Innovation: Rewarded. . .and not so much.

[sireric]

barycentric-sampling

explanation please

We support, when doing MSAA, either sampling in the center of the pixel (to generate color), or in the barycenter of the samples (centroid sampling).

 

[tEd]

barycentric-sampling

explanation please

We support, when doing MSAA, either sampling in the center of the pixel (to generate color), or in the barycenter of the samples (centroid sampling).

ah ok barycenter = centroid

thanks

 

[euan]

barycentric-sampling

explanation please

We support, when doing MSAA, either sampling in the center of the pixel (to generate color), or in the barycenter of the samples (centroid sampling).

ah ok barycenter = centroid

thanks

Named after Bary, who had the idea.

 

[tEd]

barycentric-sampling

explanation please

We support, when doing MSAA, either sampling in the center of the pixel (to generate color), or in the barycenter of the samples (centroid sampling).

ah ok barycenter = centroid

thanks

Named after Bary, who had the idea.

for real or just joking? :?

 

[Pete]

Bary always had to be the center of attn.

Anyway, have you people forgotten about Google already? Sheesh.

 

[3dcgi]

Absolutely "Trueform" was an ATI innovation. The concept itself was not a revolution, but the implementation ATI came up with (through the mind of Alex Vlachos) and implemented in their hardware made it entirely useable in a time where other concepts of the idea were not at all feasibly useful.

Npatches was ATI's innovation? Wow.

I believe Ati at least sponsored the research as an Ati employee coauthored a paper about N-patches.

 

[Ailuros]

barycentric-sampling

explanation please

We support, when doing MSAA, either sampling in the center of the pixel (to generate color), or in the barycenter of the samples (centroid sampling).

ah ok barycenter = centroid

thanks

Named after Bary, who had the idea.

for real or just joking? :?

Let a greek guess for you (besides the obvious joke some of you didn't get); bary in that case should stand for the equivalent greek word that means "weight".
Barycenter = ie put the weight into the center.

 

[WaltC]

*pokes sireric* F-Buffer please! *pokes sireric* F-Buffer please! *pokes sireric* F-Buffer please! *pokes sireric* F-Buffer please! (etc. . .)

Honestly, and I could surely be wrong, but I always perceived the R3x0 F-Buffer effort not so much as an innovation but as much more of a knee-jerk reaction to the PR numbers circulated relative to the maximum "shader-chain instruction-counts" advertised for the competitive nV gpus at the time. I'd rather see much less of a "numbers game" from the IHVs and instead a much stronger effort to educate their respective markets on why their products are designed as they are (ie, R3x0 really didn't need F-buffer regardless of the instruction specs nV published for nV3x, or so I certainly think, and in a sense the hurried F-buffer announcement from ATi probably provided the nV3x instruction-count advertising with at least a patina of legitimacy that it didn't deserve.)

In the case of F-Buffer, for instance, I think ATi wasted a lot of time and energy in coming up with it when it would have been far more efficient to simply explain why products like R3x0 and nV3x weren't going to be used for running "long" shader-instruction chains inside of 3d games--ever--regardless of the "maximum" shader-chain instruction counts advertised for those products as marketing bullets at their introductions. If such information is presented competently it need not appear self-serving or prejudicial or misleading at all, and I think is far more effective for ongoing and future product promotion than is the conduct of playing tit-for-tat feature-implementation "tag" with competitors just to feel less insecure in the marketing wars.

Some numbers, of course, are important to cite in specification advertising--like "pixel pipelines," for instance. But other numbers really aren't worth the emphasis they get, as I've said. The educational part of the picture really needs work from the IHV perspective, imo, especially where several of the popular 3d-related websites are concerned (apart from B3d, of course, which seems less affected by IHV marketing-speak than any site I know of covering the 3d industry.)

Almost daily it seems I read material presented on these other sites (sites I will graciously decline to name--at least here...) which is steeped in marketing lore dredged from the past that often bears little resemblance to currently relevant and worthwhile product information. The problem is that these sites just often don't know that their perspectives are either misleading or else invalid (or, they could very well know it--but that's another subject)--and of course it goes without saying that if the sites themselves are unaware of the skewed nature of the information they provide then of course many people who read those sites will walk away from them having actually learned far less of value than they surmise.

Of the two major 3d IHVs currently, on the whole for the last two years I think ATi has done a far better job of educating its 3d markets to its products than has nVidia (nVidia for most of that time having been content with avoiding factual and concise product info in favor of product misrepresentation and obfuscation, and of course all of the vacuous API protests most of us can recite by rote--which have hopefully ended with nV40.) I also think that ATi's overall self confidence in its clearly emerging market position is somewhat higher today than it was when F-buffer was announced, and so believe that ATi will be henceforth much slower to counter nV's "innovations" in a knee-jerk fashion with a look-alike marketing feature, at least until it can be determined whether a sound basis exists for such a product.

This leads me to at least the one "innovation" that recently has received a degree of publicity far out of proportion to its 3d market viability and unit sales to date--nV's SLI. Its one of many achilles heels being a requirement for dual-slotted PCI-e mboards, I would certainly hate to see ATi do another knee jerk and go hurly burly into a similar look-alike product announcement anytime soon (or, at least one requiring two PCIe slots.) I mentioned it earlier with respect to F-buffer, and I think it's just as true for SLI if not doubly so--but these kinds of look-alike product announcements from ATi actually serve to lend credence and credibility to what nV is doing--regardless of how ill-conceived that may be. So in hopes of countering nV's publicity about non-existent products with announcements of its own similar non-existent products, and in the process to level the advance-publicity playing field, it could be that what ATi is inadvertantly doing is simply to lend nV's product directions a degree of credence and credibility they otherwise might not have. IMO, of course...

 

[Entropy]

Bary always had to be the center of attn.

Anyway, have you people forgotten about Google already? Sheesh.

Möbius, 1827.
Anyone knows about a good biography of Möbius? That's just one time too many he turns up. Just from the different areas where I've seen his name pop up, the man must have had an awesome mind.

 

[LeGreg]

Do you know what is the center of gravity ?

 

[rwolf]

Didn't ATI help with the design of GDDR3 memory? That innovation gave us performance and cost reduction.