PDA

View Full Version : IXBT's article on trilinear

Pages : 1 2 [3]
ChrisRay
05-26-04, 06:37 PM
The 9600 was supposed to match up to the 5600, which it it did quite well. What the 5700 being faster than the 9600 has nothing to do with this thread. People get off saying that ATI cards have better IQ because that's their opinion. Nothing to do with hell or damnation. The FX is junk IMHO, fair enough it does the job it's designed to do, but comparing it to R3x0 or NV40...


Actually the 9600 was supposed to match up to the 9500 Pro, Which it never successfully did,

And with these optimisations. It makes you question just how much slower it actually was.
Jarred
05-26-04, 06:39 PM
Actually the 9600 was supposed to match up to the 9500 Pro, Which it never successfully did,

And with these optimisations. It makes you question just how much slower it actually was.

so whats your opinion on the russian article?
ChrisRay
05-26-04, 06:42 PM
so whats your opinion on the russian article?


I'm Still Evaluating multiple sources and filtering tests before I reach a conclusion. Besides, Opinions are like butts, Everyone has one and they usually stink.


I'm going to try and see if its noticable to me. I mean you can hold my opinion of Brillinear being good enough and looking fine, If you like. I havent seen this implementation in motion yet.
Jarred
05-26-04, 06:45 PM
I'm Still Evaluating multiple sources and filtering tests before I reach a conclusion. Besides, Opinions are like butts, Everyone has one and they usually stink.


I'm going to try and see if its noticable to me. I mean you can hold my opinion of Brillinear being good enough and looking fine, If you like. I havent seen this implementation in motion yet.

hmmm... I'm wondering when other sites will put up reviews, it's been a while now, plenty of time to compare the two cards under most circumstances. I'm not going to discredit the Russian site, but it will be good see more reviews.

I'd say it's safe to say that the IQ in Ati's filtering is in "question" :P
RobHague
05-26-04, 08:09 PM
I can't find where that was mentioned in the article, can you point it out?

I dunno, like i said i gave up on the translations as they just made it even more confusing than it was when it was in russia, hah.

at the risk of being in breech of the unwritten rule ;)...
http://www.rage3d.com/board/showthread.php?s=&threadid=33760559&perpage=30&pagenumber=1

ATI applies this algorithm in most cases, as you say. It doesn't apply the algorithm in certain situations that ATI knows the algorithm does not produce good results. So IXBT publishes an article, and forces the algorithm to be forced in ALL cases. So in those cases where ATI was not applying it, we can see that the IQ is lower.

actually your not getting what he is saying. he is asing why is the website disabling the detection in order to apply the optimizations in all cases, regardless if it produces a worse image. that isnt how the algorithm is supposed to work. so the reason everything is brilinear in the pics is because of what the website disabled, not because of ati. a better test would have been for thm to completely turn off the optimizations so that we get full trilinear and then compare those to the default images. i guess since they couldn't do that they pull this crap.

Is what i was refering too.... like i said i wasent willing to wrestle with the jibberish translations google gives lol :)
Jarred
05-26-04, 08:18 PM
I dunno, like i said i gave up on the translations as they just made it even more confusing than it was when it was in russia, hah.

at the risk of being in breech of the unwritten rule ;)...
http://www.rage3d.com/board/showthread.php?s=&threadid=33760559&perpage=30&pagenumber=1





Is what i was refering too.... like i said i wasent willing to wrestle with the jibberish translations google gives lol :)


hmmm... well, I donno where they got that in the article, guess we'll wait and see.

hmmm.ok, after reading the intire thread I think I understand what they're saying, however, even if they had forced it to be on all the time they claimed it would be on 90% of the time anyway, so it's the extra 10% thats gonna contribute 100% to the IQ degration?

I also think the reason they did this was so they could compare both cards optimizations when they're both on all the time.

although, I have to admit, I want to see more reviews now.
Richthofen
05-27-04, 05:37 PM
another article from those german guys who discovered ATIs cheat

http://www.computerbase.de/artikel/hardware/grafikkarten/radeon_x800_texturen/

ATI seems to get more and more under pressure with every day.

Its about time that they offer an option to disable their optimization stuff and that reviewers start to benchmark those cards again against the competition under fair conditions!
Jarred
05-27-04, 05:51 PM
another article from those german guys who discovered ATIs cheat

http://www.computerbase.de/artikel/hardware/grafikkarten/radeon_x800_texturen/

ATI seems to get more and more under pressure with every day.

Its about time that they offer an option to disable their optimization stuff and that reviewers start to benchmark those cards again against the competition under fair conditions!

they claimed that they arrived at the same conclusions that the Russians did at XBitlabs

so, now we have the Russians and the Germans claiming Ati's filtering is degrading IQ.

I hoping some american reviews will come soon.
RobHague
05-27-04, 08:14 PM
they claimed that they arrived at the same conclusions that the Russians did at XBitlabs

so, now we have the Russians and the Germans claiming Ati's filtering is degrading IQ.

I hoping some american reviews will come soon.

You can't trust the russians :kill:, everyone knows that ;) they joined sides with the germans until they started losing... and the Germans themselves? pfff - they started the war in the first place. They prolly tied up ATi staff with piano wire and forced a confession out of them. Not to be trusted at all :lol:...

:Edit:

Gotta love the title once its coverted too : Radeon X800 and the textures
How much does optimization stand humans? :lol2:

Anyone speak/read german and can give us a proper translation.
Jarred
05-27-04, 08:33 PM
You can't trust the russians :kill:, everyone knows that ;) they joined sides with the germans until they started losing... and the Germans themselves? pfff - they started the war in the first place. They prolly tied up ATi staff with piano wire and forced a confession out of them. Not to be trusted at all :lol:...

huh? are you talking WW2? I don't think the Russians were loosing, in fact, all of Europe could be changed right now if the russians had not declared war on the germans.

The largest human sacrifice in WWII was made by the russians, in my opinion, they won the war.

sorry bit off topic. :)
RobHague
05-27-04, 08:36 PM
No i meant until the germans started losing. I always thought the russians were with the germans and then switched sides, or was is that they just didnt get involved until it looked like they were next.. hmm. School history lessions are such a distant memory...

Anyways, as i said anyone speak german and care to provide a decent translation on the main points? :)
Jarred
05-27-04, 08:38 PM
No i meant until the germans started losing. ;)

Anyways, as i said anyone speak german and care to provide a decent translation on the main points? :)

:) the germans weren't loosing untill the Russians declared war on them.
ChrisRay
05-27-04, 08:41 PM
But Germany Declared War on Russia, Not Vice Versa.

Germany Broke its no aggression pact with Russia, And thus ended forcing Germany to fight a 2 way war. It was germany who betrayed Russia First. Not Vice Versa.
Jarred
05-27-04, 08:42 PM
aye I was just going to correct that thank you

"On June 22nd Hitler ordered the invasion of the USSR"

and Hitler made the same mistake that Napoleon made over a hundred years ago.
RobHague
05-27-04, 08:43 PM
:rofl

Looks like were gonna get into a history dibate now... :lol:
ChrisRay
05-27-04, 08:44 PM
Nothing really to debate, other than wasnt Russia who betrayed Germany First ^^
Jarred
05-27-04, 08:45 PM
But Germany Declared War on Russia, Not Vice Versa.

Germany Broke its no aggression pact with Russia, And thus ended forcing Germany to fight a 2 way war. It was germany who betrayed Russia First. Not Vice Versa.

there are romurs that Hitler had some sort of disease, that he knew he was going to die from, and he wanted all or nothing, he didn't want to share it with Russia, becasue it really doesn't make sense why he started a two front war, ...why not wait untill europe is secured?

interesting bit on Napoleons invasion

http://uts.cc.utexas.edu/~jrubarth/gslis/lis385t.16/Napoleon/

sorry, no more on this topic for me. ...
cthellis
05-27-04, 10:42 PM
Nothing really to debate, other than wasnt Russia who betrayed Germany First ^^
Yeah, but first Russia leaked accusations of Germany's illegal optimizations to the press, so really they were asking for it.
Shamrock
05-27-04, 11:54 PM
Dont dis the Russians guys, they are prettty cunning and intelligent. I have gotten MANY hardware hacks from the Russians....like getting a firmware update for my discontinued modem a couple years ago. Those guys really know how to program, and modify things. Oh yeah, I also got a firmware DOWNGRADE, because Plextor decided to put DRM or some crap into my firmware for my 12/10/32a, they modified the firmware for the Plextor and viola!

So, I think the Russians know what they're talkin about. :)
Quasar
05-30-04, 05:33 PM
Gotta love the title once its coverted too : Radeon X800 and the textures
How much does optimization stand humans? :lol2:

Anyone speak/read german and can give us a proper translation.

Not word-by-word:
X800 and (the) textures: How much optimization can you bear?

And yes, basically, ATis "intelligent algorithm" does nothing more than calculate a checksum of the contrasting bits between the mipmaps. If that contrast is not exceeded, the mip-levels can differ so much as they want.

Especially interesting it gets, when you take Mip5 and Mip6 of a 1024² texture. Even with a b/w-contrast there's no way that a white Mip6 (except for one pixel!) from an all-black Mip5 will get trilinear filtering - ATi does not think it's enough difference.

Furthermore, they play around with LOD when the optimization of trilinear filtering is enabled. Thus, they can filter with 2xAF where an LOD-correct (as per Nyquist-Criteria) implementation would already require 4 AF-Samples. Texture-shimmering is the logical consequence.

Sorry, but the article is too long to translate completely, but if there are questions regarding single paragraphs, pictures or sentences, i'll do my best to help you out.