This whole R600/G80 benchmarks thing is nonsense.

[josh6079]

Actually artifacting of any kind tends be be a very large problem indeed for video cards since it normally indicates they are dying.

Not all cases of artifacting imply the card is going dead. Sometimes the issue can be resolved with a driver update. Why you're so objective to this simple fact is beyond logic.

You are being retarded (as per usual)

Kinda like your arguments.

Tell us again, oh so wise IQ analyzer, is it the more colored rings the better the AF?

...as the result of a source engine bug nvidia was forced to fix...

Why shouldn't they have fixed it? It only happened on their G80. And if it was purely an engine bug I don't think nVidia could have fixed it. Turns out a driver release solved the issue because it was a driver conflict.

 

[Gstanfor]

Good lord you two are pathetic. Defend ATi/AMD to the bitter end (that end is very close now, too). You should live together under a bridge as a couple.

You haven't tried the 165.01's, have you BFG10K? They fix lots and lots of things that have been broken/missing in nvidia's drivers for a long time now. Not all things mind you, but most.

 

[Cookie Monster]

Isnt the source fog fixed? i think so with couple of drivers back.

 

[josh6079]

Defend ATi/AMD to the bitter end (that end is very close now, too).

Fine, let's talk about the TOPIC then.

Man that G80 Ultra is a good card. Definitely worth the price tag...

 

[BFG10K]

Isnt the source fog fixed?

Yes and that's my point. Gstanfor was trying to tell us that artifacts imply faulty hardware.

Also the problem persisted for about six months before it was fixed; meanwhile he's harping on about a chain-link problem on a card that hasn't been released yet.

 

[Gstanfor]

There is a difference between a driver bug and an artifact, no matter what Josh might like to think.

I don't care if R600 is released or not. Unless the leaked AMD slides are an elaborate hoax, ATi's new CFAA (F for f**cked?) is a blur filter. I don't care where the filtering is done, ROPS or Shaders, mixing data from unrelated pixels doesn't constitute an acceptable AA solution in 2007 in my books, and I'll bet I'm not alone with that opinion. Heck, Quincunx in 2001 wasn't really that acceptable, but at least nvidia had the good excuse of lack of bandwidth with GF3 -- something whch, R600, the 105gb memory bandwidth wonder that then resorts to blurry AA can't claim.

 

[Gstanfor]

Originally posted by: BlizzardOne
1. i ain't your mate.

2. i may have only been registered a little while, but i've been around this place plenty long enough.

like i said, if you can back up your claim, great.

I figured the location couldn't be too far from my last save (near the start of Highway17) and it wasn't (just before the bridge).

Half-Life 2 Savegame so others can verify on different hardware (would like a screenshot from someone who has an R600).

screenshot from savefile as close as I can be bothered matching the image.

Some further images from the surrounding area follow.
Text
Text
Text
Text

You'll notice there is a big bridge in the background. One other possible way of explaining the misrenedered polygon if it isn't an artifact is an error on the part of the hardware determining the z-distance of that polygon, brought about by ATi's Transparency AA algorithm, or in other words Hyper-Z is suffering from hyper tension

Drivers were 165.01, HL2 profile from nHancer for the skeptics.

 

[josh6079]

You haven't tried the 165.01's, have you BFG10K? They fix lots and lots of things that have been broken/missing in nvidia's drivers for a long time now.

You mean like artifacts?

There is a difference between a driver bug and an artifact, no matter what Josh might like to think.

Point to where I said they aren't different. All I've said is that sometimes artifacts can be corrected through software updates, whether they be a driver, a patch, a bios, etc.

Jumping to the conclusion that said artifact in HL2 is the result of malfunctioning hardware when the card and driver have yet to be released is idiotic. And doing it in a thread that's not even about the R600's soft-launch (mainly because it hasn't had any launch) just highlights your agenda.

 

[Lord Banshee]

oh oh more benchmarks

http://www.xtremesystems.org/forums/showthread.php?p=2171205#post2171205

Also confirmation of a HD2900XT 1GB? i thought they were only going to release a 512MB version. If it true that is great, i like options.

GRR the link is down before i could finish reading it lol

 

[Extelleron]

Originally posted by: Lord Banshee
oh oh more benchmarks

http://www.xtremesystems.org/forums/showthread.php?p=2171205#post2171205

Also confirmation of a HD2900XT 1GB? i thought they were only going to release a 512MB version. If it true that is great, i like options.

Link doesn't work? Xtremesystems seems to be down, like usual.

Anyway the HD 2900XTX is supposed to be a 1GB part with GDDR4 memory and the lower end XT is supposed to have 512MB of GDDR3 memory. There's been some SPECULATION that the XTX will not be released and only the XT will be released at the $400 price point.

 

[Matt2]

404 R600 not found

 

[Lord Banshee]

lol hopefully in a week and two days we will know if they fixed that "404" error.

 

[korgan844]

always 404 error...

 

[BFG10K]

There is a difference between a driver bug and an artifact, no matter what Josh might like to think.

But Josh doesn't have a problem understanding that, you do.

I don't care if R600 is released or not. Unless the leaked AMD slides are an elaborate hoax, ATi's new CFAA (F for f**cked?) is a blur filter.

If you aren't being reimbursed for your antics I truly pity you.

One other possible way of explaining the misrenedered polygon if it isn't an artifact is an error on the part of the hardware determining the z-distance of that polygon, brought about by ATi's Transparency AA algorithm, or in other words Hyper-Z is suffering from hyper tension

So in otherwords a driver problem which nobody was ever denying.

What was taken issue with was your claims of faulty hardware and also with you constantly ignoring nVidia driver problems.

 

[Gstanfor]

If it is a Hyper-Z problem, you need to ask yourself why it only affected 1 polygon, and not all the polygon near the misrendered one which would have had very similar Z conditions.

If it is a driver problem then it only serves to make me even more curious as to just what is inside that 320mb driver of ATi's...

Oh, and Hyper-Z occurs on chip, not in the drivers.

 

[BFG10K]

If it is a Hyper-Z problem, you need to ask yourself why it only affected 1 polygon, and not all the polygon near the misrendered one which would have had very similar Z conditions.

When was it was established to be Hyper-Z? Is the Red Faction issue a problem with nVidia's hardware occlusion culling?

Or are you just rambling your usual random rhetoric?

If it is a driver problem then it only serves to make me even more curious as to just what is inside that 320mb driver of ATi's...

How does driver size determine whether a chain-link is rendered properly or not?

Honestly, do you put any thought into your posts or do they just spray out on a whim?

 

[Gstanfor]

How does driver size determine whether a chain-link is rendered properly or not?

Honestly, do you put any thought into your posts or do they just spray out on a whim?

Normally it doesn't, unless, of course, the driver is stuffed to the gills with "optimized" rendering portions of game engines which is about the only reason I can see for a 320mb driver....

quote:
If it is a Hyper-Z problem, you need to ask yourself why it only affected 1 polygon, and not all the polygon near the misrendered one which would have had very similar Z conditions.


When was it was established to be Hyper-Z? Is the Red Faction issue a problem with nVidia's hardware occlusion culling?

Or are you just rambling your usual random rhetoric?

Take a look at the first word you quoted - "If". I'm only speculating what the problem might be - I don't know for sure what it actually is. I know damn well there *is* a problem, of some description and I also know that Hyper-Z has caused problems for games in the past for the Radeon series.

Have I also mentioned that the fact that this problem affects ATi's beloved Half-life 2 makes the problem all the more entertaining to me?

 

[BFG10K]

Normally it doesn't, unless, of course, the driver is stuffed to the gills with "optimized" rendering portions of game engines which is about the only reason I can see for a 320mb driver....

Or it could just be .NET + WDM + international versions of both R6xx and legacy drivers.

Also do you have the driver? If not how do you know it's 320 MB?

If". I'm only speculating what the problem might be - I don't know for sure what it actually is.

Yet no such speculation on your part when nVidia has artifacts. Instead from you we either get denial or you blaming the user and/or developer.

 

[Gstanfor]

Also do you have the driver? If not how do you know it's 320 MB?

No, I d on't, and until the driver is made publically available for inspection and vetting it will remain extremely suspect in my eyes. One certainly hopes that all future drivers will be 320mb in size in order to allay suspicions. Ah, memories of QUAK, winbench98 and Final Reality....

As for Red Faction, I've got that installed on one of my drives somewhere. Haven't touched it in years. Post up a savegame in a problem area and I'l ltest it out for you.

 

[BFG10K]

and until the driver is made publically available for inspection and vetting it will remain extremely suspect in my eyes.

Of course it will. Nevermind the fact that it might not actually be that big. :roll:

Ah, memories of QUAK, winbench98 and Final Reality....

Ah, the memories of nVidia substituting shaders in dozens if not hundreds of games.

As for Red Faction, I've got that installed on one of my drives somewhere. Haven't touched it in years. Post up a savegame in a problem area and I'l ltest it out for you.

You're wasting your time unless you have G80. If you have one just start a new game and you'll see it straight away.

 

[Gstanfor]

Ah, memories of QUAK, winbench98 and Final Reality....


Ah, the memories of nVidia substituting shaders in dozens if not hundreds of games.

Who was first though, and who had to lower their previous high standards to keep up (something I've never agreed with by the way, although I'm fine with user controllable optimizations).

You really should prove such claims too (something the fantaics never have). Humus's Doom 3 shader replacement was proveable though.

 

[palindrome]

Originally posted by: Gstanfor
The following images are apparently using R600 16x CFAA Anti-aliasing

image1
image2

I did notice what appears to be some sort of rendering artifact on the images (marked with red circle) - discrepancy

Please tell me you really aren't THAT stupid....nvm this is another R600 thread, I guess I should come to expect this now...

(hint: the "artifact" you see in the 16xCSAA is actually a white object in the backround, it can clearly be seen in both pictures)

I hate trolls...

 

[Gstanfor]

Originally posted by: palindrome

Originally posted by: Gstanfor
The following images are apparently using R600 16x CFAA Anti-aliasing

image1
image2

I did notice what appears to be some sort of rendering artifact on the images (marked with red circle) - discrepancy

Please tell me you really aren't THAT stupid....nvm this is another R600 thread, I guess I should come to expect this now...

(hint: the "artifact" you see in the 16xCSAA is actually a white object in the backround, it can clearly be seen in both pictures)

I hate trolls...

Keep reading palindrome. Check out my comparison image. There is a savegame to go along with that as well as other images, if you care to actually read...

There is no "white object in the background" and if there where it would be extremely unlikely to be composed of only one polygon.

I hate idiots.....

 

[BFG10K]

Who was first though, and who had to lower their previous high standards to keep up

:roll:

You really should prove such claims too (something the fantaics never have).

Sure, how about some from 3DMark's audit report?

In game test 4, the water pixel shader (M_Water.psh) is detected. The driver uses this detection to artificially achieve a large performance boost - more than doubling the early frame rate on some systems. In our inspection we noticed a difference in the rendering when compared either to the DirectX reference rasterizer or to those of other hardware. It appears the water shader is being totally discarded and replaced with an alternative more efficient shader implemented in the drivers themselves. The drivers produce a similar looking rendering, but not an identical one.

In game test 4 there is detection of a pixel shader (m_HDRSky.psh). Again it appears the shader is being totally discarded and replaced with an alternative more efficient shader in a similar fashion to the water pixel shader above. The rendering looks similar, but it is not identical.

Humus's Doom 3 shader replacement was proveable though.

Of course it was provable given no attempt was made to hide it. You can also disable it with Catalyst AI.

How did we disable nVidia's 3DMark cheats at the time?

 

[Gstanfor]

3dmark isn't a game, Doom3 is. I'm not disputing that nvidia replaced shaders there - they made several public statements about their intention to do so at the time.