ATi is ch**t**g in Filtering

Marc said:
Borsti said:
Well the whole discussion started because the guys at Computerbase found out that the X800 is loosing performance when colored mipmaps are used.
And the discussion still going on because there is a real difference in the trilinear :/

Cand you do the same test on R3xx ?

And take a look at the differences beetween R420 8x AF Performance shoot and R3xx same shot via compressonator ?

Sure, but we´re already at pure bilinear so a 9800xt should behave in the same way. I´m not sure what this would proove.

Lars - THG
 
mikechai said:
Could it be the driver detects when colored mipmap is used and force trilinear no matter what the CP setup or application request?

No, you can visibly see that Bilinear is enabled with coloured mip maps with all drivers / boards (as as what is seen with the Cat3.4 shots I took earlier). You can also see that with the Cat3.4's "Brilinear" is on when the texture slider is knocked down a notch on 9600.
 
mikechai said:
Could it be the driver detects when colored mipmap is used and force trilinear no matter what the CP setup or application request?

You can clearly see that the card is using bilinear.
 
Hanners said:
Evildeus said:
Yeah sorry, confusion :oops:
But Hanners post just confirms that there's no reason that people should complain, as they need to change the default filtering mode.

On the 9600 yes, but I'm assuming from the shots posted here that it seems to be the default mode on X800 boards, hence the complaints. I should have made that bit clearer in my last post.

Except that from the straight Bilinear / Trilinear shots the 9600 appears to be doing the same thing as X800 (and has been since Cat3.4), however this does not appear to be the case when coloured mips are used.
 
So, both the 9600 and X800 are doing 'brilinear' now in later drivers, even on the highest settings on the texture quality slider?

My brain isn't made for thinking on a Monday morning, I'm sure using it so soon in the week is doing me no good at all. :p
 
Hanners said:
So, both the 9600 and X800 are doing 'brilinear' now in later drivers, even on the highest settings on the texture quality slider?
Yay. :(

But when you're using colored mipmaps, this brilinear filtering will be disabled. The driver seems to contain a colored-mipmap detection mechanism.
 
Hanners said:
Evildeus said:
Yeah sorry, confusion :oops:
But Hanners post just confirms that there's no reason that people should complain, as they need to change the default filtering mode.

On the 9600 yes, but I'm assuming from the shots posted here that it seems to be the default mode on X800 boards, hence the complaints. I should have made that bit clearer in my last post.
Yeah, sure, that's the issue.
 
Exxtreme said:
But when you're using colored mipmaps, this brilinear filtering will be disabled. The driver seems to contain a colored-mipmap detection mechanism.

This seems to be the point of contention though - From the shots I've seen, the differences are still visible with coloured mipmaps turned on.
 
Hanners said:
So, both the 9600 and X800 are doing 'brilinear' now in later drivers, even on the highest settings on the texture quality slider?

No.

Bit comparisons of Trilinear vs Bilinear appear to show that X800 is doing "Brilinear" filtering by default - because the colouring of mipmaps appear to show that full trilinear is in operation the assumption is that when you look at normal images "Brilinear" is forced on by default but "removed" when coloured mips are in use.

However, 9600 shows similar behaviour under bit comparisons, and appears to have done so since its release (but without anyone noticing any "Brilinear" effects in games, as yet). But, when you turn down the texture slider on 9600 and the early drivers the texture quality is still lowered in comparisons of Bi-vs-Trilinear and the coloured mip maps show "Brilinear".
 
Exxtreme said:
But when you're using colored mipmaps, this brilinear filtering will be disabled. The driver seems to contain a colored-mipmap detection mechanism.

Except when the texture quality slider is dropped one level on the 9600 :!: :?:
 
DaveBaumann said:
Exxtreme said:
But when you're using colored mipmaps, this brilinear filtering will be disabled. The driver seems to contain a colored-mipmap detection mechanism.

Except when the texture quality slider is dropped one level on the 9600 :!: :?:

Has anyone tried dropping the slider a notch on the X800 to see if the same rule applies?
 
Umm, with respect to texture ops preventing ALU1 being used *for the entire duration of the texture processing*, I think this is a mistaken belief. A texture op already takes hundreds of cycles. This latency is hidden by pipelining, and I don't believe that the ALU is respronsible for doing 100% of the texture operation. More than likely, it does some processing on the texture coordinates, the LOD, and hands it off to a work queue where it is processed by a fixed function filtering unit.

A more plausible explaination is that a texture op uses up additional context resources, which may interfere with register limits and introduce a penalty, because it holds onto those resources longer before returning them back for reuse.
 
K.I.L.E.R said:
So are Ati's drivers detecting coloured mipmaps to cheat?
I will not use the bad 'c'-word :D but it seems that ATi's drivers are able to detect coloured mipmaps. If so, then the "brilinear" filter will be disabled.
 
Evildeus said:
BTW, if it's not visible how come you did find out this trick? :?
It's easy. You make two screenshots ... one with bilinear filtering and one with trilinear/brilinear filtering. Then you make a picture compare with photoshop or other tools. After this you see the areas where the pictures are different and you see where the pictures equivalent.
 
Yes of course, the question was more like, why did you make those comparisons? But i realise that's a bit dumb, if you found some differencies, sure you would investigate more and more to see what's going on. Sorry for asking for something so obvious :oops:
 
Back
Top