Higher 3DMark05 scores = Better Equipped for Future Games?

[Scali]

I voted not sure since there are several things (including DST, lack of 3dc, heavy vertex workload) that Futuremark employed that may never be widely realized in games during the life of this benchmark. If the Doom and Source engines rule the playing fields for the next two years then 3dmark05 is utterly useless.

I think your logic is flawed.
3DMark03 already predicted the performance of Doom3/Source. 3DMark05 is one step further. And there is no 'validity interval' of 2 years or anything. Where do you get that 'two years' from?
Even if there is another 3DMark released after this one, before new engines arrive, 3DMark05 could still be useful, if the new 3DMark does not test the exact same things (which it probably won't, the next 3DMark will probably have even heavier workload and use next-gen hardware features).

 

[martrox]

Jesus, Scali...... don't you think you've beaten this horse to death yet?

 

[Scali]

Jesus, Scali...... don't you think you've beaten this horse to death yet?

What is your problem anyway? You keep posting nonsense aimed at me in threads.

 

[DeanoC]

What if in the near future all games are going to use DST (we already have FarCry)?
That would both mean that Futuremark was right on the money once again, and that Beyond3D made a complete arse out of itself, not to mention a lot of people posting on forums.
Let's have another poll in 2 years time or so.

Actually its once again the PC centric view of the world. The consoles dominant the design of most engines (very few are built for PC's), ask yourself how many next gen consoles will use the same techiques as NVIDIA graphics cards. DST is not an important technique merely a quality enhancement for NV cards, all games that use shadowmapping will be tuned to pixel shader filtering NOT texture unit filtering. Nobody will reduce the number of textures samples on NV cards IMHO. The code will simple use a different texture format, using the same shaders.

As I mentioned in another thread, even NVIDIA say the same thing. Where you get 16 taps on ATI and the consoles, NVIDIA will get 64 taps.

Actually the most damning problem with 3DMark2005 is PSM. PSM is fine only in tightly controlled camera situation, something that doesn't apply to most games but does to 3DMark.

Most games will not be using PSM (Only a few shipped games have used it, and I know of NONE that will be using it).

 

[Scali]

Actually its once again the PC centric view of the world. The consoles dominant the design of most engines (very few are built for PC's), ask yourself how many next gen consoles will use the same techiques as NVIDIA graphics cards.

Since 3DMark05 is purely a PC application, do you find it odd that the discussion is PC-centric?
By the way, DST is also available on XBox and I wouldn't be surprised if it got a lot of use there.

DST is not an important technique merely a quality enhancement for NV cards, all games that use shadowmapping will be tuned to pixel shader filtering NOT texture unit filtering. Nobody will reduce the number of textures samples on NV cards IMHO. The code will simple use a different texture format, using the same shaders.

You're already wrong there, since FarCry already uses 3DMark05's approach. So at least some games won't be using DST merely to improve quality. Besides, regardless of the issue being reduced quality or reduced performance, ATi is at a disadvantage.

Actually the most damning problem with 3DMark2005 is PSM. PSM is fine only in tightly controlled camera situation, something that doesn't apply to most games but does to 3DMark.

Most games will not be using PSM (Only a few shipped games have used it, and I know of NONE that will be using it).

This may be true, we don't know. We do know that Carmack is not going to use them. But I think the exact mapping used is mainly relevant to quality, which is not an issue for a benchmark application like 3DMark05.

 

[Dave Baumann]

I agree with B3D just having to publish an article instead of talking nonsense in the forum, I already made that obvious to DaveBaumann, I think he got the message.

Thanks for giving me a message that we had already thought about, Scali - shame you gave it after we'd actually already posted these things publically.

 

[martrox]

What is your problem anyway?

And he wants to know what my problem with him is.......

 

[Scali]

Thanks for giving me a message that we had already thought about, Scali - shame you gave it after we'd actually already posted these things publically.

I was just reminding you. I still don't see the article, and I still don't see an official announcement of B3D leaving the BDP either.

 

[Dave Baumann]

I was just reminding you. I still don't see the article, and I still don't see an official announcement of B3D leaving the BDP either.

Scali, we'll do as we please, and we don't need any reminders, so please desist in these comments - you haven't told us anything we haven't already been thinking about and we haven't made necessarily made any decisions yet. Your constant petty harping will not change that.

 

[DeanoC]

We are talking about future games Futuremark is predicting games in 2006.

I know of no engine thats being designed for the PC in that era. That include Crytek's new one (i.e. Farcry isn't where they are at), Epic's and others.

The stated aim is that this tell you how your PC will play games in 2006. That means they have to write an engine like a games company today. That means they are targeting console first, PC second. If you can find anybody in the industry (except the nice Mr Carmack, who is a different position from most other games companies) to disagree with that statement please tell me.

Your right that DST is big NOW, because of XBox, no other reason (Splinter Cell is the real world example) but its not 2006 now.

 

[Scali]

Scali, we'll do as we please, and we don't need any reminders, so please desist in these comments - you haven't told us anything we haven't already been thinking about and we haven't made necessarily made any decisions yet. Your constant petty harping will not change that.

At least you now (hopefully) changed your mind about 3Dc.

 

[[maven]]

Actually the most damning problem with 3DMark2005 is PSM. PSM is fine only in tightly controlled camera situation, something that doesn't apply to most games but does to 3DMark.

Most games will not be using PSM (Only a few shipped games have used it, and I know of NONE that will be using it).

I personally think that this is not much of a problem, as I think other variants of shadow mapping will be used instead which will have roughly similar performance characteristics.
But slowly I start wondering whether there actually is a variant that does not exhibit any obvious (or that many) pathological cases.

 

[Dave Baumann]

At least you now (hopefully) changed your mind about 3Dc.

Given that 3Dc is already available on the current ATI hardware and ATI is making the next graphics core for the Xbox2, I would suggest that the justification Futuremark have already supplied to us for the inclusion of DST/PCF is likely to hold just as much water with 3Dc.

 

[Scali]

Given that 3Dc is already available on the current ATI hardware and ATI is making the next graphics core for the Xbox2, I would suggest that the justification Futuremark have already supplied to us for the inclusion of DST/PCF is likely to hold just as much water with 3Dc.

Then I take it you really didn't understand my PM. That must also be the reason why you never responded.

 

[Dave Baumann]

Then I take it you really didn't understand my PM. That must also be the reason why you never responded.

Yes, I do understand your PM - that doesn't mean I necessarily agree with it. These are also not the responses we've had from Futuremark over issue as they have just given us justification for the inclusion of DST/PCF that can easily be seen to apply elsewhere - if they wish to give us a more technical response officially then we'll take that on board and discuss it with them (however you should note that they have already agreed that in all likelyhood future versions will use 3Dc).

 

[Scali]

Yes, I do understand your PM - that doesn't mean I necessarily agree with it.

I think it does, actually, in this case.

 

[Dave Baumann]

Why would they agree that they are likely to use it in the future?

 

[Scali]

Why would they agree that they are likely to use it in the future?

Because they will be using an entirely new engine, with entirely new rendering features, entirely new content, running on entirely new hardware? Heck, there is even a chance that they will NOT use DST in their next engine. Just like they do NOT use stencilshadows in this one.

 

[Dave Baumann]

The likelyhood is that the issues you describe are just as likely to be present then as they are in the future, regarless of the desing on the engine.

Regardless though, the discussion is not about 3Dc.

 

[Miksu]

Okey, I just finished 22 pages-thread from the other part of the forums. The discussion here seems to be a direct follower to that now closed thread.

My answer to poll would have been "Yes, somewhat".