I haven't made a decision on this whole fiasco yet but I have to give ATI a lot of credit for that. At the very least it shows they care about their customers and want to alleviate their fears and concerns. I don't recall that other IHV doing anything like this when they were accused of cheating and over-optimizing...Wunderchu said:
ATI Filtering Optimisation - Genius or Disingenuous
- Thread starter pocketmoon_
- Start date
I have to say I kinda want to vote for both 
From the shots I've seen which are static screenshots and hence almost entirely inappropriate for judging IQ IMO this algorithm seems to be doing a remarkably good job.
In fact it's so good, and clearly so very very clever that it's the sort of thing I personally would expect the PR department of any major GPU manufacturer in a highly competitive marketplace to give some silly-arse name and shout from the rooftops about it's ability to cure AIDS, cancer, world hunger and poverty whilst delivering an enhanced user experience for the discerning gamer. Or something.
If the gamer is getting better IQ and better performance, great. The thrust of the benchmarks and interminable reviews of these two latest cards has been to try to judge which card gives the best performance at a given level of IQ. The review guidelines put out by ATI are clearly disingenuous in this respect -- maybe a conspiracy, maybe a cock-up, but it certainly invalidates any reviews that have assumed that setting both cards to do full, unadulterated tri-linear will give similar IQ, then presenting the frame-rates and declaring a winner.
A+ for technicals, D- for PR.
From the shots I've seen which are static screenshots and hence almost entirely inappropriate for judging IQ IMO this algorithm seems to be doing a remarkably good job.
In fact it's so good, and clearly so very very clever that it's the sort of thing I personally would expect the PR department of any major GPU manufacturer in a highly competitive marketplace to give some silly-arse name and shout from the rooftops about it's ability to cure AIDS, cancer, world hunger and poverty whilst delivering an enhanced user experience for the discerning gamer. Or something.
If the gamer is getting better IQ and better performance, great. The thrust of the benchmarks and interminable reviews of these two latest cards has been to try to judge which card gives the best performance at a given level of IQ. The review guidelines put out by ATI are clearly disingenuous in this respect -- maybe a conspiracy, maybe a cock-up, but it certainly invalidates any reviews that have assumed that setting both cards to do full, unadulterated tri-linear will give similar IQ, then presenting the frame-rates and declaring a winner.
A+ for technicals, D- for PR.
I guess my feeling on it is that I like it, but I don't like the fact I cannot choose not to use it. It would of been better had ATI had a little checkbox in the control panel to enable this type of thing. I think you would be silly not to enable it, because the screenshots do look good, but I still like having the option.
The fact they kept it almost secret makes me wonder??
I also am amazed at all the Nividiots coming out of the wood work throwing the "cheat" word around. Cheat is a bit harsh for this in my opinion? Its not like the added clipping planes. This is a straight up optimization that I feel they should give us the opportunity to use if we want, not just force it.
The fact they kept it almost secret makes me wonder??
I also am amazed at all the Nividiots coming out of the wood work throwing the "cheat" word around. Cheat is a bit harsh for this in my opinion? Its not like the added clipping planes. This is a straight up optimization that I feel they should give us the opportunity to use if we want, not just force it.
christoph said:
Oh I do. Its was pretty easy for us all to spot the brilinaer in its first form. We have 37 pages and many man hours spent by people looking for IQ issues on this. The fact that we STILL dont have any "proof" of lower IQ leads be to believe that the differences are very small. Again could be wrong but so far so good
Deepend on wheather or not you want to get into a debate or not
But the simple answer is yes. Hardware limitations are often very hard to work around if at all. Software limitations are generally a bit easy to get around. In this case the brilinear was just about speed at the sake of IQ. Swich to an older driver, Brilinear gone. Easy to get around it. Angle Depenent AF/Bi-linear is a much harder hard ware issue to get around.Ahhh look at our review here were it tells us about the bilinear af limitation: http://www.beyond3d.com/reviews/ati/radeon8500/index8.php
And we all discovered the angle thing here in the forums shortly after (with the help of pccen's program and the SS screen shot of the rotating floor room). So maybe not at the very launch of but close enough..
Where's the "both" option? It alleviates a lot of the quality issues with NVIDIA's brilinear and provides a nice healthy performance boost, but it was marketed as full trilinear (they knew that it would be called full trilinear because of the mipmap tests and never mentioned this feature). It's not full trilinear, not by any stretch of the imagination (full trilinear is necessary on principle and as a means of providing a baseline between cards), but it is an excellent optimization and one that a lot of users would probably use in upcoming games.
Well, if by "upcoming" you mean "twelve months from now"--how long have we known that the Intel integrated chips were going to support PS2.0?Bjorn said:
The Baron said:
I'm thinking more about games like Half Life 2, Doom 3, Far Cry with patch..
For sure, there will be loads of games that will be able to take advantage of AF, trilinear optimizations but those games will likely run very fast on all cards anyway. And i'm thinkin that the above games will be used rather heavily as benchmarks.
I'll reserve judgement on trylinear's IQ until I see a few more thorough analyses, but mikechai's (actually xbit's) images don't bode well. Edit: It appears the R420 shot may have inadvertently used bilinear, as CoD defaults to that when you swap boards. So hold that boding.
Not advertising this "feature" was completely disingenuous, though. ATi's PR material might have been somewhat technically correct by stating that they "defaulted" to trilinear, but ATi hasn't achieved its recent goodwill through half-measures.
Still, it appears that texture filtering is one source of possible optimisations that both IHVs are focusing on (ATi has been "playing" with trilinear since the 8500, I've just rediscovered with a quick search), so I guess we'd better get used to cut corners.
Not advertising this "feature" was completely disingenuous, though. ATi's PR material might have been somewhat technically correct by stating that they "defaulted" to trilinear, but ATi hasn't achieved its recent goodwill through half-measures.
Still, it appears that texture filtering is one source of possible optimisations that both IHVs are focusing on (ATi has been "playing" with trilinear since the 8500, I've just rediscovered with a quick search), so I guess we'd better get used to cut corners.
nutball said:
I agree with this actually. It would've sounded pretty cool if it was mentioned at launch, but I agree with Dave as mentioned in the other thread: if RV3x0 had it from day one (probably a necessity as a result of that target market) and the x800 is a supercharged RV3x0 design, then it's probably just that they just forgot it was part of the design (this target market certainly doesn't *need* the optimizations). Considering ATI's previous marketing branding, it probably would've been named SmartTrilinear or SmartFiltering or something equally inane
This would have been a lot better from the start if someone had just simply remembered this optimization was in use. Looks like bad communication all around.
AAAGGHHH!!!!!! Give me an option. All I ask. A little sentence with a little box to check and uncheck.
I see no SIGNIFICANT difference, but I do feel slighty jipped/lied to. Like buying a car with 335 HP when it was advertised as having 340 HP. You dont' feel a difference, but people sure as hell sue over that stuff. We need better honesty from both companies. Do whatever the hell you want to, just let us know so we can decide if we dont' want your product then.
I see no SIGNIFICANT difference, but I do feel slighty jipped/lied to. Like buying a car with 335 HP when it was advertised as having 340 HP. You dont' feel a difference, but people sure as hell sue over that stuff. We need better honesty from both companies. Do whatever the hell you want to, just let us know so we can decide if we dont' want your product then.
