New DX8.1 benchmark: Codecreatures

[Doomtrooper]

Maybe it's because some developers feel that there is more to DX8 or 8.1 than "mere" shading. DX8 provides for several strictly-performance related features compared to its predecessors. This could be more important for developers and I wouldn't blame them. Can't be close minded, y'know...

I TOTALLY agree but when the developer starts pimping Pixel Shaders as their main selling point of their engine, then labels it Dx 8.1 well maybe I'm just a silly old laymen but I read that as DX 8.1 pixel shader revisions of 1.3 and 1.4 not last years version of Dx8.0 Hardware.

I find it strange people state that there will never be a demo with support for 1.3 or 1.4. Yet Madonion did it last year for specific shader version, yet I see no one seeing the trend I'm getting at. I don't see what the difference is in this scenario, one year ago if I wanted to see the nature DEMO I was forced to buy a certain card which we did, I see no reason why the same can't be done here.

 

[Nappe1]

Let me ask just plain theorical question...

If we would have a card with 4 Vertex and 4 Pixel Shader units using VS 1.1/PS 1.3 and Enough memory bandwidth to run that beast on it's best, how much higher scores we could be able to get?

Let's Assume that chip runs with clock speed of 275Mhz and it has 4 Pipelines and 4 texturing units on each.

 

[PVR_Extremist]

If a benchmark tested DX8.1 to the full then strictly speaking it would be more of a system test than a video card test.

A full DX benchmark needs to test DirectSound, DirectPlay DirectInput etc etc

Just being silly....I know what you're all getting at...

 

[SlmDnk]

Is that 512MB recommendation really a necessary one?

I mean I have:

Abit KT7A, VIA 4.38
Athlon Thunderbird @ 933Mhz
256MB SDR DRAM @ 133Mhz
Creative 3DBlaster 4 Titanium 4400 128MB DDR, Detonator 2832
Win98SE

...and the results I get:

1024x768x32 No AA = 24.0FPS Average
1024x768x32 4x AA = 15.7FPS Average
1024x768x32 No AA = 25.8FPS Average [300Mhz/650Mhz]

You have posted results that have been achieved with 512MB DDR memory and Ti 4600 and the score isn't that much better. Difference is only about +4FPS.

So what is that 512MB min recommendation all about? Makes me wonder...

--

http://www.crytek.com/

On April 15 we might get yet another benchmark/demo combo from Crytek Studios....

 

[OpenGL guy]

Is that 512MB recommendation really a necessary one?

You have posted results that have been achieved with 512MB DDR memory and Ti 4600 and the score isn't that much better. Difference is only about +4FPS.

So what is that 512MB min recommendation all about? Makes me wonder...

+4 FPS sounds quite significant when you are talking ~24 FPS. That's roughly 17%. If you were talking ~100 FPS then +4 FPS would be insignificant.

Most times, minimum memory recommendations are made to take other system limitations out of the equation. Thus, you can't fairly compare your results with someone who had 512 MB because of other limiting factors in your setup.

 

[Doomtrooper]

Ok I got my reply from Codecult checking my email tonight, as I suspected this demo is NOT using PS 1.4 or PS 1.3.

Hi XXX,

thanks for your interest in the Codecult benchmark.
The benchmark uses PS 1.1 the engine itself utilizes 1.4 and future versions.
The Codecreatures-demo which was shown on NVIDIA´s GeForce 4 Ti -Launch used Pixel Shader 1.3.

I hope that will answer your questions. If you´re interested in more information about the Codecreatures Game Developement System, please get in touch with our licensing and business-development diretor Stefan Heinemann (stefan.heinemann@codecult.com).

Best regards

-Silke Wengemann
pr- & marketing-manager

codecult - to create and to protect -

Codecult software research & development GmbH ~ Josef-Haumann-Str.10 ~ 44866 Bochum

Phone +49-2327-997178
Fax +49-2327-997188
eMail silke.wengemann@codecult.com

www.codecult.com www.codecreatures.com
-----Ursprüngliche Nachricht-----
Gesendet: Freitag, 5. April 2002 23:55
An: silke.wengemann@codecult.com
Betreff: Codecult Benchmark
Wichtigkeit: Hoch

 

[LeStoffer]

...
The benchmark uses PS 1.1 the engine itself utilizes 1.4 and future versions.
The Codecreatures-demo which was shown on NVIDIA´s GeForce 4 Ti -Launch used Pixel Shader 1.3...

Jeezz, these guys are really fooling around with this benchmark.

1) The engine itself utilizes PS 1.4, but the benchmark don't. If the benchmark is not based entirely on the engine, then what the hell is it based on?

2) They must have put PS 1.3 support into the engine in order to make the nVidia GF4 demo. But again the benchmark doesn't use anything but PS 1.1. What gives?

What good is a benchmark if it's not based on the very engine that people can license? Huh?

 

[Entropy]

This is not a benchmark. It's a demo.

But if you are trying to launch a new 3D-engine, and want to attract attention and create a buzz, releasing a demo is pretty much useless. However, if you release a benchmark and make it really heavy on the system, you may get a lot of review sites to not only report on its' existance, but perhaps actually use it in their reviews of other products. This gives your demo a enormously better coverage, and may even create the impression of significance.

Entropy

 

[Doomtrooper]

This is not a benchmark. It's a demo.

But if you are trying to launch a new 3D-engine, and want to attract attention and create a buzz, releasing a demo is pretty much useless. However, if you release a benchmark and make it really heavy on the system, you may get a lot of review sites to not only report on its' existance, but perhaps actually use it in their reviews of other products. This gives your demo a enormously better coverage, and may even create the impression of significance.

Entropy

Entropy your first sentence is not correct, look at the 1st line of the reply,
thanks for your interest in the Codecult benchmark. They are marketing this as a benchmark and not a demo.

From my understanding this engine has the capability to be optimized for any Pixel Shader version including the upcoming 2.0.

 

[Reverend]

Wow, Doomtrooper's real name is XXX.

Why so secretive?!?

 

[Humus]

...
The benchmark uses PS 1.1 the engine itself utilizes 1.4 and future versions.
The Codecreatures-demo which was shown on NVIDIA´s GeForce 4 Ti -Launch used Pixel Shader 1.3...

Jeezz, these guys are really fooling around with this benchmark.

1) The engine itself utilizes PS 1.4, but the benchmark don't. If the benchmark is not based entirely on the engine, then what the hell is it based on?

2) They must have put PS 1.3 support into the engine in order to make the nVidia GF4 demo. But again the benchmark doesn't use anything but PS 1.1. What gives?

What good is a benchmark if it's not based on the very engine that people can license? Huh?

It doesn't need to mean that the benchmark isn't based on the engine, it just means that they didn't use those features in the benchmark even though the engine supports it.

 

[jb]

This is not a benchmark. It's a demo.

But if you are trying to launch a new 3D-engine, and want to attract attention and create a buzz, releasing a demo is pretty much useless. However, if you release a benchmark and make it really heavy on the system, you may get a lot of review sites to not only report on its' existance, but perhaps actually use it in their reviews of other products. This gives your demo a enormously better coverage, and may even create the impression of significance.

Entropy

Maybe,

but look at Croteam, didnt they do both with the release of the First Serrious Sam Test? I thought that was playable as well as having away to benchmark it. Darn forgot now ....

 

[Entropy]

Maybe,

but look at Croteam, didnt they do both with the release of the First Serrious Sam Test? I thought that was playable as well as having away to benchmark it. Darn forgot now ....

Only Serious Sam is a game. Croteam included a way of benchmarking their game along with their test-version. (Which makes excellent sense btw.)

This benchmark however does NOT relate to any application whatsoever. It is built using an engine which is used by noone. Even if they manage to find licensees for the engine, it would take quite a while for an application to show up on shelves. Will they find takers? Will the takers produce product? When will the product be available? Will it have a significant market impact? Will the engine, at that point in time, be accurately modeled by this benchmark? Will the application use the engine in a way that is accurately modeled by the benchmark?

I think it should be glaringly obvious why the Codecult benchmark is totally useless. The only purpose it serves is as a vehicle for Codecult to get exposure for their new engine. Nothing wrong with them trying to get some attention, but lets get real about the significance of this as a benchmark. It's zilch.

Entropy

 

[jb]

Yea my bad. Should have had coffee frist I did not think before I typed this morning...doh!

 

[Doomtrooper]

Wow, Doomtrooper's real name is XXX.

Why so secretive?!?

I didn't think my name was of Real Importance as the content of email is what I wanted to post but If you wish I can put it in.

 

[Sabastian]

It would seem to me that if the benchmark is indeed a DX 8.1 benchmark that it would include a bench for PS 1.4. I realise that there is much more to DX8.1 then just PS 1.4 but from what I have heard about PS 1.4 is a substantial developement. Why wouldn't they include PS 1.4 in the benchmark of DX 8.1 when PS 1.4 is an important aspect of DX 8.1? It kind of sounds like a partial DX 8.3/4 benchmark! 8)