View Full Version : IXBT's article on trilinear
Evildeus
05-24-04, 12:49 PM
I supposed we will see the english article in the future, but here's the russian one:
The russian article (http://www.ixbt.com/video2/nv40-rx800-3.shtml)
Have a look:
http://www.ixbt.com/video2/images/r420-nv40-Trilinear/util.jpg
I supposed we will see the english article in the future, but here's the russian one:
The russian article (http://www.ixbt.com/video2/nv40-rx800-3.shtml)
Have a look:
They stated that you can notice the diifference between IQ qualities in stills, and that even while playing the Ati card may have a bit more of a moire pattern and some dithering.
(my wife is russian)
so there you have it, the Russians say it's degrading the IQ.
http://www.freetranslation.com/web.asp
http://www.freetranslation.com/web.asp
arrg! I couldn't get it to work.
Lfctony
05-24-04, 04:22 PM
"the images received on chips ATI, will sometimes look more precise, and the images received on chips NVIDIA, - more dim, but at movement of the chamber of video, received on chips ATI, will be more inclined to occurrence of moire and "песочка" (dithering), arising because of an insufficient filtration (under-filtering)."
There goes the "IQ is the same" theory.
"the images received on chips ATI, will sometimes look more precise, and the images received on chips NVIDIA, - more dim, but at movement of the chamber of video, received on chips ATI, will be more inclined to occurrence of moire and "песочка" (dithering), arising because of an insufficient filtration (under-filtering)."
There goes the "IQ is the same" theory.
thanx for getting it to work, least now I know my wife will suffice for Russian translating.
so, now I'm guessing that fanATIc's will fallback to the excuse, "you can barley tell" instead of "the IQ is the same"
it's official, that I am now officaly pissed off about the whole situation, I was holding out, and trying to give Ati the benifit of the doubt, but.
shame on Ati.
or maybe i'm just pissed becasue it's monday... :)
It's monday (aaa) LOL
Weird that you had a problem getting that to work Jarred, I tried like 3 others, and that was the one that worked NP.
Lfctony
05-24-04, 04:31 PM
It's not so much the filtering, after all, Nvidia is doing it too. It always has been and will be they way they hid it, instructing everyone to use Full Trilinear on Nvidia cards to ensure a "fair" comparison, with them using their optimised algorithm. Well, it appears the comparison wasn't so fair after all.
It's monday (aaa) LOL
Weird that you had a problem getting that to work Jarred, I tried like 3 others, and that was the one that worked NP.
it spat out a bunch of weird text, maybe I don't have any of the lanuage packs installed or something,
suff like this
Oèëüòð*öèÿ de Oðèëè*åé**ÿ (trilinear que filtra (trilinear MIP que traza)) - ôèëüòð*öèè de áèëè*åé*îé de ð*çâèòèå de ä*ëü*åéøåå (bilinear que filtra) MIP-ì*ïïè*ãîì C (MIP que traza). Oåêñòóðû de MIP-óðîâ*åé de ñîñåä*èõ de èç de âûáîðîê de áèëè*åé*ûõ de äâóõ de ðåçóëüò*ò*ìè de ìåæäó de ç**÷å*èå de ñðåä*åå de âçâåøå**îå de ÿâëÿåòñÿ de ôèëüòð*öèè de òðèëè*åé*îé de Ðåçóëüò*òîì. Aûáîðîê de áèëè*åé*ûõ de äâóõ de èç de îä*îé de ç**÷å*èå de ïðåîáë*ä*åò de ïèêñåëü òåêñåëü de ð*çìåðîâ de ñîîò*îøå*èÿ de îò de ç*âèñèìîñòè de A. O*êîé ìåòîä ïðåäîòâð*ù*åò ìåðö**èå è ðåçêó ñìå*ó ÷åòêîñòè òåêñòóð ïðè äâèæå*èè ê*ìåðû
I don't have any LP installed either, so I dunno whats up with that.
Hmmm
O well
spanker
05-24-04, 05:01 PM
instructing everyone to use Full Trilinear on Nvidia cards to ensure a "fair" comparison, with them using their optimised algorithm. Well, it appears the comparison wasn't so fair after all.
they told reviewers to compare the cards at equal iq, they never said to compare ati's tri vs nv's tri, you got to blame the incompetent reviewers that thought nv's tri and ati's tri was comparable iq wise.
they told reviewers to compare the cards at equal iq, they never said to compare ati's tri vs nv's tri, you got to blame the incompetent reviewers that thought nv's tri and ati's tri was comparable iq wise.
You should read this
http://techreport.com/etc/2004q2/filtering/index.x?pg=1
spanker
05-24-04, 05:34 PM
You should read this
http://techreport.com/etc/2004q2/filtering/index.x?pg=1
its on those slides they ask the reviewers to compare the cards at equal IQ
To ensure a fair comparison between products, we believe that the RADEON X800 should be compared with card running at a similar level of image quality
its on those slides they ask the reviewers to compare the cards at equal IQ
ok thats fine, blame the reviewers, Ati didn't lie, they just failed to offer valueable information that would make reviewers jobs easier.
for the last week or so, eveyones been saying the IQ is the same, and well...it's not.
So, Ati implimented an optimization to their trilinear filtering that gave them an increase in performance while degrating the IQ, and didn't tell anyone about it.
yeah, its the reviewers fault. :lame:
its on those slides they ask the reviewers to compare the cards at equal IQ
Okay :rolleyes: I think you should read it again..
Couldn't reviewers simply compare screenshots / actual gameplay / slowmotion video captures / electron microscopy then decide that x is worse than y?
Or does the reviewer first need to know that x uses a different algorithm before they can proclaim that it is indeed worse than y? :bash: If this algorithm is indeed so evil, we would all have noticed it a long time ago.
Not to mention out in the real world companies do a thing called research and development. It is a companies prerogative to find ways of doing things better. It would be pretty silly to broadcast to the world your secret new way of saving raw computing time, just for your competitor to have your algorithm in their spring refresh. The only real issue here is the extent you take your optimisation. All algorithms have parameters for adjusting the ammount and aggressiveness of the optimisation, whether at design time or runtime. Think Jpeg compression. There is a balance between quality and performace. At some point the marketing people get involved and tell engineering where to put that slider. Whether or not it permanently effects hardware, or is set in ASIC registers it doesn't matter. If you don't like where company x sticks their knob. Go elsewhere. (if you can).
Okay :rolleyes: I think you should read it again..
Note the words "we believe". IE their opinion. It isn't gospel. Any sane reviewer would see that phrase as a free license to do as they wish.
Couldn't reviewers simply compare screenshots / actual gameplay / slowmotion video captures / electron microscopy then decide that x is worse than y?
Or does the reviewer first need to know that x uses a different algorithm before they can proclaim that it is indeed worse than y? :bash: If this algorithm is indeed so evil, we would all have noticed it a long time ago.
Not to mention out in the real world companies do a thing called research and development. It is a companies prerogative to find ways of doing things better. It would be pretty silly to broadcast to the world your secret new way of saving raw computing time, just for your competitor to have your algorithm in their spring refresh. The only real issue here is the extent you take your optimisation. All algorithms have parameters for adjusting the ammount and aggressiveness of the optimisation, whether at design time or runtime. Think Jpeg compression. There is a balance between quality and performace. At some point the marketing people get involved and tell engineering where to put that slider. Whether or not it permanently effects hardware, or is set in ASIC registers it doesn't matter. If you don't like where company x sticks their knob. Go elsewhere. (if you can).
huh? I understand the whole marking people dictating where "sliders" go but how is this relevant to what Ati has done?
your basically stating that what Ati did was ok because it's a secret, and other companies shouldn't know about it, you know there's this thing called patent laws that protect things like that right?
and I don't think this is a matter of where company X sticks their knob, your missing the point entirely.
huh? I understand the whole marking people dictating where "sliders" go but how is this relevant to what Ati has done?
your basically stating that what Ati did was ok because it's a secret, and other companies shouldn't know about it, you know there's this thing called patent laws that protect things like that right?
and I don't think this is a matter of where company X sticks their knob, your missing the point entirely.
It's not got it's patent yet. I'm and engineer, so I care where they stick their knob. Being "lied" too means absolutely nothing to me. :)
They stated that you can notice the diifference between IQ qualities in stills, and that even while playing the Ati card may have a bit more of a moire pattern and some dithering.
(my wife is russian)
so there you have it, the Russians say it's degrading the IQ.
Shhhhhh!!! lets keep that in secret.. :)
let ATi folks be aware of that by themselves .radeon9600xt "optimizations" in a highend card best quality settings . :screwy:
It's not got it's patent yet. I'm and engineer, so I care where they stick their knob. Being "lied" too means absolutely nothing to me. :)
ok, so... they can put whatever on the box and you'd be cool with it? Like, it gives GREAT head or something, heck, they could even say it will make you toast?
Shhhhhh!!! lets keep that in secret.. :)
bah what do they know anyway, bunch of communists! :P
ok, so... they can put whatever on the box and you'd be cool with it? Like, it gives GREAT head or something, heck, they could even say it will make you toast?
What is on the box is irrelvant, as long as it passes trading standards. If it gets good reviews in the press, and demos well at the PC shop then that's good enough for me. If it makes good toast, I'd expect that to be on the box, and have some good reviews.
What is on the box is irrelvant, as long as it passes trading standards. If it gets good reviews in the press, and demos well at the PC shop then that's good enough for me. If it makes good toast, I'd expect that to be on the box, and have some good reviews.
well I don't think some reviewers are too happy about Ati's new trilinear methods... but make your own judgement.
well I don't think some reviewers are too happy about Ati's new trilinear methods... but make your own judgement.
Anything that makes news is good for reviewers and their websites in their eyes.
vBulletin® v3.7.1, Copyright ©2000-2009, Jelsoft Enterprises Ltd.