Aquamark3 What about those techniques?

K

Klux

Newcomer
Recently they release technical documentation about Aquamark 3 at their official site www.aquamark3.com. The pdf gives some really interesting insights to some pretty cool techniques.

But honestly I do not understand all of what I read. What is the difference between the SVIST technique and the OVIST technique? What exactly is SVIST for (at least the screenshot of this technique looks interesting).

Any ideas over here? :?:
 
Klux said:
Recently they release technical documentation about Aquamark 3 at their official site www.aquamark3.com. The pdf gives some really interesting insights to some pretty cool techniques.

But honestly I do not understand all of what I read. What is the difference between the SVIST technique and the OVIST technique? What exactly is SVIST for (at least the screenshot of this technique looks interesting).

Any ideas over here? :?:
Click to expand...

SVIST marks the areas of the scene which use pixel shader effects. By the color you can decipher if it's using PS2.0 or PS1.X shaders (or if it's not using any at all)

OVIST simply colors the overdraw levels. Red for an example is an area where there's many overlapping objects.

Maybe I'm just not positive enough but I honestly don't see the need for SVIST. As long as all graphics cards use the same shaders (which they should unless someone is cheating or you're comparing a DX8 to a DX9 board) the test is basically useless IMHO. (Or does someone see a reason why it could be interesting except for curiousity?).

OVIST might be interesting to see how efficient various boards HSR might be though.
 
I got it :D :D :D

Maybe I'm just not positive enough but I honestly don't see the need for SVIST. As long as all graphics cards use the same shaders (which they should unless someone is cheating or you're comparing a DX8 to a DX9 board) the test is basically useless IMHO. (Or does someone see a reason why it could be interesting except for curiousity?).
Click to expand...

If SVIST works like you described it I think it is in fact a cool feature! I had a look at the technical document again and found some interesting statement. They claim that it is not 'nessecary or recommended to implement the pixelshader in the highest availible shading version' and (even more interesting) that they 'implemented a test where they !!count!! the pixels rendered through a specific shader version'. I conclude that if they count the pixels of each shader they have something like a statistic for the benchmark where they have a percentage of pixels for each shader version. If they really have this ... awesome! I would be like: Aquamark 3 uses 50% pixel shader 2.0 30% pixel shader 1.x 20% no pixel shader! I do not know of any benchmark or game at all which exposes or shows :) this to the user. :LOL:

Now I'm wondering why anybody would do 3000 runs with the AUTT technique :oops: this probably would kill your system after some days :?
 
Looking at just one engine from one benchmark with SVIST just gives you more accurate information about what the one benchmark is doing. Depending on the method and what it detects exactly (i.e., whether it color codes based on, for example, the precision of rendering on the screen, or color codes based on what the application tells the drivers to draw on the screen...I'm a bit unclear after glancing at the document), it might range from simply being a convincing placebo giving a false sense of security to the user (unless they recognize and emphasize that if fails to be effective in measuring what the cards actually do), to being an informative cheat detection method (if they manage to succeed in measuring what the card actually does), AFAICS. I'd hope their goal is the latter and they tried to avoid the mistake of only achieving the former, so it should be a good thing.

I'm guessing maybe the lack of emphasis (maybe I just missed it?) of "cheat detection" is an effort to avoid legal entanglements with words like "cheat" involved in the test, and just provide more information for people to evaluate for themselves while pushing the boundaries of benchmarking forward a bit more.

Klux: I noticed you say the word "synthetic" like it was inherently bad ("Sure it is not a synthetic benchmark like 3DMark but IMHO it is not comfortable enough." when referring to Gunmetal). I ask you to consider that this SVIST testing methodology concentrates on making the test more synthetic, not less...what game color codes based on differing shader levels? A synthetic test either conveys useful information, or does not, to some degree of accuracy....the synthetic part is to highlight the characteristic being measured for the user more effectively, when in a game situation many factors would work together. Being based on a game engine does not inherently affect that or make a benchmark more useful except as far as the one specific game engine...the general usefulness is determined solely by the implementation decisions. Game/synthetic labelling of benchmarks to determine relevance just sidesteps evaluating the validity of the implementation decisions, please don't make that mistake.
 
I don't really see why it'd be a good cheat detection method.

I think its primary uses would be for the developer, to prioritize performance optimizations, and to the artist, to meet planned development limitations (similar to a triangle count limitation).

It would also be beneficial as a diagnostic or benchmarking tool: why is card X performing so poorly in scene A?
 
What I would like to know:

1) WHEN will it be released

2) What exaclty will the Aquamark Result Comparator be? They give a hint at the end of the technical document about this and I think this is quite exiting because it will be ONLINE :D

3) Do they think that Aquamark 3 is better than 3DMark03?

4) What is their position when it comes to 'cheating' in drivers?

5) Do they work together with graphic card manufacturers like ATI and NVIDIA?

6) Will Aquamark 3 be free or is there a charge?

That's what I would like to know :D
 
AquaMark 3 WILL be made publicly avaliable, which I wasnt sure of, and the guru3d.com is going to be a public download mirror.
20030702-02.jpg
20030702-03.jpg
 
This preview is obviously a preview of a leaked version! Guru3D already sent news that Massive tracked down the source of the leak!
 
So ... we have to be careful with this preview. I'm anticipating Aquamark 3 but this preview seems to be a little bit too positive. We just have to be patient until we can evaluate it on our own and share our thoughts about it. I think it would be bad if some illegal preview would create a hype or too high expectations. I just want to have a new good benchmark.
 
True, but scince when do we hype other than as a clearly designated poll or speculation about the future?

One of the things I find so refreshing here is the clear analysis of facts to an amazing level until the truth pops out - something few other sites do so well.
 
I would like to return to the question: when do we see those "interesting techniques" in games? :?:
So far even aquamark 1 level has not been reached quite often.

In a way this makes me wonder if these benchmark/demo views are even possible? We had alpha-textures used with the NVIDIA creature - and we have the same technique in StarWars: Galacies wookiees... but somehow it looks laaame. Not up to the demo level. Surprise?
 
TKorho said:
I would like to return to the question: when do we see those "interesting techniques" in games? :?:
So far even aquamark 1 level has not been reached quite often.

In a way this makes me wonder if these benchmark/demo views are even possible? We had alpha-textures used with the NVIDIA creature - and we have the same technique in StarWars: Galacies wookiees... but somehow it looks laaame. Not up to the demo level. Surprise?
Click to expand...

Didn't Aquanox looked like Aquamark and vice versa? So what about Aquamark 3 and Aquanox 2? If I look at the Aquanox 2 screenshots it looks very similar to what I saw from Aquamark 3. Even if you compare the Aquanox 2 in game footage and the Aquamark 3 footage it does look similar. Interestingly there is a Dx9 patch for the German Aquanox 2 available since end of last year. I assume that Aquanox 2 is like Aquamark 3 and vice versa. Probably the benchmark puts more stress on the hardware. It is very nice that Aquamark 3 runs on Dx7,Dx8 and Dx9 hardware which is IMHO better than 3DMark03. This is typical for a game engine nowadays. I think it is a good move that they create a benchmark beased on a game engine.
 
Klux said:
Probably the benchmark puts more stress on the hardware. It is very nice that Aquamark 3 runs on Dx7,Dx8 and Dx9 hardware which is IMHO better than 3DMark03. This is typical for a game engine nowadays. I think it is a good move that they create a benchmark beased on a game engine.
Click to expand...

Whats your fasination with this being "better than 3DMark03"? Its almost like you're trying to push a line. Auqamark is one benchmark, 3DMark03 is another - they both have their merits and inevitably their downsides.
 
whql said:
Klux said:
Probably the benchmark puts more stress on the hardware. It is very nice that Aquamark 3 runs on Dx7,Dx8 and Dx9 hardware which is IMHO better than 3DMark03. This is typical for a game engine nowadays. I think it is a good move that they create a benchmark beased on a game engine.
Click to expand...

Whats your fasination with this being "better than 3DMark03"? Its almost like you're trying to push a line. Auqamark is one benchmark, 3DMark03 is another - they both have their merits and inevitably their downsides.
Click to expand...

I totally agree. I really like 3DMark03 and I do use it a lot. And I like Aquamark 3 at least the things I know about it. However there is a tendency that even Futuremark admits that their benchmarks may be of a different kind in the future. Many of the things said feel more like Aquamark 3 than 3DMark03. So I'm really looking forward to get Aquamark 3 on my system and judge for myself which benchmark I will use in the future. Chances are high that I will use both.
 
Sorry to take this thread OT, but can anyone point me to a working link for Aquamark 2? I'd like to take a look first-hand, now that I have a DX8 card.
 
Pete said:
Sorry to take this thread OT, but can anyone point me to a working link for Aquamark 2? I'd like to take a look first-hand, now that I have a DX8 card.
Click to expand...
Dude, we are now in the DX9 ages.

:LOL:
 
You must log in or register to reply here.

Similar threads

S
Discussion about lighting (usage of lighting techniques).
2
Replies
21
Views
2K
Frenetic Pony
F
I
The first Evil Within still has remarkably sophisticated lighting
Replies
2
Views
734
Phantom88
P
D
Hardware implementation of threading models in contemporary GPUs
Replies
17
Views
3K
Lurkmass
L
Rys
Reopening Architecture and Products + rules and guidelines changes
3 4 5
Replies
90
Views
14K
Shifty Geezer
Shifty Geezer
L
What is Forwards Compatibility? *spawn*
Replies
15
Views
2K
Jwm
J
Back
Top