Are you ready ? (Nv 30 256 bit bus ?)

[Bigus Dickus]

Another brute-force 256-bit IMR would be, well, predictable.

As predictable as the Parhelia's performance, right?

 

[Vince]

If the NV30 implimentation of CineFx does indeed have a deferred or otherwise exotic architecture, I find this whole 128-256bit kind of ridiculous. Back in the day, 3dfx/GP anticipated a 10X reduction is bandwith needs by using a region based deferred rendering scheme - this is my only starting point. Even if nVidia can only achieve 40% of that, using the same underlying/baseline architecture, they'd achieve NOTHING by moving to a 256-bit bus except for added costs and having their supporters toting around higher numbers/nomenclature.

Now, that being said, you can increase the output of said underlying architecture to meet the added bandwith effeciency and have a design that would utilize a 256-bit bus and be quite a bit faster than the competition.

I realise I'm not intune and don'r hear as much as many here (heh), but I've seen alot of indications that nVidia is taking this road.

 

[Testiculus Giganticus]

Efficiency will be the buzzword in coming weeks

 

[Testiculus Giganticus]

edited

 

[alexsok]

Efficiency will be the buzzword in coming weeks

Correct. After all, if they can suprass R300 in terms of perfomance (with AA, ANISO of course) while still utilizing a 128-bit memory bus (all rumours folks ), then I see no point in using 256-bit memory bus (the chips would be cheaper to make as well, which means further potential for lower prices over time...).

The coming weeks will be filled with many useless buzzwords coming directly from the tons of marketing papers NVIDIA will release and the countless amount of tech demos they have prepared...

The unique aspect of this board is that people in here usually look beyond the "buzzwords", discussing the potential of the underlying architecture, it's cons & prons, it's limitations and so on... that's what I'm waiting to hear about from the folks on this board, so no worries, these marketing papers and "buzzwords" are not directed at us!

 

[Simon F]

NVidia is already talking smack about ATI's 24-bit FP precision vs NVidia's 32-bit FP and about DX9 limits. Here come the endless flame wars about what's "enough" and "not enough" I personally think that while 32-bit FP is preferred, 24-bit is so much better than what we used to have that it doesn't really matter for games all that much.

There's a possibility, however, that Nvidia will be agressively pushing their 16bit floats rather than the 32bit because of the relative performance advantage. The 'better quality' argument might not be so valid <shrug>.

 

[DemoCoder]

I bet NVidia's spin on this is "quality where you need it" or "only pay a performance penalty where you need to"

 

[Randell]

At least its a choice between speed or quality I suppose, and better quality than under 'performance' than went before. They should promote the NV30 flexibility rather than say the competition isnt adequate, bearing in mind prior comments about how much advanced R300 is than anything else currently available in that department.

 

[Kristof]

As I have said before it will lead to a mess of benchmarks... NV30 either supports 16 bit per component or 32 bit per component, one is low and one is high... some will say one is too low and one is too high. R9700 sits in the middle with 22 or 24 bits (Can't remember). So what do you compare with what ? No matter how you twist and turn it will be apples and oranges...

K-

 

[Laa-Yosh]

Correct. After all, if they can suprass R300 in terms of perfomance (with AA, ANISO of course) while still utilizing a 128-bit memory bus (all rumours folks ), then I see no point in using 256-bit memory bus (the chips would be cheaper to make as well, which means further potential for lower prices over time...).

Why not? It can instantly double your theoretical performance. It's almost like asking "why do we need a V8-engine, if the Honda S2000 can do 220Hp with a 4-cylinder one?"

 

[Nagorak]

Correct. After all, if they can suprass R300 in terms of perfomance (with AA, ANISO of course) while still utilizing a 128-bit memory bus (all rumours folks ), then I see no point in using 256-bit memory bus (the chips would be cheaper to make as well, which means further potential for lower prices over time...).

Why not? It can instantly double your theoretical performance. It's almost like asking "why do we need a V8-engine, if the Honda S2000 can do 220Hp with a 4-cylinder one?"

Price.

 

[demalion]

Correct. After all, if they can suprass R300 in terms of perfomance (with AA, ANISO of course) while still utilizing a 128-bit memory bus (all rumours folks ), then I see no point in using 256-bit memory bus (the chips would be cheaper to make as well, which means further potential for lower prices over time...).

Why not? It can instantly double your theoretical performance. It's almost like asking "why do we need a V8-engine, if the Honda S2000 can do 220Hp with a 4-cylinder one?"

Well, depending on the way the rest of the vehicle is engineered, a 4 cylinder might conceivably be faster than a v8. If it costs you less to make a 4 cylinder and you can engineer the rest of the vehicle to offer the features the consumer wants while doing so, that's a valid reason "why not".

Nothing new in that observation, of course, but I felt your analogy warranted this response.

On the one hand, it seems pretty feasible that it would be 256-bit, on the other, their wordings in some places and many rumors seem to indicate (to me) 128-bit. We'll see soon enough I guess.

 

[Laa-Yosh]

The Radeon 9700 doesn't seem to cost that much for me; at least it's not that much more expensive than the Ti 4600 was half a year ago. And don't forget we're talking about highend cards here

 

[Randell]

The 9700Pro is now cheaper than the 4600 was when the 9700Pro first launched in the UK.

 

[Laa-Yosh]

Well, depending on the way the rest of the vehicle is engineered, a 4 cylinder might conceivably be faster than a v8. If it costs you less to make a 4 cylinder and you can engineer the rest of the vehicle to offer the features the consumer wants while doing

Er, my point was - to continue with the analogy - that if the additional cost of making your engine a V8 instead of 4 cylinders is reasonably low, than why shouldn't you do it? And we can see by the example of the Radeon 9700 that a 256-bit bus does not make a card considerably more expensive; however producing a 0.13 micron chip with seriously low yields could...

 

[dksuiko]

Correct. After all, if they can suprass R300 in terms of perfomance (with AA, ANISO of course) while still utilizing a 128-bit memory bus (all rumours folks ), then I see no point in using 256-bit memory bus (the chips would be cheaper to make as well, which means further potential for lower prices over time...).

Why not? It can instantly double your theoretical performance. It's almost like asking "why do we need a V8-engine, if the Honda S2000 can do 220Hp with a 4-cylinder one?"

Torque.

 

[demalion]

Well, depending on the way the rest of the vehicle is engineered, a 4 cylinder might conceivably be faster than a v8. If it costs you less to make a 4 cylinder and you can engineer the rest of the vehicle to offer the features the consumer wants while doing

Er, my point was - to continue with the analogy - that if the additional cost of making your engine a V8 instead of 4 cylinders is reasonably low, than why shouldn't you do it?

Notice my bolded "If" and "and", why'd you simply replace them with an if that was the opposite? You asked "why not", and the statement answered that...

And we can see by the example of the Radeon 9700 that a 256-bit bus does not make a card considerably more expensive; however producing a 0.13 micron chip with seriously low yields could...

Are we talking about expensive to manufacture or expensive to the consumer? We really don't have any idea (atleast as presented in this thread) how expensive it is for ATi to make (though of course ATi or a board partner could clear that up), and we further have no idea how expensive 256-bit would be for nVidia to implement for a different graphics chip. Does nVidia depend on a larger profit margin for their business model? A lesser one?

 

[jb]

Efficiency will be the buzzword in coming weeks

It was the buzzword when the K2 first launched and kept up with the GF2 given its TNT2-Ultra "like" specs

 

[no_way]

Why not? It can instantly double your theoretical performance. It's almost like asking "why do we need a V8-engine, if the Honda S2000 can do 220Hp with a 4-cylinder one?"

Honda doesnt sound like 'Vette

 

[Nappe1]

Correct. After all, if they can suprass R300 in terms of perfomance (with AA, ANISO of course) while still utilizing a 128-bit memory bus (all rumours folks ), then I see no point in using 256-bit memory bus (the chips would be cheaper to make as well, which means further potential for lower prices over time...).

Why not? It can instantly double your theoretical performance. It's almost like asking "why do we need a V8-engine, if the Honda S2000 can do 220Hp with a 4-cylinder one?"

Torque.

well, actually... High tuned Japanese Inline engines (4 as well as 6 cylinders) give pretty awesome torque as well as the power. But the key thing is that those engines are a bit high rpm ones: you need get the rpm pretty high to get all out of it. (without even mentioning Mazdas Wankels which love everything over 6000 ) With more cylinders and with V-construction, you get those torques in much lower rpm's... Everyone who has tested Range Rover outside of the something that could be called road, knows what I am taliking about. Another thing is that V-engine does not make car "shake" like the old Inline engines did without overhead camshafts did. (nowadays there isn't really much difference as long as your Inline has at least two overhead camshafts.)

and IF we are talking about cars with Inline engine and extremely good power/engine size ratio, you must Remember the very first shopping rocket: Mini Cooper S. with properly tuned that bit under 1300ccm engine gave more than 125 hp and when the car weights less than 700 kg's it was a damn fast on it's time. even the legendary 1966 Mustang GT does not have so good Power/weight ratio as Cooper S had on the same years.