ATi & nVidia Anti-Aliasing Performance

[Leonidas]

Wäre wichtig. Und bitte mehr als nur 2 Leute.

http://www.3dcenter.org/artikel/ati_nvidia_aa_performance/

[Lost Prophet]

ich denk ich schmeiss mich mal an die erste seite, nachdem ich mir zuerstmal jetzt den artikel reinzieh...

cya, axel

hab jetzt angefangen, steck ein bisschen mit "samplemuster". ist "subpixel-mask" das selbe??

[Lost Prophet]

Introduction

Both ATi and nVidia like to speak of their anti-aliasing solutions as of something miraculous - for example, that they are the best one available on the market, or that they'd work with (virtually) no performance-loss. Both companies have a very different approach: While ATi uses three Multisampling modes, nVidia works with - depending on the driver - with up to 10 different modes, of which most are a combination between Super- and Multisampling.

In most hardware-reviews, only few of these are benched, simply because there isn't enough time for a more in-depth look. We want to make up for that with that article, which will exclusively look at the anti-aliasing performance of current ATi and nVidia graphic-chips. Doing so, our emphasis will be on the performance of the more rarely-benched anti-aliasing modes in comparison to the lower anti-aliasing modes as well as a general view on the overall-performance.

Before actually benchmarking, looking at the differences in image quality of the tested modes is essential, to see whether they are comparable. Unfortunately, no modes between ATi and nVidia really are. Even with 2x Anti-Aliasing, where ATi and nVidia use a similar-quality subpixel-mask, ATi has the advandage of gamma-corrected anti-aliasing (http://www.3dcenter.de/artikel/Multisampling_anti-aliasing/index6.php) to offer, which nVidia is incapable of with its current GeForceFX-generation. Hence, on a close look, no anti-aliasing modes between ATi and nVidia are directly comparable. This has to be kept in mind for the following performance-comparisons.

In advance we have to note, that all chips in this article were tested on principle - hence also in tests without anti-aliasing - with 8x (nVidia) and 16x (ATi) anisotropic ffilter. This happened solely on the reason of making the article its-self more realistic. The thought behind that is, that one who switches on anti-aliasing on such a highend graphic-chip, will as well choose to add the superior texture-quality achieved through anisotropic filtering - already because today's highend-chips have no problem to do so. Of course we could have benchmarked anti-aliasing alone, without the anisotropic filtering, and through that have achieved a possibly preciser result concerning anti-aliasing. But we think of the measuring with anisotropic filter as much closer to reality, because in most cases on highend graphic-cards both features are enabled.

The basis for ATi and nVidia does not change through the enabling of anisotropic filtering anyway, because both of the mainly tested cards, - Radeon 9800 Pro 256MB and GeForceFX 5900 Ultra 256MB - achieve a nearly identical performance under this setting. Thus, this article will not measure the theoretical anti-aliasing performance (for which flyby-benchmarks in Unreal Tournament 2003 would probably be best), rather concentrate only on the anti-aliasing performance in day-to-day gaming, where switching on the anisotropic filter should be standard already.

For a more detailed look in theory about anti-aliasing we recommend our previous articles link and link

ATi Anti-Aliasing Modes

With ATi the question about existing anti-aliasing modes of the R3x0 and RV3x0 Graphic-chip generation (Radeon 9500/9600/9700/9800) is easy to answer: The Canadian graphic-chip developer offers 2x, 4x and 6x gamma-corrected Multisampling ind Direct3D as well as in OpenGL. ATi uses the following subpixel-masks and reaches image qualities as seen on the right, measured by the FSAA-Tester (Click on Thumb-nail opens bigger picture)

Modes
Comments
Subpixel-mask
Image quality

without AA
2x Multisampling rotated grid EER 2x2 gamma-corrected
4x Multisampling rotated grid EER 4x4 gamma-corrected
6x Multisampling manyally optimised sparse grid EER 6x6 gamma-corrected

To sum up , ATi offers good anti-aliasing possibilities, where every mode was tried to get the maximum of efficiency through intelligent subpixel-masks out of. In theory, even the sample-positions of the ATi anti-aliasing are programmable, but this possibilty doesn't exist in the Windows-drivers of ATi so far. At most, there is an 8x mode lacking, since between 4x and 6x anti-aliasing there is only a marginal gain, and it is not until 8x anti-aliasing that there is a major jump in image quality again.


_________________________

ich hoffe es passt einigermassen, bitte umbeding wer korrektur lesen, ich hab mir in einigen faellen ein bisschen freiheit genommen, aber immer ATi, nVidia und anti-aliasing geschrieben wenn leo das auch tat.... :D

ich selbst bin mit der uebersetzung nicht wirklich zufrieden, hab aber momentan keine zeit noch laenger daran zu arbeiten. es ist umm.... schlampig uebersetzt, bzw weniger uebersetzt als geschrieben. ich wette ich habe unzaehlige satz-zeichen vergessen und gross-klein-schreibung verschi**en.

so far, so g00t.

cya, axel

ps. sorry, zu faul fuer links...

[Leonidas]

Wo ist der Rest des Teams? CannedCaptain, wo steckst Du?

[mapel110]

ich kann morgen mal einiges machen. sag dann noch bescheid.

[CannedCaptain]

hiho, mein Dasein fristete ich die Woche in der Zivildienstschule Braunschweig und konnte dementsprechend nicht ans Netz.

Ich werde mal ein paar Leutz zusammentrommeln!
Am Sonntag fange ich mit der Fortsetzung an. Es wäre nett, wenn einer der Gurus (z.B. Zeckbag) Korrektur liest.

[zeckensack]

Grmpf :)

Ich schaue mal, wie weit ich heute noch komme.

[zeckensack]

Seite eins, von Purple_Stain, erste Lesung
______
##Titel: ATi & nVidia anti-aliasing performance
Anti-aliasing quality and performance compared while using anisotropic filtering

Introduction

Both ATi and nVidia like to present their anti-aliasing solutions as something miraculous - for example, that they are the best available on the market, or that they'd work with (virtually) no performance hit. Both companies use very different approaches: While ATi uses three multisampling modes, nVidia employs up to 10 different modes - depending on driver version -, of which most are combinations of supersampling and multisampling.

In most hardware reviews only few of these are benchmarked, simply because there isn't enough time for a more in-depth look. We want to make up for that with this article, which will exclusively look at the anti-aliasing performance of current ATi and nVidia graphics chips, while also keeping an eye on the performance of these modes. While doing so, our emphasis will be on the performance of the more rarely benchmarked anti-aliasing modes in comparison to the lower anti-aliasing modes as well as a general view on the overall performance characteristics.

Before actually benchmarking, we'll have to look at the differences in image quality of the tested modes, to see whether they are comparable. Unfortunately, between ATi and nVidia none really are. Even with 2x anti-aliasing, where ATi and nVidia use qually good subpixel masks, ATi has the advandage in offering gamma corrected anti-aliasing (http://www.3dcenter.de/artikel/Multisampling_anti-aliasing/index6.php), which nVidia is incapable of with its current GeForceFX generation. Under closest scrutiny no anti-aliasing modes between ATi and nVidia are directly comparable. This has to be kept in mind for the following performance comparisons.

In advance we have to note that throughout this article, all chips were tested with 8x (nVidia) and 16x (ATi) anisotropic filtering respectively - also in tests without anti-aliasing. This happened solely on the reason of making the comparison itself more realistic. The idea was that if you switch on anti-aliasing on such a highend graphics chip, you will likely also choose to enhance texture quality with anisotropic filtering - simply because today's high end chips have no problem to do so. Of course, we could have benchmarked anti-aliasing alone without the anisotropic filtering, and this way could have achieved a possibly more precise result concerning anti-aliasing. But we think measuring with anisotropic filter is much closer to reality, because in most cases on highend graphics cards both features are used together.

The basis for ATi and nVidia does not change through the enabling of anisotropic filtering anyway, because both the cards we primarily tested - Radeon 9800 Pro 256MB and GeForceFX 5900 Ultra 256MB - achieve nearly identical performance under this setting. So this article will not measure the theoretical anti-aliasing performance (for which flyby benchmarks in Unreal Tournament 2003 would probably be best), rather concentrate only on the anti-aliasing performance in day-to-day gaming, where switching on the anisotropic filter should be standard already.

For more details on anti-aliasing theory we recommend our previous articles link and link.

ATi anti-aliasing modes

With ATi's R3x0 and RV3x0 generation of graphics chips, the question about existing anti-aliasing modes is a quick one to answer: the Canadian graphics chip developer offers 2x, 4x and 6x gamma corrected multisampling in Direct3D as well as in OpenGL. ATi uses the following subpixel-masks and reaches image qualities as seen on the right, measured by the FSAA-Tester (clicking on a thumbnail opens a larger image)

Modes
Notes
Subpixel grid
Image quality

without AA
2x multisampling rotated grid EER 2x2 gamma-corrected
4x multisampling rotated grid EER 4x4 gamma-corrected
6x multisampling hand-optimised sparse grid EER 6x6 gamma-corrected

To sum it up, ATi offers good anti-aliasing choices, where every mode was optimised to get the maximum effect through intelligently chosen subpixel grid. In theory, the sampling positions of ATi's anti-aliasing are even programmable, but this possibilty doesn't exist in the Windows drivers so far. The only thing we might complain about is the lack of an 8x mode, since because of the rule of diminishing returns this would be required to get a significant increase in image quality versus the 4x mode while the quality gains to be had by using 6x versus 4x are rather small.

_________________________

Zu Anfang kennzeichnen Bold-Tags alle meine Änderungen, das wurde mir bei den vielen kleinen Fehlern (Bindestriche, Gross-/Kleinschreibung) aber dann doch zu mühselig :bäh:

repeat after me:
high end graphics chip

[zeckensack]

Seite 2 - fertig

nVidia anti-aliasing modes

There are a lot more anti-aliasing modes on NV3x (GeForceFX 5200/5600/5700/5800/5950) and - using the right drivers and tools - the NV2x family (GeForce3 /Ti and GeForce4 Ti). Available anti-aliasing modes may differ from one driver version to the next, even with GeforceFX cards. We used driver version 45.23 which obviously isn't the most recent one, but it offers all anti-aliasing modes we wanted to use.

As nVidia never exposed all possible anti-aliasing modes through the control panel, and because the Californian chip designers sometimes rename modes, we'll stick to the naming scheme used by our own aTuner. nVidia uses the following sampling grids and achieves the respective image qualities, again determined with the FSAA Tester tool (clicking opens larger images). We skipped the "Quincunx" and "4x 9 tap" modes as usual, because they additionally blur the image.

<Tabelle>
Mode | Notes | Sampling grid | Image quality

No AA | -
1x2 | Supersampling, ordered grid, EER 1x2
2x1 | Supersampling, ordered grid, EER 2x1
2x | Multisampling, rotated grid, EER 2x2
4x | Multisampling, ordered grid, EER 2x2
4xS | Hybrid, partially rotated grid, EER 2x4 (Direct3D only)
4xSS | Supersampling, ordered grid, EER 2x2 (Direct3D only)
6x | Hybrid, ordered grid, EER 3x2 (Direct3D only)
6xS | Hybrid, paritally rotated grid, EER 3x4, may produce artifacts (Direct3D only)
8x | Hybrid, ordered grid, EER 4x2 (Direct3D only)
8xS | Hybrid, partially rotated grid, EER 4x4
12x | Hybrid, ordered grid, EER 3x4 (Direct3D only)
16x | Hybrid, ordered grid, EER 4x4 (OpenGL only) | this mode is OpenGL only, so there's no comparison screenshot

</Tabelle>

The multitude of nVidia's anti-aliasing modes unfortunately doesn't imply that there are more choices that actually make sense. The Quincunx and 4x 9-tap modes are thrown out due to the additional blurring, the latter mode has also been removed from the series 50 drivers. Also dubios is the 6xS mode because it tends to produce artifacts. The 6x mode has been removed from the drivers while 6xS with its issues remained, but note that nVidia call this mode "6x" in the control panel.

Also missing from the series 50 drivers is the 8x mode. This isn't a big loss because the mode isn't competitive in it's smoothing effect due to its ordered sampling grid. That's in contrast to the 8xS mode, which, ironically, nVidia expose in the series 50 drivers instead of the old one, under the same "8x" name (for OpenGL there has only been the 8xS mode anyway). New additions to series 50 are the (rather uninspiring) supersampling modes 1x2 and 2x1.

The 16x anti-aliasing mode for OpenGL has only been in the drivers for a short time, even with driver version 45.23 we used here it didn't work anymore. This isn't a huge loss either because, as we are going to see, even 8xS puts tremendous strains on the nVidia cards and the further improvement in anti-aliasing quality that could have been achieved with the 16x mode would only have led to frame rates completely inadequate for gaming.

So we're left with the following sensible anti-aliasing modes on nVidia hardware: 2x, 4x, 4xS, 4xSS, 6x, 8x, 8xS and 12x, where the 8x mode is only supported by series 40 drivers. Out of these modes, 4xS, 4xSS, 6x, 8x and 12x are also only available for Direct3D.

[zeckensack]

Seite 3 - fertig

Quality comparison

What we've outlined so far about the technical specifications of ATi's and nVidia's respective anti-aliasing modes only serves one purpose: to show that these modes cannot be directly compared. The edge quality of the various modes is simply too different, particularly so with 4x anti-aliasing. That's why we won't draw any such direct comparisons when doing our benchmarks, but we'll rather benchmark all anti-aliasing modes we can and then look how it all stacks up.

We'd like to apply a certain quality preference, even though it is, of course, according to our own subjective perception of quality. The following chart presents a relative ordering of both ATi's and nVidia's anti-aliasing modes according to their smoothing quality. 100 per cent would be a theoretical peak that only something on the order of 64x anti-aliasing could hope to achieve. The quality valuation is based on both technical specs (EER and sampling grid) and the comparison screen shots we've already shown (for nVidia we're again exclusively using aTuner (http://www.3dcenter.org/atuner/index_e.php)'s naming scheme):

<Tabelle>
Mode | Ati | nVidia
<restliche Tabelle aus dem Original kopieren>
</Tabelle>

The influence of gamma correction on anti-aliasing, that's exclusive to the ATi chips, has been disregarded for this comparison. It's hard to express the effect in terms of edge smoothing quality, but it's, of course, an additional advantage for ATi's anti-aliasing modes. To a certain extent, nVidia's supersampling hybrid modes can counter this effect, because they automatically enhance texture quality (while multisampling strictly affects edges). But then, this effect wouldn't be right to place in the above chart either. It doesn't have anything to do with edge smoothing quality, and that's what we're looking for (it's rather an appreciated byproduct).

For the performance comparison between ATi's and nVidia's anti-aliasing modes, we ended up picking the following: ATi 2x, 4x and 6x, and nVidia 2x, 4x, 4xS, 8xS and 12x. We also benchmarked nVidia's 8x mode, but as its quality doesn't compare favorably to 4xS while its performance is significantly worse, these numbers didn't make it into the comparison. They are, however, present in the appendix (http://lin.k).

Performance comparison

We used the same test system as we did here (http://www.3dcenter.org/artikel/r9500+9700+9800_vs_gffx5200+5600+5800/zi_testumgebung.php), the only difference is that we've installed DirectX 9.0b. We've used Catalyst 3.6 for ATi cards and Detonator 45.23 for nVidia cards. Both aren't exactly the most recent versions, but as performance increases in later drivers are largely constrained to shader limited benchmarks (which we don't use here), we hope to get away with this.

We used Aquanox 2, Comanche 4, Dungeon Siege, Max Payne, Serious Sam: The Second Encounter, Star Trek: Elite Force II, Unreal Tournament 2003 and Wolfenstein: Enemy Territory for benchmarking. Initially we planned on using No One Lives Forever 2 and Splinter Cell as well but had to skip these, because some of the anti-aliasing modes we wanted to use didn't work with these games. As noted in the introduction, throughout all benchmarks, we've used 16x anisotropic filtering on ATi cards (full trilinear has been forced with rTool (http://www.3dcenter.org/downloads/rtool.php)) and 8x anisotropic filtering on nVidia cards.

The main contenders were a Radeon 9800 Pro 256MB and a GeforceFX 5900 Ultra 256MB. To compare 128MB versus 256MB we added a Radeon 9800 Pro 128MB and a GeforceFX 5900 128MB. For cross comparison to ATi's and nVidia's respective lesser performance cards we also added a Radeon 9700 128MB and a Geforce FX 5800 128MB. Initially we wanted to have a more comprehensive list of cards, full results of (almost) every high end card from ATi and nVidia are in the appendix.

So we don't want to step through single benchmarks and instead skip directly to the overall performance. For this purpose we simply averaged all benchmark results generated with the aforementioned eight games. Because none of the games in our suite are high fps games (100+), this should be sufficient. For the two Direct3D-only anti-aliasing modes offered by nVidia (4xS and 12x), we only took the Direct3D games for calculating the average (5x Direct3D, 3xOpenGL).

__________________


Leo,
Der letzte Satz ist mir im Original schon zu unklar formuliert, ausserdem hat's da IMO noch zwei Fehler:

"Bei den nVidia Anti-Aliasing Modi, welche allein unter Direct3D zur Verfügung stehen (4xS und 12x), wurde der Unterschied der Performance unter Anti-Aliasing Performance gegenüber jener ohne Anti-Aliasing rein der Direct3D-Spiele auf alle Test-Spiele (5x Direct3, 3x OpenGL) hochgerechnet."

=>

"Bei den nVidia Anti-Aliasing Modi, welche allein unter Direct3D zur Verfügung stehen (4xS und 12x), wurde der Unterschied der Performance unter Anti-Aliasing gegenüber jener ohne Anti-Aliasing rein der Direct3D-Spiele auf alle Test-Spiele (5x Direct3D, 3x OpenGL) hochgerechnet."

Wäre auch schön, wenn du mir nochmal erklären könntest, was du damit nun eigentlich sagen wolltest :D

[zeckensack]

Seite 4 - fertig

Performance comparison (cont'd)

This is what we got in 1024x768:

<Büld>

This chart should be obvious enough to almost ignore the (following) single performance figures. Anti-aliasing performance characteristics clearly differ between the ATi and nVidia graphics chips: while all ATi chips - including the comparatively modestly clocked Radeon 9700 128MB - stand their ground and lose very little performance even with 6x anti-aliasing, the nVidia chips can only keep pace up to a maximum setting of 4x anti-aliasing.

With higer anti-aliasing modes the nVidia cards take a real dive - certainly a hint towards the bad "effect to cost ratio" of nVidia's hybrid modes incorporating supersampling when compared to the pure multisampling modes (up to 4x on nVidia chips). This is where nVidia's chances are getting slim. When put up against the higher 4xS, 12x and 8xS modes, ATi's comparable 4x and 6x modes are in places significantly faster than the GeforceFX 5900 Ultra 256MB - even a GeforceFX 5950 Ultra 256MB won't be able to change that.

<Tabölle>
Ø of all games
<Röst der Tabölle kopierön>
</Tabölle>

While comparing graphics cards with 128MB vs 256MB we couldn't find anything of substance, the average performance of the respective contenders didn't deviate much at all. The only thing that reacted to more memory in a meaningful way was Unreal Tournament 2003 in higher modes of anti-aliasing, both on ATi and nVidia hardware:

<Tabälle unveränd0rt aus dem Original kopieren>

The clock speed differences between cards should be taken into account for this comparison: the Radeon 9800 Pro 256MB's memory clocked is higher by 10MHz than on the Radeon 9800 Pro 128MB (we can probably safely disregard this), while the GeForceFX 5900 Ultra 256MB at 450/425MHz has a 50MHz higher core clock than the GeForceFX 5900 128MB at 400/425MHz. This obviously doesn't explain the differences in performance we observe here - this must surely be the larger memory on the 256MB models. As we noted, this only happened with Unreal Tournament 2003, in 1024x768 to boot.

[zeckensack]

Seite 5 - fertig

Performance comparison (cont'd)

For the record we repeated the benchmarks in 1280x960 (Aquanox 2, Max Payne, Serious Sam: The Second Encounter, Unreal Tournament 2003) and 1280x1024 (Comanche 4, Dungeon Siege, Star Trek: Elite Force II, Wolfenstein: Enemy Territory), again with 8x (nVidia) or 16x (ATi) anisotropic filtering, which yielded the following:

<Bild>

In general these are the same characteristics as we've already seen, for both ATi and nVidia, but at a lower overall performance level. We find it interesting to note that the "old" GeForceFX 5800 128MB with its 128 bit memory interface shows a pattern that's no different from the GeForceFX 5900 128MB with its 256 bit memory interface.

<Tabelle>
Ø of all games
<Rest wieder kopieren>
</Tabelle>

Again there aren't any big differences between cards with 128MB and 256MB, even though the differences between the two versions of the Radeon 9800 Pro seem to be based on its larger memory, not on its faster (by only 10 MHz) memory. And once again we have to make an exception for Unreal Tournament 2003, which showed significant differences:

<Tabelle kopieren>

None of the other games in our selection reacted to increased graphics memory sizes, not even in 1280x960 or 1280x1024.

[zeckensack]

Seite 6 - fertig

Conclusion

We have to name ATi the clear winner here: without anti-aliasing and with 2x anti-aliasing all chips perform similarly, or according to their relative prices. Once we go above 2x anti-aliasing the fun is over for the nVidia chips: while these are spending ever increasing amounts of performance for small gains in edge quality, the ATi chips are leaps and bounds ahead with their 4x and 6x modes, in both performance and edge smoothing quality.

Using the performance figures of the Radeon 9800 Pro 256MB and the GeForceFX 5900 Ultra 256MB we can make the following comparison which presents the edge smoothing quality of various anti-aliasing modes alongside their respective performance penalties (for both smoothing and performance: higher is better; the performance percentage is normalised to the performance without anti-alising, 1024x768 and 1280x960/1024 were averaged):

<Tabelle>

| ATi | nVidia
Mode | smoothing | performance | smoothing | performance
<Rest kopieren>
</Tabelle>

This chart clearly demonstrates ATi's anti-aliasing leadership: with all of ATi's anti-aliasing modes, the performance hit is well justified by the achieved edge smoothing quality. With nVidia this only applies to the lesser modes 2x, 4x and 4xS, with these the performance hit fits the quality gain. We can't claim that for the higher modes of anti-aliasing at all, 12x and 8xS reduce performance far too much for their respective quality gains, they are quite inefficient.

nVidia's problem is, none of the efficient 2x, 4x and 4xS modes can compare to ATi's "middle of the road" mode 4x when it comes to edge smoothing quality. That means nVidia is only competitive in lower to medium quality anti-aliasing modes and falls short on higher quality modes, sometimes significantly so. As soon as higher anti-aliasing modes are required, ATi clearly edges out nVidia in both performance and image quality terms.

So currently ATi offers clearly better anti-aliasing than nVidia - at least in the high end segment where you can actually use 4x anti-aliasing or more in practice. nVidia would need to offer dramatically higher raw performance than ATi to make up for the difference in anti-aliasing - due to the current near-parity in "normal benchmarks" this seems out of the question.

This conclusion only theoretically applies to the mainstream segment, of course, as it's populated by cards that generally lack the performance for any anti-aliasing beyond 2x. And, as we have observed, at 2x anti-aliasing ATi and nVidia offer equal edge smoothing with only small differences in performance. The overall message - ATi's clear leadership in anti-aliasing - thus in practice only applies to high end cards, because that's what you need if you want to generally (be able to) use any more than 2x anti-aliasing.



In fact, the dramatic results we gathered have even surprised us: most hardware reviews on the web currently suggest that ATi and nVidia are about equal - also concernig anti-aliasing. This might stem from the fact that anti-aliasing modes with equal names are often compared directly - e.g. ATi's 4x anti-aliasing versus nVidia's 4x anti-aliasing.

But we may only compare two things if they are of equal value, in this case that would mean equal image quality or at least equal edge smoothing. The naming conventions for ATi's and nVidia's antialiasing modes alone, however, do not make a comparison valid, the quality offered by two anti-aliasing modes going by the same name can indeed be very different. In particular: testing ATi's 4x anti-aliasing versus nVidia's anti-aliasing is not a comparison with equal image quality, it's not even close. If you use image quality or the amount of edge smoothing as a basis for comparisons you end up at very different conclusions - as we did.

Of course there's a reason for this humiliating defeat of nVidia: since the GeForce 3 nVidia didn't work on anti-aliasing much at all. The GeForce4 Ti's Accuview anti-aliasing is just a very slight enhancement - and the GeForceFX did in fact add nothing to these technologies (yet it did gain some modes). The plethora of new modes offered on the GeForceFX are all driver controlled and are available and just as functional on older cards, down to the GeForce 3, if you're using a tweaking tool such as aTuner (http://www.3dcenter.org/atuner/index_e.php).

So when it comes to anti-aliasing, nVidia is still in 2001, the year the GeForce3 was launched. The next generation of graphics chips must change this, as ATi are far ahead already with the R3x0 series and they will certainly try to keep the heat turned up. We can only hope that nVidia will return the favor of more intelligent sampling grids, higher multisampling modes and gamma correction for anti-aliasing with the NV40, without removing the sometimes handy supersampling/multisampling hybrid modes from the drivers.

__________________________


Leo,
unter der Voraussetzung dass du bald mal wieder
1)einen Artikel schreibst der Anti-Aliasing thematisiert,
2)mindestens 20 mal "AA" darin vorkommt
und
3)keinmal "Anti-Aliasing" (oder orthographische Derivate[TM]) darin vorkommt,
wäre ich bereit, dir eine Tüte Erdnüsse zu schenken :freak:

[zeckensack]

Seite 7

Appendix: benchmarks

All benchmarks were run with 16x anisotropic filtering on ATi cards (full trilinear has been forced with rTool (http://www.3dcenter.org/downloads/rtool.php)) and 8x anisotropic filtering on nVidia cards:

____________________

Seite 8 ff

Appendix: benchmarks (cont'd)


_____________________


Done.

[CannedCaptain]

Was soll man da sagen? 1a Leistung Zeckensack!!! Du brauchst Dich aber nicht immer allein opfern!

[Lost Prophet]

Original geschrieben von zeckensack
Zu Anfang kennzeichnen Bold-Tags alle meine Änderungen, das wurde mir bei den vielen kleinen Fehlern (Bindestriche, Gross-/Kleinschreibung) aber dann doch zu mühselig :bäh:

repeat after me:
high end graphics chip

high end graphics chip :D

hm, wahrscheinlich war das korrigieren mehr arbeit als es selbst zu machen ;(
naja, jedenfalls ist es genau das was ich mit schlampig meinte
ich war jedenfalls mit leos HighEnd-Graphikchips verwirrt, vor allem weil graphicchips mit den 2 "c" bloed aussieht und highend-graphic-chips mit 2 bindestrichen auch nicht besser wird.
+ ich hab viel von der grossschreibung am anfang uebernommen und spaeter vergessen zu korrigieren.
aber genug der rechtfertigung, ich bin ... uhm ... enttaeuscht.

:up: 4 z-bag

cya, axel

[Lost Prophet]

warum ist der artikel noch nicht online?
oder andere frage: was spricht dagegen den artikel online zu stellen?

cya, axel

[Leonidas]

Original geschrieben von zeckensack
Wäre auch schön, wenn du mir nochmal erklären könntest, was du damit nun eigentlich sagen wolltest :D



[Summe D3D-Spiele mit 12xAA] / [Summe D3D-Spiele ohne AA] * [Summe aller Spiele ohne AA] = Summe aller Spiele mit 12xAA

[Leonidas]

Original geschrieben von zeckensack
Leo,
unter der Voraussetzung dass du bald mal wieder
1)einen Artikel schreibst der Anti-Aliasing thematisiert,
2)mindestens 20 mal "AA" darin vorkommt
und
3)keinmal "Anti-Aliasing" (oder orthographische Derivate[TM]) darin vorkommt,
wäre ich bereit, dir eine Tüte Erdnüsse zu schenken :freak:



Ähm, warum?

[Leonidas]

Original geschrieben von Purple_Stain
warum ist der artikel noch nicht online?
oder andere frage: was spricht dagegen den artikel online zu stellen?




Ich war ein wenig ko. Nun aber: Online. Thx @ all, speziell Z-Bag.

[KillerCookie]

Hallo,
da leonidas nicht antwortet (PN!) schreib ichs halt hier rein: ich will mitglied des übersetzungsteams werden! was soll ich machen? bitte helft mir doch :)...

MfG Jason

[zeckensack]

Original geschrieben von Jason15
Hallo,
da leonidas nicht antwortet (PN!) schreib ichs halt hier rein: ich will mitglied des übersetzungsteams werden! was soll ich machen? bitte helft mir doch :)...

MfG Jason Mach einfach mit wenn das nächste mal ein Artikel zur Übersetzung ansteht :)
Es wird dich niemand abhalten (ich jedenfalls nicht).