16 pipes for R420 too?

My post was directed toward people always buying into marketing hype and getting their expectations too high.

NV30 and 35 were both attempts to regain the performance crown nVidia lost in September of '02.

 

But is that really the case? I tend to think they only "lost the crown" when their competing product was released and wasn't up to scratch. Sure, R300 was out since the previous September, and was the fastest at the time, but had no comparable "next gen" NVIDIA part to beat.

That's the way I tend to look at it anyway.

 

Well in the strict sense there is no FUD. There is not released product from anyone yet.

 

madshi,
I'm aware of this definition, that's why i'm using FUD. It's interesting that you chose the part of the definition that is convinient to you. May i do so?

 

There's no reason to split any hairs what that dumb performance crown concerns anyway. Fact is that NVIDIA faced for the time ever since the R300 such fierce competition for quite a long time and it doesn't seem like ATI is willing to give in or that the picture will change any time soon.

What has changed is that NVIDIA is no longer in the clear lead or alone at the top, however hard that would be for some people to digest (not you of course).

 

Why do you need a programmable 2d engine?
My understanding is that at least ATI cards can use their programmable 3d engine for doing flashy 2d effects too.

Its certainly going to be interesting to see some real benchmarks

 

Now a question concerning the 800 XT. We know that it's 16 pipe, meaning that theorically it could be 25% faster than the pro version, is this correct? To be able to accomplish that, it will need to have 25% more bandwich than the pro part, if both card are clock equally, right? Do you think the XT part will be clock blow, equally, higher than the pro part? If higher, do you think the memory will be available to help this beast, as DB hinted on memory as a possible reason for the late XT launch?

 

powervr !?

 

Doubt it, although it will do better re: bandwidth than NV40, which is... lacking at times. Not NVs fault though. Stupid Memory Manufacturers, huh?

 

This isn't directed at anyone in particular, but what boggles my mind is the apparent attitude from the fan-boys regarding this competitive situation. It's as if they don't realize this benefits us all, and would consciously choose to see their preferred IHV completely conquer the other and then raise their own prices.

 

I tend to feel that ATi fans are more guilty of this (just my impression), with their constant "must overthrow the evil NVIDIA overlords" attitude. Whilst they may not agree with the tactics NV have used, running them out of business isn't going to help anyone - until a real, viable alternative is presented, at least. And even then I'd argue against it.

 

Smart shader just manipulated the frame buffer and all the NV40 2d engine does is manipulate the frame buffer. This 2D engine sounds like marketing BS to me.

I'm sure you could make a plugin for photoshop that uses pixel shaders on this generation of hardware.

 

A not unreasonable point. However, if (that teeny word that means so much) it is true that NV40 boards are being given their workouts under NDA by reviewers across the world as I write, but these 16-pipe R420 boards are not out there yet, then in my view it is still FUD --aimed at reviewer mind-share and getting them to write more cautious reviews of the NV40, ala Anand's famous habit of reviewing ATI boards with a "but NV's new xxx is just around the corner and it is looking really good" ending.

 

Well, going back to an earlier post of mine:

If we assume that R420 and R300 are similar enough in architecture so that they require the same "bandwidth to fill rate" ratio to be effective, then, to give you some idea:

1) A 450 Mhz 12 Pipe X800 Pro would need 256 bit, 600 Mhz GDDR3.
2) A 450 Mhz 16 pipe X800 XT would need 256 bit, 800 Mhz GRRD3.

Dave B has not so subtly hinted that 800 Mhz GDDR-3 is a "pipedream" at this time...though it's not clear if there would be enough quantity for a limited launch of say, the XT a couple months down the line.

Now, it's very well possible that there are some more meory bandwidth optimizations that reduce the bandwidth to fill-rate ratio in the R420 vs. the R300, so that memory speeds wouldn't need to be quite so high to "satisfy" the 12 and 16 pipe configurations.

It's also possible that the parts will be simply be more bandwidth limited than the R3xx parts. No way to really know at this time...

 

Not "overthrow" but rather "force to change their tactics, preferrably for the better".

 

That so called "sound cancellation device" on the NV40 is no such thing.
Its a copper spool with platic on it, its for curcuit electronic noise reduction, not "sound cancellation device". Thought most here would know that since its been in use on most nVidia (and other graphic cards) for a long time now....

 

Yeah you are right Joe. But i was thinking that the 800 pro could be @ 400 MHz with 500MHz memory part, and the 800 XT same clock and 600MHz memory part. The issue being yield of course, but as HB says that Ati is willing to sacrifice yield on 800 XT...

 

Of course it helps everyone....the absolute worst thing that can happen would be for one company to be in complete control. It's by far best that bot companies - and more if possible, be as competitive as can be. However, it would be so nice for the consumer to be able to make a purchasing decision on a level playing field..... and while nVidia has the ability to build some wonderful and world beating products, the last thing they seem to want is a level playing field - EVEN if they hold the upper hand in product performance. As I have said before, many times, I will buy the best product for my money - no matter who makes it. BUT - that doesn't mean I have to love - or even like the company that manufactures it. The last month can be seen as nothing more - or less - than a respin of what went on in 2002 - when ATI brought out the original R300. - so far! I'll be real interested to actually see and own BOTH these products.

 

Well I think it's a bit more complex than that. If Nvidia had come out with a storming 16x1 design and stomped on an ATI Extreme 8x1, I think everyone would have thought "back to business as usual" and ATI would have been relegated to also-ran status. Nvidia's "optimisations" would be water under the bridge, and Nvidia could thank their PR based stalling technique to have saved them from a nasty period in their history they would rather like to forget.

By ATI bringing out another Nvidia beater, it consolidates ATI's position and brings the two companies into a position of longterm competition on equivalent footing, rather than R3x0 being seen as nothing but a blip to the continued natural dominance of Nvidia.

For this reason ATI not being soundly beaten this cycle is actually better for the long term benefit of the consumer as I think we'll see in the next couple of product cycles. We'll have two companies that are perceived by the general public as well as the enthusiasts as being strong and successful, rather than the single 900lb gorilla we had before the R3x0 watershed.

 

Really? Do you know what it is called? (I'm not questioning the accuracy of your input, just looking for a bit more info so I can google it. )