R520 Infomania

The new level isn't using floating point blending, it's using 16 bit integer buffers.

 

I wouldn't bet on Valve using an implementation that's not supported by ATI's products. If they do support both methods and allow the use of floating point blending on 6xxx I would be highly impressed.

 

mboeller, 4 or 5 I hope.

Subtlesnake, this is not the full truth

 

This is what they've said:

Is the HDR rendering utilising float buffers? (i.e. will this work on NV hardware under the DX9 path if they haven't enabled float target support.)

[Brian Jacobson] Yes, although we're working with NVIDIA to get a non-float buffer version working.

[Gary McTaggart] Actually, with ATI, we use 16 bit integer buffers. We don't use a float buffer. NVIDIA is adding support for buffers that are lower than 16-bit integer.

 

How old is this?

 

Far too old to be using it to state what the new Lost Coast level is or isn't doing technology wise.

 

Well if Lost Coast is using FP blending and assuming Valve has a r520 on hand, Did anybody notice any AA being applied in the demo? It would be nice if the r520 could do AA and FP blending. 8)

 

And leave R3xx->R4xx users out in the cold? The horror!!

 

Demirug, since you're under NDA, was my guess with the Ati Level and R520 "launch" correct, or you don't know anything about it.

(I know "I can't give an answer to that" is yes...)

 

I don't know anything about the release date for this level or the R520 launch date. The NDA I have is not affected by this events. But it looks like that I am now free to talk about they stuff I am know but I will double check this before I get in some trouble.

 

It looks like w'll have to wait a verly long long time time to see the R520 (or G70) and I think it will a autum availability on earliest..

 

Say it ain't so.

 

OK, here we go.

This informations are from Gary McTaggart (senior software engineer at Valve) and part of a Q&A for the german PC Games Hardware magazin.

There will be two new modes that kann be used with the "Lost Coast" Level: Best Quality and Lower Quality.

For Best Quality you will need a chip with:
- FP16 filtered textures
- FP16 targets with alpha blend support.
- SM 3.0

For Lower Quality:
- FX16 filtered textures
- SM 2.0

In both cases the lighting informations are stored in textures with higher range (FX16 for lower; FP16 for best) Best will use FP16 targets, too.


IMHO I am still not sure if I want to call the "Lower Quality" Mode a HDR Mode. It looks more like a common bloom postprocessing with a FX16 lightmap.

 

The Bloom effect is allready in source.
I think "Gary's Mod" exposes it, it ads a nice bloom effect, a la Far Cry on my x800.
Up to now, I don't think we were able to turn it on or off within any source game...
Also, the demo's of source that were released last year and the year before didn't show it, so it was probably updated in the last months, I guess since March.

Now, SM3.0 for "high quality"...
Really, unless Valve is popping it up Ati's behind after accepting the 6 million, my bet is that they WILL use it for the r520 showcase.
And I wouldn't be surprised if they used something that would bring the nv40's to their knees...

 

Intentionally out of deference to ATI or out of necessity for the effect(s) they're trying to accomplish?

 

Number 1.

A *LOT* of people over at ati want to have front page impact like the nv40 had with Doom3.
Valve has both cards, they can look where it hurt and later (as in 2-3 weeks) release a patch that fixes aforementioned problem.
They can say "w00ps, sorry" and no one will care a month later, but the impact this line will have "r520 humiliates 680Ultra SLI" is bigger, a LOT bigger than the 6 million ati pumped into Valve, agree?

After all, it's still a "Get in the game" project and "you can expect the best visual results and speed on ati hardware"

 

You do realize though that in essence the GITG and TWIMTBP programs are not at all about "humiliating" the competition and especially any possible developing funds that might float into an ISVs or game publishers pockets. If that should be the case NVIDIA managed to "pay off" for quite a bit of "self-humiliation" with their TWIMTBP program in several spots during the NV3x-era.

I could accept that an ISV could result on inferior performance on one family of accelerators because:

a) said hardware (or even driver) has inevitable weaknesses or is lacklustering in features.
b) out of ignorance (because there's no IHV developer initiative present to inform developers during the development) or even just a simple mistake.


I have severe doubts that an ISV would result to such ventures on purpose; not only is it unethical, it would be most likely exposed in no time and it would be the worst publicity slap for an ISV and in extension the IHV indirectly.

Some of you folks are taking those hypothetical "graphic wars" way too seriously.

 

Half Life 2 was, but that doesn't necessarily follow that any subsequent updates actually have to be.

 

That's true, but it can still be heavily optimized for certain peace of HW (meaning R520 obviously).

Or not. We'll see

 

TWIMTBP is still a laughable logo on most products.

Most of the time, it is the vendor promoting specific routines or functions that work best on their hardware; volumetric-shadows being the best example as it kills 9700pro's in X2 the threat, but has a mid-range nv card, the fx5700 for example "shine like a star."
More recent saw Doom3 display the same behaviour.
Right now, I guess Rome: Total War is the only other TWIMTBP game that exhibits "strange" behaviour on ati cards, as in 20 to 30 fps difference is a lot in a dx8 game.

There are reasons why Ati likes to show off Colin McRea / Race Driver (toca/dtm) games at prodcut launches and demonstrations, because it's simply optimized for their hardware, the first r420 benchmarks of that game showed astonnishing performances with that card. (http://graphics.tomshardware.com/graphic/20041004/vga_charts-12.html)
I mean, in which every day game would a card lead with twice the number of frames?

In the world of sales, ethics comes in 8th, after carbon copy paper..