Did anyone do a Filtering Image Quality test "AF" on GTX 470/480??

[ajaidev]

For people who forgot this is the one in which 5870 got perfect marks



http://www.hardocp.com/article/2009/..._card_review/6

Anyone with a GTX 480/470 try the test if any review did it post a link...

[demonkevy666]

For people who forgot this is the one in which 5870 got perfect marks



http://www.hardocp.com/article/2009/..._card_review/6

Anyone with a GTX 480/470 try the test if any review did it post a link...

yeah I was wonder this my self it seems no one even bothered with it :/

found one anandtech did it.

http://www.anandtech.com/show/2977/n...th-the-wait-/7

I just noticed this but it looks like the 285 gtx still has higher texter fill rate

51 Vs 42 that's not an improvement at all...

[ajaidev]

Thanks demonkevy666 and wow it looks bad i mean GTX 480/470 is suppose to be better than GTX 285 but this seems like they are not even trying.

Bottom line is the Radon does have better AF implementation even if it is in theory.

[Jowy Atreides]

So, fail for thermi.

6 months lead time.

[Chumbucket843]

the idea of filtering is to take unwanted signals out of an image which involves sampling the texture and using the correct mip map. just because ATi is angle independent doesnt necessarily make their filter higher quality.

check this out.
http://www.pcgameshardware.com/aid,6...eviews/?page=5

if anyone with a 5870 would like to make a comparison to nvidia's filter there is a better way.
download this.
http://www.3dcenter.org/download/3dcenter-filter-tester
and this 8192x8192 massive .dds texture then use it in the filter test.
http://www.mediafire.com/file/yizwgmxxbmj/bigtex.dds

[Chumbucket843]

So, fail for thermi.

6 months lead time.

why do you even bother opening nvidia section? for the lulz?

[demonkevy666]

this card lacks ROP's

48 isn't much of an upgrade from 32 :/ it's only like 33% more pixel fill sure about 65% more.

should be up near 64 ROP's, though that's a bit much I would went with a re design of shaders that's what the 4,000 series was from the 3,000.

sure they great at what they do but the where taking up a lot of space, the design made them smaller and pack more in there.
( yes I know it's really 160 shaders for the 4870 and 320 for the 5870)

[Musho]

the idea of filtering is to take unwanted signals out of an image which involves sampling the texture and using the correct mip map. just because ATi is angle independent doesnt necessarily make their filter higher quality.

check this out.
http://www.pcgameshardware.com/aid,6...eviews/?page=5

if anyone with a 5870 would like to make a comparison to nvidia's filter there is a better way.
download this.
http://www.3dcenter.org/download/3dcenter-filter-tester
and this 8192x8192 massive .dds texture then use it in the filter test.
http://www.mediafire.com/file/yizwgmxxbmj/bigtex.dds

I did that with my 5870. What exactly should I see, though? How can I know whether the filtering is any good?

[Chumbucket843]

the left half is 16xAF HQ on a gtx260. the other is alu rendering at the settings displayed. look for sharp clear textures. if you move set texture movement you can see some flickering that can be noticeable in games.

[Musho]

This is what I got with a 5870. Not sure if this is considered good or not

[zalbard]

this card lacks ROP's

Looks like decent drivers and improved memory bandwidth is what it really needs! Performance wise at least.

[Chumbucket843]

rops are fine. ATi didnt increase rops from 16 since r420.modern games use much more shading and texturing.

This is what I got with a 5870. Not sure if this is considered good or not ]

thanks a lot. i got some interesting results.

to see the difference you need to magnify the detail. the image on the right is enlarged 700%. from left to right the order is HD5870, GTX260, ALU 16x, ALU 512x. you can see that nvidia is closer to the perfect ALU filter. when you get to more extreme angles (where it approaches gray) ATi is slightly better. these differences are small but noticeable and if you are an image quality junky like me it is important.

[demonkevy666]

rops are fine. ATi didnt increase rops from 16 since r420.modern games use much more shading and texturing.


thanks a lot. i got some interesting results.

to see the difference you need to magnify the detail. the image on the right is enlarged 700%. from left to right the order is HD5870, GTX260, ALU 16x, ALU 512x. you can see that nvidia is closer to the perfect ALU filter. when you get to more extreme angles (where it approaches gray) ATi is slightly better. these differences are small but noticeable and if you are an image quality junky like me it is important.

I can't see anything :/

you might have to circle it lol

VS you'll a lot of double, maybe slightly more.

VS

almost double but not quite.
included 280 gtx too

[Chumbucket843]

I can't see anything :/

you might have to circle it lol

it is very hard to see. compare each row of pixels to eachother. the radeon is slightly off. its so small its almost negligble. this just puts to rest all of the arguments about AF. you have to try and look for a difference to even see. my eyes might be playing tricks on me.

[QuadDamage]

How to get the same textures you guys do? Anyway here's my GTX480.

[Carfax]

QuadDamage, what kind of cooling are you using on your 480?

[Jowy Atreides]

why do you even bother opening nvidia section? for the lulz?

Just to converse with you, my sweetheart.

Perhaps I have an interest in graphics cards. No, that's too "out there"

[QuadDamage]

QuadDamage, what kind of cooling are you using on your 480?

Stock cooling and stock TIM in Cosmos S, CPU on water. She's still a virgin heh. Usually runs a little cooler but my room is pretty warm at the moment.

[Carfax]

Stock cooling and stock TIM in Cosmos S, CPU on water. She's still a virgin heh. Usually runs a little cooler but my room is pretty warm at the moment.

Thats a pretty good overclock for stock cooling

What are your temps like at load?

[Musho]

How to get the same textures you guys do? Anyway here's my GTX480.

[img].http://img707.imageshack.us/img707/5094/affilter3.png[/img]

[img].http://img220.imageshack.us/img220/917/aftester0.png[/img]

By downloading this file: http://www.mediafire.com/file/yizwgmxxbmj/bigtex.dds
And by opening it with the filter tester program. Good luck!

[QuadDamage]

Thats a pretty good overclock for stock cooling

What are your temps like at load?

82C max while gaming. Pushed it to 87C or so with Furmark at 1920x1080 and it hits 84C in Vantage with fan at AUTO of course.

Here's the png screenie with the new textures. BTW i'm getting 69fps average in this new Stone Giant DX11 demo and from what i've seen it twice as fast compared to the 5870.

[Carfax]

Those temps are still good, if thats overclocked with the fan speed on auto..

A far cry from the 95c load temps the professional reviewers were getting.

[QuadDamage]

Truth be told most of the so called pro reviewers these days are a bunch of noobs. None of them managed to squize more than 806Mhz core on an open bench failing to realize that a well vented case with a 200mm quiet fan on the side panel works much better.

I can bench at 860/2000 with the fan set to 100% but that's way to noisy.. the good news is at 860Mhz the card isn't voltage limited so if you're planning on slapping w/c on your GTX480 you'll prolly be able to hit 850Mhz+ core for 24/7 gaming/benching easy.

[AAbenson]

QuadDamage whats the ambient temp in your room?
82c matters very little if you live in Finland/Norway/Canada/Alaska

[QuadDamage]

Relax my room is at 22C most of the time and never gets colder than 19C. I'm not as hardcore as i used to be back in the day.