Batman AA, the battle continues...

http://www.hexus.net/content/item.php?item=20991

The last paragraph sums it up but I'm missing something:

"The entire episode can be viewed one of two ways. On the one hand, NVIDIA's spent money and resources in enabling AA for an engine - Unreal Engine 3.5 - that doesn't natively support it, so why should AMD be given the code, gratis, to enable AA on its hardware? The counterpoint, as AMD may well argue, is that the company is (at a later stage than NVIDIA) prepared to put in the legwork but is apparently claiming it isn't being given an opportunity of doing so."

I didn't see anything, in the article at least, that would prevent anyone else from adding AA functionality with their own independent code optimizations.

CF AA = Custom Filtering Edge-detect filter (ATI)
AD AA = Adaptive Antialiasing at the Supersampling Quality level (ATI)
CS AA = Coverage Sampled Antialiasing (NVIDIA)
TR SSAA = Transparency Supersampling AA (NVIDIA)

Not all AA algorithms are created equal, with some giving better quality then others. NVIDIA's is supported, ATI's is not.

Considering DirectX comes from MS I'd like to see AMD carry out that threat, it could be interesting.

Ok so most of us know that each teams AA works differently, now i freely admit that I'm not very knowledgeable when it comes to driver level stuff so and please be gentle with me, what good does getting the vendor IP filter taken off of Nvidia's optimised AA coding do ATI ? don't they need to supply Eidos with their own AA optimised code for ATI AA to work properly ?

Mactronix

1 DX is owned by microsoft, microsoft showing a bais would be fun to see how that pans out
2 AMD is AMD they just lock Nvidia out of IGP department on AMD boards just like Intel has done to Nvidia for the time being. Nvidia has no friends it's why you don't see too many igp from nvidia on mobo's ION is basically their last go at it in that department until intel pushes them out of that and developed decent graphics for their atom to run with.

And why doesn't AMD just supply edios with their AA code to insert into the game?

AMD do not and never have locked Nvidia out of anything. It was Nvidia who abandoned AMD in favour of the intel platform, and it served them right to be treated like dirt by intel so now they have no chipset business worth calling that.

This is getting interesting

http://forums.hexus.net/hexus-net/ [...] ost1808312

That has what to do with the topic?

Well, nVidia has said, its selling its SW, talks CUDA, talks CUDA cores, talks Fermi, talks gpgpu, talks about how DX10.1 and DX11 arent that important.
So, I guess they want money for anything they do, including something as simple as SLI.
Not that theyre money grabbing here. Coming in with 600$+ cards etc.
Keeping their overpriced, underperforming G200s priced high etc etc
Its not about the money, remember that, its for you, the consumer, all these things

This was already said by ATI weeks ago, as they attempted to work with the devs at that time.
I guess Huddy just raised it up a notch is all.
Its really a shame, and for a company that claims their cards are sooooo much better by using their proprietary IP and charging for it, and cutting off usage of other, again, etc etc, theyre losing their own base.
In one of the comments/emails, Lars says:
With your comment regarding locking DX11, do you try to indicate that AMD invented DX11 and could have been an AMD-only feature?? DirectX 11 is a new version of DirectX, that will be fully supported by Fermi, as we announced at GTC. It seems that AMD tries to create the perception that DX11 is a AMD only feature. It is not.



My problem with that is, if its ATIs IP making these things work in DX11, then, according to him, why not?
AA is NOT owned by nVidia either, nor is it proprietary.
Eidos, pull your head out, and next time, say, thanks but no thanks

Wrong. Where has ATI been offered as much, using the same API?
Wrong, nVidia "thought" they could charge that much for their G200s, and rebates prove them wrong.
And, as DX11 games are devved, and its usage is propelled forwards by ATI devrel helping those devs, why should nVidia be able to use any of that?
It wouldnt have been there unless they work on it.
Sorry Jack, this is like being the kid with the only baseball, and hes decided to take it home.
Some things need to be overlooked, just for compliancy in a PC gaming world

So it seems it was Nvidia code to block out ATI, and not Eidos? Since AA is standard both companies abide, its no wonder when you remove block, AA works just fine on Radeons too. Thats why funny counterpoint is AMD should block out Geforce owners from DX11 and any features they help developers with I wonder how those who defend Nvidia would react then, I can bet it wouldnt be "they have a right to do it" or "fair enough"

Not really. AA is a really basic math function, all things considered. Runs slow as molasses though, which is why both ATI and NVIDIA came up with alternative methods to achieve the same exact effect (SuperSampling, Multisampling, Edge Detect, etc). Tell me, why should ATI be allowed to hijack NVIDIA's own functionallity?

So the primary issue at hand is WHY ATI's implementation didn't make it in game. Could be Edios didn't want to have to test/support ATI hardware; maybe NVIDIA locked them out as part of the deal to work with NVIDIA. We'll probably never know.

If you had a company and your legal department advised you not to do something would you ? Thats what i mean by fair enough but yea you take on it works too and i would say fair enough either way.

Mactronix

Cheers mate, The other thing notable in it's absence is the proof that ATi did proved code and or engineers and that they were not used, as they are quite happy to show us emails that are being sent back and forth where is the copy of the shipping note, plane tickets or whatever receipt you would have after sending code or engineers to the dev's? I've not seen anything like that yet, have you?

Exactly there is all sorts of speculation and inferance but no actual proof that ATI even lifted a finger to get this sorted. Im not an anyones side here just calling it as i see it.
Just for those who like to brand people fanboys for what they post i would like to point out that if anything i personally favour ATI as agraphics solution and am just calling it like it reads.

Mactronix

I really don't know the answer to this, nor do I really care since the game doesn't interest me, but in the above email:

"I believe this technique is very closely related to a technique which we've seen NVIDIA recommend before now - so actually it may well fit very well with the code that they've given you... "

Note how he says he "believes 'this' technique...may well fit very well [with NVidia Code]". That tells me ATI gave them something, otherwise why would he be talking about something working well with NVidia code, if none of his code is in the game? After that though, maybe it wasn't deemed acceptable, or some darker reason, or who cares.

So, DX11 specific is OK?
You know, nVidia does sabe alot of monies at the hands of ATI.
I think this may work out after all.
Make nVidia come up with its own DX11 implementation in every game, no crossover, and maybe then , TWIMTBP wont have all the time and money to mess with other things

Fine by me, you see the way I'm reading this bit in the email:

suggests to me that this has happened before with other games and that rather than use their own code ATi have waited for the game to be released, reviewed and benched and then have ripped the code apart found the vendor tags and written a workaround into their next set of drivers thus getting the performance boost that is always being touted as an ATi strong point (it will only get better as the drivers mature, sound familiar?), simply because they have always been using the Nv specific code rather than supplying their own. The burden of proof here is at ATi's door, where is the code that they provided? what proof do they have that it was offered and turned down or sent and not used? and if the cards are so different because ATi cards are so much better with DX10.1 and now DX11 capabilities and because of their internal architecture which is so much more refined than Nv's big power hungry rubbish, why would they want to use Nv code anyway?

Yeah mm, I agree it is fairly unclear. I could read it both ways, but I feel putting so much thought into something that the guy probably spent 5 seconds writing is a case of reading too far into it.

But, just to explain what I meant a little further:

"so actually it [what it? This is where I am assuming he is referring to his code. Obviously though if that is true, somewhere that code lost its way, as it isn't in the game currently (though who knows, could be in testing] may well fit very well with the code that they've [and like you said, this is referring to NV code] given you..."

So, I guess the question is, what is the it? I could see it being either ATI AA implementation or the ATI guy saying that NV code would "fit" well with the way the ATI cards do AA. If the latter is the case, then I am fine with ATI being barred until they 'contribute' in some way. However, if it is referring to code and Eidos decided not to use it (assuming it worked) then shame on them. Oh well, as long as AC2 and SH5 aren't AA-less, I won't lose any sleep. Interesting emails, regardless of what they mean.

This was one of the main reasons I picked it up for PS3 instead of PC (that and trophies, hah, I'm such a T-whore) as I use an ATI card in my gaming rig, and really wasn't happy at all with not having that option. If I have to play with no AA, might as well do it on a console, where there are other rewards, heh.

Like I said earlier, this is all for us, consumer.
They deliver, for a cost.
At this point, PC gaming needs more team than individuals, and since theres been rumors of nVidias ignoring the gaming area, this too plays right into it.

This just in!!!!!!!!!!!!
CUDA buys nVidia!!!!!!!!!!!!!!!!!!!

2 things about this I find troubling, besides the obvious.
One is, having the exclusive to me would be great for MSI, so ditching it doesnt make sense.
The other is Intels involvement with LH, unless Intel was using this as alternatives to SLI if they couldnt settle it out with nVidia