PDA

View Full Version : NV50 specs???


Pages : 1 [2]

Greg
02-19-05, 10:18 PM
Just a few comments to clarify things Nutty has already mentioned...

A single game frame may contain a few hundred thousand vertices, but a few million pixels may be drawn, with each pixel possibly made up of multiple fragments. So the pixel shader usually does more work than the vertex shader.

Since VS and PS 2.0, the inputs and outputs of both shaders are floating point vectors. The instructions are also very similar. In the past, the vertex shader needed to do more complex operations since it did the transform and lighting calcs. Now, many of those calcs are done per pixel.

PS 1.0 (and possibly up to 1.4) were not implemented in hardware as instructions interpreted and executed by a GPU, they were implemented as a configuration state block, just like the old fixed function. That is why on the XBox it is possible to fill out the configuration block instead of writing a PS program. It also explains the extreme limitations of the PS 1.0 set, such as instruction order, sampling and calculation limits.

Now having said that, you can see that both the VS and PS are becoming more like standard DSP processors, like a SSE unit on a pentium 4 or such, and recently, with branching operations. Unless there is a specific reason (such as asyncronous processing of logical VS and PS operations), there is little reason not to merge the processes into shared units. The actualy hardware implementation is really up to the engineers in terms of cost and real world benefit. You remember from the whole NV30 saga that having fancy hardware doesn't mean you can sell it, if it doesn't run current games real fast (at least compared to your competition).

From a coding point of view, although the VS and PS are still logically separate operations, they belong to the same material, or surface effect. I notice one of the projects at work has almost one pixel shader for every vertex shader, showing that they could almost be stored or at least considered together. There is also now a stronger dependance between VS and PS due to arbitrary data passing between them, rather than fixed sizes and semantics.

jAkUp
02-19-05, 10:27 PM
Well if this is true, it makes my SLI setup look like a Super NES. LOL

Razor1
02-19-05, 10:30 PM
What options ? I dont see many differences between the 2 at the moment.


I think thats more to do with the fact that pixel shaders are more heavily tied into how an object's appearance is created, so you need to cater for more variations with pixel shaders, whereas you can feed them all from a generic vertex shader.


Because fragment shaders were lagging behind vertex shaders in the terms of programmable flexibility.

Theres also the point that vertex shaders worked on Floating point, whereas fragment shaders were fixed point. So it made sense to create custom hardware for them, to exploit the savings in fixed point. But fragment shaders are floating point now too, so there litterally is hardly any difference in what they do, which...


.. allows the hardware engineers to concentrate on a single unifed shader engine, instead of creating 2 custom ones, where custom ones aren't needed anymore.

here are some diffferences from PS vs VS

PS has these extra features just for SM 3.0,

Shader antialiasing, Back face registry, Fog and specular, texture coordinate count, Inerpolated color format all these things the vertex shader has nothing to do with.

The fragment shader was never lagging behind (well at least after it was created), without a fragment shader we couldn't do any per pixel attenuation.


BTW Greg its still not true pre pixel calculation, everything is derieved from the vertex normals.


I see where your coming from, its just different point of views Nutty :)

jAkUp
02-19-05, 10:30 PM
So... here is where we are at this far in the rumor bin... :)

Rumored ATI & nVIDIA Next Generation Card Specs

ATi
"R520"
24 "Pipelines"
32 Texture Units
96 Arithmetic Logic Units (ALU)
192 Shader Operations per Cycle
700MHz Core
134.4 Billion Shader Operations per Second (at 700MHz)
256-bit 512MB 1.8GHz GDDR3 Memory
57.6 GB/sec Bandwidth (at 1.8GHz)
300-350 Million Transistors
90nm Manufacturing
Shader Model 3.0
ATI HyperMemory
ATI Multi Rendering Technology (AMR)
Launch: Q2 2005
Performance: Over 3x Radeon X800 XT !!! (for single R520)
16x stochastic FSAA
FP32 blending, texturing
Programmable Primitive Processor/Tesselator

nVIDIA
"NV50"
375-400 million of transistor on 90-nm process
550-600 MHz core
1600 MHz 256Mb (Support for upto 512Mb) GDDR3
Graphics Core 256-bit
Memory Interface 256-bit
32 (24 real) pixels pipelines (on the Ultra model)
32 Vertex Shader Engines
51.2 GB/sec Bandwidth (GDDR 3)
Fillrate 12.8 billion texetls /sec.
Textures per Pixel 32 (maximum in a single redering pass)
Nvidia's LMA - IV
DirectX 9.0d UMI-23
Release date Q3 2005

MUYA
02-19-05, 10:35 PM
U forgot NV48s ;)

SH64
02-20-05, 03:31 PM
U forgot NV48s ;)

& umm .. NV47s :p

MUYA
02-20-05, 11:29 PM
I do now think Nv50 might have been a differnt design philosophy to what we are used and got canned as they didn't see the benefits of this? just a shot in the dark. NV have a few graphics projects running at the same time and probably ended up staying loyal to the architectures that have worked for them. Unified model maybe a little to early to move to?

I dunno.

Razor1
02-20-05, 11:53 PM
Mostly your right Muya, since they will need WGF to use a unified system, and its not fully ready yet and possible delayed till next year, there really is no use for a unified shader system right now. What I don't get is what will happen to Ogl when WFG comes out. Nvidia hopefully will make sure Ogl will stay competitive.

ToxicTaZ
02-24-05, 01:09 PM
U forgot NV48s ;)


YES! Very true...NV-48 is still real, they just put it again in the next Forceware Drivers (75xx) and coming this march :) and as for NV-47 well :rolleyes: show me the forceware Driver that has NV-47 suport in it?

Forceware 65xx ?
Forceware 70xx ?
Forceware 75xx ?

NV-48 code is in all of those drivers! :angel:

As for NV-50 Specs in this thread they are fake :thumbdwn:
NV-50 would look more like this?

- GeForce 7800
- 0.09-micron process made @ TSMC or IBM
- 32 pixel pipelines
- 600MHz core speed
- 256-bit 1000MHz GDDR3 1ns (SAMSUNG) 64GBps, Total of 2000MHz of video RAM :cool:
- 512MB and 256MB Models of Video Ram
- 12 Hardware Vertex Shaders
- 325 million Transistors
- UltraShadow III Technology
- Intellisample 4.0 HCT Technology
- CineFX 4.0 Engine & DirectX 10a PS/VS 4.0 (shader models 4) and OpenGL 2.0 Support
- Dual 600MHz RAMDACs display resolutions up to 2048x1536 @120Hz
- Dual DVI for HDTV in & out
- DVD and HDTV-ready MPEG-4 decoding up to 1920x1080 ( 1080p/i ) resolutions.
- PCI Express and AGP 8x Support
- SLI

All this could be out by years end X-mas 2005?
See don't this look alot better? :D

ToxicTaZ
02-24-05, 01:27 PM
Oh ya I better put NV-48 specs here too.

- GeForce 6900 Ultra AGP, GeForce 6900 GT PCI-E, (NV-48)
- 0.11-micron process made @ TSMC
- 24 pixel pipelines
- 500MHz core speed
- Under clocked RAM @ 750MHz from 256-bit 800MHz GDDR 3 Ram (SAMSUNG) K4J55323QF
- 512MB and 256MB Models of Video Ram
- 8 Hardware Vertex Shaders
- 260 million Transistors
- DirectX 9c PS/VS 3.0 (shader models 3) and OpenGL 1.5 Support
- Dual DVI for HDTV in & out
- SLI

Remi
03-01-05, 10:00 AM
Unified model maybe a little to early to move to? I dunno.Same feeling here... By the way, it's strange to see that ATI is going that way with the R520 (which looks like a challenge - I will be curious to see how they'll handle PPP expansion and rasterizer expansion for instance), when it seems to me that nVidia has a better suited architecture to handle that. I surely will be watching this with interest!

Mostly your right Muya, since they will need WGF to use a unified system...Hum... Interesting. But you lost me there. What would WGF bring to the unification table?

Razor1
03-01-05, 02:43 PM
Hum... Interesting. But you lost me there. What would WGF bring to the unification table?


WGF will bring the unified shader system to the table without WGF there is no unified shader system :) well there still could be but we will need an api to access it.

Remi
03-01-05, 04:35 PM
Oh-oh!... Looks like it's time to refresh my readings of Longhorn's SDK... :D Thanks!