When will Nvidia finally get out of the NV30 rut?

Bjorn said:
jvd said:
which may equal nothing if p.s 3 on nv40 ends up slower or the same speed as p.s 2.0 on the r420. Not only that if used in other ways it may decrease performance , so who knows.

That is true and it'll probably take a while before we know how this pans out. Although the Far Cry SM2.0/3.0 patch should at least say something, whenever it's released.
well that said it was only speed and you wont see a diffrence .

Not only that but it apears the nv40 right now is running ps 1.1 shaders in that game . so who knows.
 
davepermen said:
to all: sm3.0 is not _REALLY_ ment to gain speed. it _can_ give some minor improvements. but its MAINLY only for having more shading-capability-freedom in terms of coding.

That's true and even Nvidia themselves admits this. But there are some features that could possibly gain a lot of performance. Geometry instancing f.e which i'm guessing will be supported in the Far Cry patch/addon.
 
Regarding FarCry, the "final" judgement will be made when the patch and DX9.c is out.
As far as the gamebenches etz have fallen too nV or ATI(most ATI with aniso on) i am not really paying that much attention to the the benches before we will see D3, Stalker and HL2 benches, that´s what matter´s too me atleast.
 
This feature allows a subset of the input registers to be initialized at a less frequent rate. This feature is supported by all devices that support vs_3_0.

[...]

Stream frequency is ignored for indexed primitives and if the current vertex shader set is less than version 3_0 (including fixed function).
 
Bjorn said:
jvd said:
which may equal nothing if p.s 3 on nv40 ends up slower or the same speed as p.s 2.0 on the r420. Not only that if used in other ways it may decrease performance , so who knows.

That is true and it'll probably take a while before we know how this pans out. Although the Far Cry SM2.0/3.0 patch should at least say something, whenever it's released.

Add that to the fact that this Far Cry patch will also feature 3Dc support, and things will be very interesting indeed.
 
overclocked said:
Regarding FarCry, the "final" judgement will be made when the patch and DX9.c is out.
As far as the gamebenches etz have fallen too nV or ATI(most ATI with aniso on) i am not really paying that much attention to the the benches before we will see D3, Stalker and HL2 benches, that´s what matter´s too me atleast.

mmm...right! lets make our final judgement AFTER most of us buy cards!

that way, all the people who bought a 6800 can whine about it, and we can laugh at them! :LOL:
 
N40 is nowhere near as bad as the NV30.

NV30 was a year late, hot, noisy, and was slower than the competitors card. Crippled on a 128-bit memory bus, and uber-expensive using DDR2 (arguably it decreased performance while costing more to make)

The NV40 beat the competition (at least for a couple weeks), its still hot but its not slow. It is at least a resonable alternative compared to X800.

Running around ~120 watts compared to the ATi around 60 watts is an area of concern though.
 
Bjorn said:
Nick Spolec said:
While the NV40 is leaps and strides above their previous offering, it appears that Nvidia is STILL suffering from their NV30 debacle, and their new chip is still carrying some of the traits/flaws introduced with the NV30 (IQ/Performance ratio, heat, and power).

I'm of the opinion that it's vastly different this time. There is a huge difference between the featureset and this time clearly to Nvidias advantage. I would have tend to agree if R420 had FP32, SM3.0 support though.
Its one thing to look at features and say *WOW*.

Its an entirely different matter to have an actual marketable product that will make your company money and not lose its butt. Which is the ledge Nvidia is currently messing around on.

They currently cant produce them in mass. They are not getting many good samples from the batches they do have. they are switching foundries. they need to re- tape out.

They have to Huge a power requirement for major OEMs to even touch them with a 10' pole. Where will the sales be generated? Where is the Margin for any profit at all??

Now, We have not seen their mid and low range parts yet. But will they also have all the features of the Nv40 this time around? Will they also have power and heat issues if they do? Will they have the performance needed to Compete?

From where I am sitting Things look pretty far from pretty, From a Business standpoint for Nvidia. Having SM3.0 or not on thier High end card wont help them here at all.
 
Kombatant said:
Bjorn said:
jvd said:
which may equal nothing if p.s 3 on nv40 ends up slower or the same speed as p.s 2.0 on the r420. Not only that if used in other ways it may decrease performance , so who knows.

That is true and it'll probably take a while before we know how this pans out. Although the Far Cry SM2.0/3.0 patch should at least say something, whenever it's released.

Add that to the fact that this Far Cry patch will also feature 3Dc support, and things will be very interesting indeed.

3dc support doesn't do much if don't have the compressed normal maps in advance to use it. Hence to support 3dc you would also need to download all compressed normal maps and i don't think crytek will do that
 
Hellbinder said:
Its an entirely different matter to have an actual marketable product that will make your company money and not lose its butt. Which is the ledge Nvidia is currently messing around on.

If Nvidia want's to sell cards at a loss just to gain some marketshare (or at least the developers support) then it's fine by me. I'm a consumer of graphics cards and not a stock holder in any of the companies.

Now, We have not seen their mid and low range parts yet. But will they also have all the features of the Nv40 this time around? Will they also have power and heat issues if they do? Will they have the performance needed to Compete?

I rather doubt that the mid end cards will have the same power and heat issues. And i very much doubt that they won't be SM3.0 capable. And from the reviews i've seen, the power and heat issues aren't exactly as bad as some people seem to think.

From where I am sitting Things look pretty far fro pretty From a Business standpoint for Nvidia. Having SM3.0 or not on hteir High end card wont help them here at all.

That's your opinion of course. I don't agree.
 
tEd said:
Kombatant said:
Bjorn said:
jvd said:
which may equal nothing if p.s 3 on nv40 ends up slower or the same speed as p.s 2.0 on the r420. Not only that if used in other ways it may decrease performance , so who knows.

That is true and it'll probably take a while before we know how this pans out. Although the Far Cry SM2.0/3.0 patch should at least say something, whenever it's released.

Add that to the fact that this Far Cry patch will also feature 3Dc support, and things will be very interesting indeed.

3dc support doesn't do much if don't have the compressed normal maps in advance to use it. Hence to support 3dc you would also need to download all compressed normal maps and i don't think crytek will do that

I think there's a pretty good chance it will, but that's just me.
 
ZenOps said:
NV30 was a year late, hot, noisy, and was slower than the competitors card. Crippled on a 128-bit memory bus, and uber-expensive using DDR2 (arguably it decreased performance while costing more to make)

The Nv40 is big, consumes more power, and is slower than the competitors
11137.jpg


The NV40 beat the competition (at least for a couple weeks), its still hot but its not slow. It is at least a resonable alternative compared to X800.

nv40 beat what competition? last generation?[/quote]
 
Bjorn said:
Hellbinder said:
Its an entirely different matter to have an actual marketable product that will make your company money and not lose its butt. Which is the ledge Nvidia is currently messing around on.

If Nvidia want's to sell cards at a loss just to gain some marketshare (or at least the developers support) then it's fine by me. I'm a consumer of graphics cards and not a stock holder in any of the companies.

Now, We have not seen their mid and low range parts yet. But will they also have all the features of the Nv40 this time around? Will they also have power and heat issues if they do? Will they have the performance needed to Compete?

I rather doubt that the mid end cards will have the same power and heat issues. And i very much doubt that they won't be SM3.0 capable. And from the reviews i've seen, the power and heat issues aren't exactly as bad as some people seem to think.

From where I am sitting Things look pretty far fro pretty From a Business standpoint for Nvidia. Having SM3.0 or not on hteir High end card wont help them here at all.

That's your opinion of course. I don't agree.
If it takes so mannnnnnyy transistors to pull of full SM3.0 support. How are you going to do that on a part with 1/2 the Trasistor count? Without totally sacrificing (for lack of a better term) *pipelines*.

Trust me The Power is a BIG deal. Not for people like you or me. But the real money for these cards is in the OEM segment. Which will simply not tollerate this kind of thing. Perhaps Alienware?

The Idea that you can keep sacrificing profit per board at an ever increasing rate, while at the same time clamping off your ability to generate OEM support because of your board needs... is well...

Just a wee tad Suicidal?
 
Theoretical question here: if that rumor on the IBM limited run of only 10,000 units is correct, and no tapeout of Nv40 at TMSC has occured yet, how long would it be before people can actually buy a 6800 at a store?

Even the Nv30 had a reported run of about 100,000 and those were kinda rare, so most 6800's will probably end up with devs or on ebay, right?
 
Hellbinder said:
The Idea that you can keep sacrificing profit per board at an ever increasing rate, while at the same time clamping off your ability to generate OEM support because of your board needs... is well...

Just a wee tad Suicidal?

They can of course not do that for anything but the 6800 Ultra.
 
thatdude90210 said:
Theoretical question here: if that rumor on the IBM limited run of only 10,000 units is correct, and no tapeout of Nv40 at TMSC has occured yet, how long would it be before people can actually buy a 6800 at a store?

Supposedly, the NV45 taped out at the same time as the NV40. Don't know if it's true or if it is, if it's at TSMC or IBM. But it definitely looks like it'll take some time before we can buy the 6800. .
 
Trust me The Power is a BIG deal. Not for people like you or me. But the real money for these cards is in the OEM segment. Which will simply not tollerate this kind of thing. Perhaps Alienware?

funny you should say that, since alienware had the nv40 as an option in their PC's the day of the launch. :p
 
Back
Top