Mod, please lock, this is no longer on topic.

Page 6 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Elcs

Diamond Member
Apr 27, 2002
6,278
6
81
Originally posted by: nemesismk2
Is this topic back to talking about the Kyro? Guess not! :(

Yeah, some idiots found this place and decided to make it their personal flame thread.

The Kyro 2 is a legendary card and Id like to see some of the techniques used by the manufacturers used today. Could be fun.
 

BFG10K

Lifer
Aug 14, 2000
22,709
2,997
126
Because 7800s have p!ss poor filtering. AA is insignificant in terms of IQ compared to AF.
So again you don't want best possible reproduction, you want best possible filtering and will gladly give up everything else for it.

Why don't you just admit this whole thread is about you trashing optimized AF instead of hiding behind some fictional holy grail which makes it appear your opinion is somehow balanced and rational when it isn't?

Because from what I've seen is AOE3's gameplay sucks while Sacs rocks
But is that mathematically provable?

No, that is a joke. We don't have the option to remaster existing recordings.
No but we can make existing music sound better by buying better equipment.

Actually- the higher end gear you move to at that point the poorer it tends to sound(as you can pick up even more of the limitations of the original source).
So now you have to advocate low-end gear in order to keep your original claim intact. That's quite a conundrum you've got yourself there. Or rather yet another hole you've dug yourself into.

I'm not buying parts with disgustingly poor texture filtering- how exactly is that not practicing what I preach?
You didn't ask for best possible filtering, you asked for best possible reproduction that is available. 7800 SLI matches that claim since the AF you demand is for all intents and purposes no longer available (and thus should be dropped from topic according to your rule addendum #4) and the superior AA is available and it's far better than what your holy grail card offered.

As for not buying such cards, you purchased a 9800 Pro didn't you?

Best possible reproduction of gameplay kind of explicitly implies that it need be playable
Of course, which is why I pointed out that real AF and SSAA are generally not viable because of this very reason. You then responded by showing low detail benchmarks - benchmarks that violated your own rules for both reproduction and playability - to "prove" I was wrong. In otherwords you contradicted yourself while trying to prove me wrong!

No, the image sampled to the monitor is the same resolution as what is being outputted. The backbuffer is four times the resolution which is also the case with multi sampling
Why are you stating the obvious to make it sound I said something else? You know darned well the AA-sampled image is higher than the final image so following your previous simpleton spin on AA you would be forced to come to the flawed conclusion that the concept of AA in general is useless.

It is possible it is not enabled.
No, it isn't posisble. For all intents and purposes real AF is not available because the optimizations are locked at the hardware level which means according to your rule addendum #4 (must be available) you won't be pursuing the issue any further. In fact according to that rule you should've never even been talking about it to begin with.

So you are on record saying that dropping 40FPS does not matter-
Where did that strawman come from? More specifically what does it have to do with the semantic games you're playing in order to ignore the cumulative performance hit of the two?
 

BenSkywalker

Diamond Member
Oct 9, 1999
9,140
67
91
So again you don't want best possible reproduction, you want best possible filtering and will gladly give up everything else for it.

Give up what exactly? I can run Sac now @2048x1536 w/AA- moving to 7800GTX SLI setup gains me a couple of notches worth of AA- which to me isn't worth the reduction in quality of the filtering.

Why don't you just admit this whole thread is about you trashing optimized AF instead of hiding behind some fictional holy grail which makes it appear your opinion is somehow balanced and rational when it isn't?

You are the one being irrational here. Easily demonstrated by statements like-

But is that mathematically provable?

In relation to gameplay quite clearly indicates who is the irrational one.

No but we can make existing music sound better by buying better equipment.

You have real low end gear? You hit a certain sweet spot for a given recording- even when looking at newer recordings- in terms of the accurate reproduction of the accoustical range involved. Beyond that, you start exposing the shortcomings of the reocrding you are listening to. As a generalized example people rave about the iPod's music playback that you purchase through their website which sounds fine on an iPod but piped through a remotely decent setup sounds completely unacceptable(complete loss of all sound stage, sound range is seriously cut off at the high end low ends and middle has turned into mud). You take existing outdated recordings and they sound better up to a point, then your equiptment simply serves to amplify all of their shortcomings.

So now you have to advocate low-end gear in order to keep your original claim intact. That's quite a conundrum you've got yourself there. Or rather yet another hole you've dug yourself into.

Where have I ever done any such thing? Reading comprehension is an important element. I pointed out reality- don't take my word for it- go to any audiophile site or forum and broach the topic.

You didn't ask for best possible filtering, you asked for best possible reproduction that is available.

When tube amps went out of production an awful lot of people just stopped buying new ones. Why? Tube amps were still available used.

7800 SLI matches that claim since the AF you demand is for all intents and purposes no longer available (and thus should be dropped from topic according to your rule addendum #4) and the superior AA is available and it's far better than what your holy grail card offered.

For which games exactly? Anything worth playing? When I run into something that I have trouble playing that has trouble running on my rig then I'll be forced to look for something else. The latest graphics showcases have been extremely boring titles at best. Gee whiz, another online shooter with nigh identical mechanics to dozens of others..... no thanks.

As for not buying such cards, you purchased a 9800 Pro didn't you?

Yep, POS too.

Of course, which is why I pointed out that real AF and SSAA are generally not viable because of this very reason.

Except they are- kind of blows your whole line of discussion out of the water.

You then responded by showing low detail benchmarks - benchmarks that violated your own rules for both reproduction and playability - to "prove" I was wrong. In otherwords you contradicted yourself while trying to prove me wrong!

Except, and I'll try explaining this more clearly- what I showed you were percentage performance hits that such implementations took when running proper filtering and AA. You could take the performance hit that those benches show and then apply those to the performance level of a 7800GTX SLI setup. Given that the performance difference between 4xS w/AF and 4x MSAA w/o AF on the NV25 tended to be in the less then 20% range(although this did go higher in certain instances) I would tend to say that the 7800GTX in SLI would have no problem at all handling the strain.

Why are you stating the obvious to make it sound I said something else? You know darned well the AA-sampled image is higher than the final image so following your previous simpleton spin on AA you would be forced to come to the flawed conclusion that the concept of AA in general is useless.

No- you left out a few rather important elements. One is the amount of corrective adjustment a 4x AA blend during flip will have on the front buffered image and how the sampling of an analog display will be impacted when trying to display a resolution beyond the pixel density that the screen allows. You went simpleton so I replied in kind. If you want a real answer then deal with the actual issue. Signal theory and diminishing return back me all the way on this.

No, it isn't posisble. For all intents and purposes real AF is not available because the optimizations are locked at the hardware level which means according to your rule addendum #4 (must be available) you won't be pursuing the issue any further.

Did you see that on the moving picture box? Why are you dropping to simpleton BS when you know it is? The AF optimization technique they utilize is hardware based- but why on Earth would you try to imply that that could not be completely disabled and they revert to brute force? Why are you trying to even pretend that this is not completely and easily within their capability? They are fully capable of not only resorting to brute force pure AF they also could manage to do it with FP32 level of precision for their filtering across the entire chip. Now, that isn't viable as of now because the performance hit would be as big as you are making proper AA and AF out to be.

Where did that strawman come from?

From you. If you keep using cumulative then you are firmly standing behind the fact that 40FPS it totally irrelevant. You are brutally misuing basic language.

More specifically what does it have to do with the semantic games you're playing in order to ignore the cumulative performance hit of the two?

Show me a cumulative performance hit. If you are claiming that you are coming remotely close to using cumulative in the proper sense then you are saying that 40FPS does not matter at all. The x850 takes a 90% performance hit when using any AF at all. If I claim that your stating it is less then 10% is simply a matter of semantics and I decided to use 90% as I see fit you would have no problem with that? That is exactly what you are doing.
 

xtknight

Elite Member
Oct 15, 2004
12,974
0
71
Originally posted by: BenSkywalker
Except, and I'll try explaining this more clearly- what I showed you were percentage performance hits that such implementations took when running proper filtering and AA. You could take the performance hit that those benches show and then apply those to the performance level of a 7800GTX SLI setup. Given that the performance difference between 4xS w/AF and 4x MSAA w/o AF on the NV25 tended to be in the less then 20% range(although this did go higher in certain instances) I would tend to say that the 7800GTX in SLI would have no problem at all handling the strain.

But SSAA would depend on texture detail and geometry detail while MSAA would depend on geometry detail still. With either you can't compare the percents to a high detail benchmark unless I'm sorely mistaken.

Edit: NM, SSAA just processes the whole image in one piece so the texture detail doesn't matter, but the MSAA still depends on geometry detail?
 

BenSkywalker

Diamond Member
Oct 9, 1999
9,140
67
91
But SSAA would depend on texture detail and geometry detail while MSAA would depend on geometry detail still. With either you can't compare the percents to a high detail benchmark unless I'm sorely mistaken.

His low detail comment was the benches I linked to topped out @1600x1200 for resolution- unfortunately this is all most sites tested up to until extremely recently.

Edit: NM, SSAA just processes the whole image in one piece so the texture detail doesn't matter, but the MSAA still depends on geometry detail?

Z compares. Geometric edges- not in general geometry.
 

nemesismk2

Diamond Member
Sep 29, 2001
4,810
5
76
www.ultimatehardware.net
Originally posted by: Elcs
Originally posted by: nemesismk2
Is this topic back to talking about the Kyro? Guess not! :(

Yeah, some idiots found this place and decided to make it their personal flame thread.

The Kyro 2 is a legendary card and Id like to see some of the techniques used by the manufacturers used today. Could be fun.

With any luck a mod will view this thread and find out how they have hijacked the topic and ban them.
 

Elcs

Diamond Member
Apr 27, 2002
6,278
6
81
Originally posted by: nemesismk2
Originally posted by: Elcs
Originally posted by: nemesismk2
Is this topic back to talking about the Kyro? Guess not! :(

Yeah, some idiots found this place and decided to make it their personal flame thread.

The Kyro 2 is a legendary card and Id like to see some of the techniques used by the manufacturers used today. Could be fun.

With any luck a mod will view this thread and find out how they have hijacked the topic and ban them.

We can only dream.....

Too much ATI/Nvidia flaming goes down without intervention, nevermind Kyro bashing.
 

BFG10K

Lifer
Aug 14, 2000
22,709
2,997
126
I can run Sac now @2048x1536 w/AA- moving to 7800GTX SLI setup gains me a couple of notches worth of AA- which to me isn't worth the reduction in quality of the filtering.
Reduction of filtering quality to what? Are you saying you run the game at 2048x1536 with 16xAF on a NV25 board?

You are the one being irrational here.
No, you are. Like I said before this thread is nothing more than your typical bash at AF.

When tube amps went out of production an awful lot of people just stopped buying new ones. Why? Tube amps were still available used.
An awful lot of people also like optimized AF so what does the above comment prove?

For which games exactly?
"Real" AF is not available so therefore according to your own rules it doesn't count. What that has to do with "which games" I'll never know.

Yep, POS too.
Woah, woah, woah. So you mean to tell me after all of you holier than thou preaching, d@mn blind and several pages of this ridiculous nonsense you DON'T EVEN OWN/USE an NV25 based board?

hy·poc·ri·sy ( P ) Pronunciation Key (h-pkr-s)
n. pl. hy·poc·ri·sies
The practice of professing beliefs, feelings, or virtues that one does not hold or possess; falseness.
An act or instance of such falseness.

Except they are- kind of blows your whole line of discussion out of the water.
No, they aren't viable and you proved that yourself by not even owning/using an NV25. That's quite a deceptive hypocrisy you've been running in this thread.

What's even more ridiculous is that you'll no doubt produce a completely unrelated counter to the above quote and think your credibility is somehow restored.

You could take the performance hit that those benches show and then apply those to the performance level of a 7800GTX SLI setup.
Except for the little issue of 1280x1024 and lower. Apply those same performance hits at at 2048x1536 with "real" 16xAF and 4xSAA and you'll get quite a different picture.

No- you left out a few rather important elements.
I most certainly did not. First you argued that 16xAA doesn't benefit 2048x1536 because it's outside of the monitor's sampling range yet then you turned around and claimed AA in general is fine despite the fact that it exceeds the monitor's current sampling range (i.e. resolution) by definition.

Signal theory and diminishing return back me all the way on this.
Diminishing returns also back me on optimized AF to which you counter that you'd have to be d@mn blind. In otherwords you simply twist facts whenever they suit you but when they don't you use your opinion like gospel. This is why your entire premise was flawed right from the start and your obtuse arguments to back it up are even worse.

Did you see that on the moving picture box?
See what? You accuse me of having reading comprehension problems yet over half of your replies have absolutely nothing to do what you quoted.

AF optimization technique they utilize is hardware based- but why on Earth would you try to imply that that could not be completely disabled and they revert to brute force?
Uh, because it can't be disabled as it's locked at the hardware level. If you think otherwise point me to a driver or application that that demonstrates real AF on a R100. Until then you are just blowing hot smoke and creating fictional scenarios and treating them like fact.

If you keep using cumulative then you are firmly standing behind the fact that 40FPS it totally irrelevant
Nonsense.

Show me a cumulative performance hit.
Done repeteatedly. If A + B is lower (or higher) than A or B by themselves then A + B is cumulative. That's simple logic.

If I claim that your stating it is less then 10% is simply a matter of semantics and I decided to use 90% as I see fit you would have no problem with that?
That is absolutely nothing like what is happening here and you know it. You are just producing obtuse semantics again.
 

BenSkywalker

Diamond Member
Oct 9, 1999
9,140
67
91
Reduction of filtering quality to what? Are you saying you run the game at 2048x1536 with 16xAF on a NV25 board?

NV25 doesn't support 16x AF.

An awful lot of people also like optimized AF so what does the above comment prove?

A lot of people where happy with solid state amps too- but those that demanded the best purchased used gear until they brought back tube amps.

"Real" AF is not available so therefore according to your own rules it doesn't count. What that has to do with "which games" I'll never know.

It isn't enabled, it is easily available if it was.

Woah, woah, woah. So you mean to tell me after all of you holier than thou preaching, d@mn blind and several pages of this ridiculous nonsense you DON'T EVEN OWN/USE an NV25 based board?

I've already told you in this thread I still have my Ti4200- it is paired with a XP2000 in an Abit NF7S.

No, they aren't viable and you proved that yourself by not even owning/using an NV25. That's quite a deceptive hypocrisy you've been running in this thread.

Except if you would recall what I've alread posted I do.

Except for the little issue of 1280x1024 and lower. Apply those same performance hits at at 2048x1536 with "real" 16xAF and 4xSAA and you'll get quite a different picture.

You think.

I most certainly did not. First you argued that 16xAA doesn't benefit 2048x1536 because it's outside of the monitor's sampling range yet then you turned around and claimed AA in general is fine despite the fact that it exceeds the monitor's current sampling range (i.e. resolution) by definition.

Signal theory- AA itself would be nigh useless on a display with a resolution of 8000x8000(small children may require 12Kx12K with older adults or those with less then optimal vision would require less). If working with resolutions lower then that you need to be able to sample at a high enough rate to eliminate the highest contrast transitional areas on screen. Running 2048x1536 you are above the sampling possibilities for a consumer display and as such aliasing is seriously reduced although there are still areas of high contrast left that can still leave some instances of aliasing on screen. Enabling a relatively decent 4x sampling pattern would eliminate those areas.

Diminishing returns also back me on optimized AF to which you counter that you'd have to be d@mn blind.

Diminishing returns would back you up in terms of adjustments to LOD bias, but not in terms of reducing aliasing as adaptive AF tends to have a considerable increase in texture aliasing over the default settings.

Uh, because it can't be disabled as it's locked at the hardware level. If you think otherwise point me to a driver or application that that demonstrates real AF on a R100. Until then you are just blowing hot smoke and creating fictional scenarios and treating them like fact.

You are trying to be obtuse with that comment. Worse case scenario if they had the hardware locked in some way in which basic filtering wasn't even possible using the default hardware you could easily use the shader hardware entirely to handle sampling and blending. You know that.

Done repeteatedly. If A + B is lower (or higher) than A or B by themselves then A + B is cumulative.

No- cumulative is A+B. What you are talking about is compounded.

That is absolutely nothing like what is happening here and you know it.

You are misusing the word repeatedly. It is a compounded performance hit, not cumulative.
 

ArchAngel777

Diamond Member
Dec 24, 2000
5,223
61
91


Done repeteatedly. If A + B is lower (or higher) than A or B by themselves then A + B is cumulative.

No- cumulative is A+B. What you are talking about is compounded.

That is absolutely nothing like what is happening here and you know it.

You are misusing the word repeatedly. It is a compounded performance hit, not cumulative.[/quote]

Actually, he is not misusing the word. The performance hit is cumulative and could also be said that it is compounded as well. Either way, both uses of the word are correct in common everyday language. Not to mention, how gay is it to be argueing over a word? Your arguement was on "Image Quality" and now it is over a simple English word? :roll:



 

BenSkywalker

Diamond Member
Oct 9, 1999
9,140
67
91
The performance hit is cumulative and could also be said that it is compounded as well.

Cumulative and compounded are not synonyms, they indicate two different things.

Not to mention, how gay is it to be argueing over a word? Your arguement was on "Image Quality" and now it is over a simple English word?

When you are talking about a difference of ~40% in terms of performance then I would say it does matter quite a bit.
 

ArchAngel777

Diamond Member
Dec 24, 2000
5,223
61
91
Originally posted by: BenSkywalker
The performance hit is cumulative and could also be said that it is compounded as well.

Cumulative and compounded are not synonyms, they indicate two different things.

Not to mention, how gay is it to be argueing over a word? Your arguement was on "Image Quality" and now it is over a simple English word?

When you are talking about a difference of ~40% in terms of performance then I would say it does matter quite a bit.


No, I dissagree. But I am not going to argue with you. It is fruitless.

Going to get a mod to lock this, many people have already voiced their irritation for this thread continueing on the way it is. It has not been on topic since page 1 or so.
 

BenSkywalker

Diamond Member
Oct 9, 1999
9,140
67
91
Sounds good- please keep this in mind at any point in the future if you post anything that is remotely off topic in a thread that it will also be reported to a mod :)
 

BenSkywalker

Diamond Member
Oct 9, 1999
9,140
67
91
dont go crying to mods like noobs!

I never have before- although I also fail to see exactly how anything in this thread warrants locking. Since the OP seems to think going off topic is a lockable or bannable offense then we should respect his choice and make sure that the board is moderated as he sees fit in terms of how his actions are handled.
 

blckgrffn

Diamond Member
May 1, 2003
9,501
3,816
136
www.teamjuchems.com
Originally posted by: BenSkywalker
dont go crying to mods like noobs!

I never have before- although I also fail to see exactly how anything in this thread warrants locking. Since the OP seems to think going off topic is a lockable or bannable offense then we should respect his choice and make sure that the board is moderated as he sees fit in terms of how his actions are handled.

:laugh: :thumbsup: :laugh:
 

JonnyBlaze

Diamond Member
May 24, 2001
3,114
1
0
Originally posted by: BenSkywalker
dont go crying to mods like noobs!

I never have before- although I also fail to see exactly how anything in this thread warrants locking. Since the OP seems to think going off topic is a lockable or bannable offense then we should respect his choice and make sure that the board is moderated as he sees fit in terms of how his actions are handled.

i was talking to him more or less anyways. ;)


 

Gstanfor

Banned
Oct 19, 1999
3,307
0
0
My god, what a thread full of rubbish this is.

I suggest Ben actually takes a look at geforce 6 & 7 before proclaiming them to possess rubbish filtering, perhaps you have spent too long in front of ATi products and are confusing the two? BFG10K's rabid ranting certainly isn't helping this thread either.

nVidia have a large range of combined AA modes in their drivers, acessible through the likes of nHancer (your best option as it works with nVidia's profiles and allows you to modify things on an application by application basis) or RivaTuner.

I'm a big fan of 16x AA myself, which is comprised of 4x MSAA and 2x 2 SSAA combined. This mode is certainly not as slow as certain people would love you to believe it might be. I'm currently replaying Vampire: Bloodlines (hardly what most people would call an undemanding game...) with this mode enabled @ 1600x1200x32 and the game flies. Bear in mind I only possess a humble Socket A cpu and a 6800GT - a 7800 series GPU with an Athlon64 wouldn't even blink at the workload going by my experience.

You can see my exact graphics settings for the game in the links below, and this is only one of many, many games (by no means all of them old or undemanding) I run @ 16x AA very comfortably.

http://img323.imageshack.us/img323/8947/enhancementspage0rv.jpg
http://img323.imageshack.us/img323/2664/optimizepage1om.jpg
http://img323.imageshack.us/img323/8103/compatiblepage0iq.jpg

Just as a general headsup for those who don't already know, nHancer has recently been updated to include even *more* functionality than it already had, truly making it a must-have utility for anyone who uses a nVidia card.
http://img274.imageshack.us/img274/6731/nhancer1223mk.jpg

get it from www.nhancer.com
 

BFG10K

Lifer
Aug 14, 2000
22,709
2,997
126
NV25 doesn't support 16x AF.
Nice sidestep; you know exactly what I mean.

I've already told you in this thread I still have my Ti4200- it is paired with a XP2000 in an Abit NF7S.
I can run Sac now @2048x1536 w/AA- moving to 7800GTX SLI setup gains me a couple of notches worth of AA- which to me isn't worth the reduction in quality of the filtering.
But you can't run Sacrifice on a Ti4200 @ 2048x1536 with maxed AF which brings us back to the original issue of the "reduced filtering" comment you made. Again I ask, "reduced filtering" from what, given you clearly aren't referring to a Ti4200?

It isn't enabled, it is easily available if it was.
What sort of semantic nonsense is this? The AF you demand is not available and therefore according to your own rules is not up for discussion. If you want to change your tune again then let's go back to 32 bit 256xAF because you should be lambasting the NV25 for not providing this.

Once again you are changing your argument whenever it suits you.

You think.
Yes I think, just like you. You have no access to a 7800 running "real" AF and neither do I so therefore either of our comments are really just speculation at this point.

Signal theory- AA itself would be nigh useless on a display with a resolution of 8000x8000(
What that has to do with 2048*1536 I'll never know.

Running 2048x1536 you are above the sampling possibilities for a consumer display and as such aliasing is seriously reduced although there are still areas of high contrast left that can still leave some instances of aliasing on screen. Enabling a relatively decent 4x sampling pattern would eliminate those areas.
And running better than 4x will reduce them even further. Whether you claim to see a difference is your opinion. Personally I'd go as high as possible.

Diminishing returns would back you up in terms of adjustments to LOD bias, but not in terms of reducing aliasing as adaptive AF tends to have a considerable increase in texture aliasing over the default settings.
That's your opinion, just like it's your opinion that greater than 4xAA is not needed at 2048x1536.

Worse case scenario if they had the hardware locked in some way in which basic filtering wasn't even possible using the default hardware you could easily use the shader hardware entirely to handle sampling and blending.
In theory. But that's all we have at the moment: theory.

No- cumulative is A+B. What you are talking about is compounded.
Goodness me, this is really getting comical now. This is why threads like this always degenerate into 20 page crap-fests.

cu·mu·la·tive ( P ) Pronunciation Key (kymy-ltv, -y-l-tv)
adj.
Increasing or enlarging by successive addition.

com·pound ( P ) Pronunciation Key (km-pound, km-, kmpound)
v. com·pound·ed, com·pound·ing, com·pounds
v. tr.
To combine so as to form a whole; mix.
To produce or create by combining two or more ingredients or parts: pharmacists compounding prescriptions.

These words are nothing more than synonyms yet only you would go so far as to claim somebody is wrong for using one and not the other.
 

BenSkywalker

Diamond Member
Oct 9, 1999
9,140
67
91
But you can't run Sacrifice on a Ti4200 @ 2048x1536 with maxed AF which brings us back to the original issue of the "reduced filtering" comment you made. Again I ask, "reduced filtering" from what, given you clearly aren't referring to a Ti4200?

Never played Sac I take it. It is playable at that setting on a Ti4200(albeit without AA). The game came out during the GF1 time era.

What sort of semantic nonsense is this? The AF you demand is not available and therefore according to your own rules is not up for discussion.

It isn't available on newly released parts you mean.

Yes I think, just like you. You have no access to a 7800 running "real" AF and neither do I so therefore either of our comments are really just speculation at this point.

The 7800 can handle Z compression up to 2048x1536 w/4x AA which the earlier parts couldn't. All of the information we have available on the part indicates that it would be more efficient at that setting then anything else available, certainly not less then the NV25.

And running better than 4x will reduce them even further. Whether you claim to see a difference is your opinion.

If I can't see- you wouldn't stand a prayer of seeing it. You are the one that was reccomending performance AF on R3x0 based parts with their incredibly obvious mip banding and aliasing.

In theory.

No, that is readily avaiable functionality documentation talking about the G70(or NV4x based parts for that matter).

Goodness me, this is really getting comical now. This is why threads like this always degenerate into 20 page crap-fests.

cu·mu·la·tive ( P ) Pronunciation Key (kymy-ltv, -y-l-tv)
adj.
Increasing or enlarging by successive addition.

com·pound ( P ) Pronunciation Key (km-pound, km-, kmpound)
v. com·pound·ed, com·pound·ing, com·pounds
v. tr.
To combine so as to form a whole; mix.
To produce or create by combining two or more ingredients or parts: pharmacists compounding prescriptions.

These words are nothing more than synonyms yet only you would go so far as to claim somebody is wrong for using one and not the other.

I suppose you never claimed to know anything at all about statistical analysis nor have you ever claimed to have much knowledge on linguistics so perhaps it shouldn't be surprising you can't see the enormous differerence betwee those two definitions. If it makes you feel better-

Cummulative

Compounded

Guess what- they are not synonyms. One clearly indicates that it is an additive function while the other is a mixed function. Every single benchmark that I have been able to find indicates that the performance stacking is a mixed performance issue and none of them indicate it is additive.

I suggest Ben actually takes a look at geforce 6 & 7 before proclaiming them to possess rubbish filtering, perhaps you have spent too long in front of ATi products and are confusing the two?

Spent quite a bit of time checking out the 6 series- they were horrible(worse then my R9800Pro even). They increase aliasing(not quite as much as my R9800) and artifacts(quite a bit worse then my R9800). From what I've seen the 7 series did not improve on this(although I don't have too much time with one as of yet).
 

coomar

Banned
Apr 4, 2005
2,431
0
0
i was wondering about that, cummulative is additive and compounded should have a multiplicative part
 

BFG10K

Lifer
Aug 14, 2000
22,709
2,997
126
Never played Sac I take it. It is playable at that setting on a Ti4200(albeit without AA).
I'd like to see benchmarks to back that claim as even GLQuake would be unplayable at those settings on a Ti4200.

Also you now appear to be claiming to run without AA so one must conclude you must be d@mn blind because aliasing is clearly visible at 2048x1536 with 0xAA. So much for "best possible reproduction" and "mathematically provable". Those statements are an equally big steaming pile of heap as they were when you first made them and the only thing more ridiculous is the charade you're still employing by hiding behind them.

And now that I think about, who even gives a crap about Sacrifice? It's your opinion the game is good which is great and all but again you're passing your opinion as fact as using that to back your claim. A Ti4200 will not run any modern RTS at the settings you claim to run Sacrifice at and in the grand context of this discussion Sacrifice means precisely nothing.

It isn't available on newly released parts you mean.
You can play with semantics or you can sum up that it's been discontinued which means it's not available.

The 7800 can handle Z compression up to 2048x1536 w/4x AA which the earlier parts couldn't.
Certainly but you can bet the architecture is geared around optimized AF. Nobody's claiming the newer cards won't run real AF faster than the old ones. My claim is that the performance hit is still unacceptable especially since optimized AF looks practically as good as real AF but has essentially zero performance hit.

You also seem to have the flawed idea that SSAA is fast so take a look at these. Running the 8xS modes at any reasonable resolution is usually restricted to very old games like Quake 2 (Quake scores are missing but they're similar to Quake 2's FYI).

You are the one that was reccomending performance AF on R3x0 based parts with their incredibly obvious mip banding and aliasing.
Likewise you were the one trying to tell us a middling resolution like 1600x1200 is acceptable and that you don't need anything above 4xSSAA. I certainly didn't make those claims because if I did I'd be d@mn blind.

No, that is readily avaiable functionality documentation talking about the G70(or NV4x based parts for that matter).
Just like there's probably documentation about the theory around higher AF modes. Until you demonstrate an application or driver that promotes viable and widespread use (i.e. not just a lab implementation) then it's not available and it's simply theory.

I suppose you never claimed to know anything at all about statistical analysis nor have you ever claimed to have much knowledge on linguistics so perhaps it shouldn't be surprising you can't see the enormous differerence betwee those two definitions.
:roll:

So let's start from the beginning, shall we?

  • You claimed AF and AA stress different parts of the card and are therefore completely disjointed.
  • I pointed out that your own numbers show the performance hit of the two combined is higher than either mode just by itself and therefore the two methods aren't completely separate like you claimed. Using simple logic if your claim was correct then the performance hit of the two combined would be the same as the higher of the two when used by itself, which wasn't the case.
  • You then claimed that because cumulative and compounded are "different" and because 90% and 10% are different we can ignore the benchmark results and hence your original claim was "correct".
This is a classic example of the obtuse arguments you constantly employ and the example above commits many logical fallacies including the most obvious, non sequitur. Since you've already had a go at my intelligence I'm not going to hold back on this one:

You clearly have no idea how to argue or how to employ even the most basic reasoning or logic.

You twist facts to suit you and when that fails you present your opinion as fact. When your numerous logical fallacies and contradictions are pointed out you then create a completely irrelevant tangent which you then proceed to argue over. Eventually when you get to the end of that irrelevant tangent you somehow manage to "conclude" that because you've argued the irrelevant tangent long enough, that means you were correct in your original claim, a claim that has absolutely nothing to do with said tangent.

Your "arguments" should be mandatory reading for all logic & reasoning students of what not to do.