nappe1, can't you speak Marketing?
This means that there is some obscure instance where the R300 is 3xGF3 performance. And that would only be true if it was official marketing. Since this is pre-release hype, you might be able to assume that there is some obscure instance where the R300 is twice as fast as an original GF3.

Of course, your own guesses are probably just as good as any "info" you gain from rumours such as these.

Entropy

Entropy: maybe that smiley in the end wasn't enough to show some irony.

So I just wanted make clear that those words don't make any sense at all from technical point of view.

and no, My infos don't come from marketing segment. And besides, after Matrox Parhelia is now out including 4 Vertex Shaders and 4 pipelines with 4 TMU's on each with 256 Bit DDR memory interface, I really don't believe that there is much problem with my reliability for those who are interested about same things as I am.

Of course, I am interested all the happenings in gfx market. I have some opinions about 1st generation DX9 Hardware based on some facts, but I doubt that they could be discussed openly without me being stamped with fanboyishm stamp. (and I don't mean that I would be stamped as ATI / nVidia lover. I have already disscussed about my thoughts at ICQ and some of the users of this board at least partially agree, but this time is better wait and see.)

 

[Entropy]

A man who has shown unreasonable optimism concerning BitBoys Oy, has to suffer the indignity of people believing him serious even when posting the most blatantly obvious irony.







Entropy

 

[Nappe1]

A man who has shown unreasonable optimism concerning BitBoys Oy, has to suffer the indignity of people believing him serious even when posting the most blatantly obvious irony.



Entropy

and "people" in that sentence considers more persons than just you?

I have my reasons to believe Bitboys, But as said many times before, there's nothing new to talk about bitboys right now. Let's get back on it when there is something without depending whatever it will be.

 

[Pete]

"[...]R300 is expected to have three times higher performance compared to the competitors."

This seems logical only if compared to the current gen (e.g., 8500), as I recall an interview with an ATi employee saying ATi typically releases new cards >=2x more powerful than their previous gen. I find it almost impossible to believe that one manufacturer will outperform a competitor's same-gen product by more than fractions of multiples, except in highly specialized cases (which may be possible if The Inquirer's next-gen specs are true regarding PS's and VS's).

As the R300 seems to be a single-chip design, the NV30's potential separate TT&L design seems to be much more potentially radical, especially with the main chip featuring as many transisitors as ATi's--that leaves NV with more transistors, which, in my relatively uninformed view, means more features.

 

[Sharkfood]

Well, they also didnt put any price point and their specification could have referred to an entire "line" of R300 products.

Who's to say a 2 or 4 chip "MAXX" version of the R300 wont be announced, that will cost an arm and a leg, yet the single chip R300 will be the mainstream, ~$300 product line most gamers will pursue.

 

[Basic]

Wasn't GF3 supposed to be 7 times faster than GF2 according to some PR from nvidia? I'm sure this is a similar number. In theory or in some highly specialized benchmarks, R300 might be 3 times faster than the current generations leader.

 

[Livecoma]

Here is a sort of related question...

How many transistors could you get @ .13 while keeping the size of the chip similar to what NV25 is @ .15?

 

[pascal]

from the info of this page http://www.tsmc.com/english/technology/t0101.htm maybe 25% more transistors (79 millions transistors total). But it has eight metal layers then it could be more complex and compact.

 

[Gunhead]

Pete, oh God, is that "Nvidia to use a separate t&L chip" bullshit still walking? I'm sure the guy who started it has just rolled over and died laughing.

Nappe darling, Parhelia isn't out yet, so you do have a bit of a problem with your credibility

<-- has seen Pyramid3D in action

 

[Pete]

So it's confirmed to be BS? It makes sense to me because 3dfx was all set to do it, but I suppose the added complexity might not go over too well with nV's card makers (plus, it doesn't mesh with their history). Also, I know nothing of video card engineering concerns. =)

 

[Nappe1]

Nappe darling, Parhelia isn't out yet, so you do have a bit of a problem with your credibility

for those who know more exactly what I knew some time ago, I don't think that being problem... and for rest of... well, I could not care less. In the eyes of them I don't have any credibility left after last autumn, so it really doesn't matter anymore.

 

[Typedef Enum]

Nappe1, you're still OK in my book...

As the old saying goes...Shi* happens. We'll see what happens in the future.

 

[mdhome]

Nappe1 was also right about aspects of Parhelia, even as early as last Xmas. Even after the BitBoys debacle, he has improved in my standing. I also think he's OK

Regards Michael

 

[Gunhead]

Yes, Nappe has had good info. No problem there. I only nitpicked at his saying "Parhelia is now out" when the product isn't. OTOH, of course the design and the features are locked and public -- I guess that's what's relevant here... Just had nothing better to do, I guess.

Additionally, I don't think Nappe had any problem in the Avalanche incident. He was the first to report what was happening, and other sources have (later) verified that he reported the events correctly. Bitboys got royally screwed by Infineon, but I didn't feel like shooting at the messenger at all. (Just too bad they couldn't send any review cards to websites.)