DeltaChrome nitro

[anaqer]

...I don't see any reason a new card should be getting half the framerate of a Ti4200 in it.

Except for the immature OpenGL drivers, maybe...?

 

[KimB]

1. We are not guaranteed significant performance increases with future driver releases.
2. Who can say whether it is a driver-related issue or just an inefficient architecture?

 

[Fodder]

1. We are not guaranteed significant performance increases with future driver releases.
2. Who can say whether it is a driver-related issue or just an inefficient architecture?

Unreal 200x screams along at a very competitive pace, and I doubt the hardware is only capable of running the one game. It smacks of driver issues to me.

 

[StealthHawk]

1. We are not guaranteed significant performance increases with future driver releases.
2. Who can say whether it is a driver-related issue or just an inefficient architecture?

Unreal 200x screams along at a very competitive pace, and I doubt the hardware is only capable of running the one game. It smacks of driver issues to me.

Yes, but the drivers may be optimized up the wazoo for UT200x. And S3 may or may not spend the same time optimizing for other applications.

 

[Fodder]

Yes, but the drivers may be optimized up the wazoo for UT200x. And S3 may or may not spend the same time optimizing for other applications.

I was more referring to his second point, to show that the hardware is actually capable of performing well. I do find it strange though that they haven't concentrated just as hard on Quake3, given how widespread that engine is.

 

[vnet]

Unreal 200x screams along at a very competitive pace, and I doubt the hardware is only capable of running the one game. It smacks of driver issues to me.

Application specific optimization, anyone?

 

[demalion]

It is unfortunate that so many people still use "cheat" and "optimization" interchangably in such discussions, especially when the difference is at the heart of the issue being discussed.

 

[Hyp-X]

It is unfortunate that so many people still use "cheat" and "optimization" interchangably in such discussions, especially when the difference is at the heart of the issue being discussed.

I agree.
Eg. an UT2003 specific optimization is not necessarily a cheat.

An IHV might analize the game find the typical bottlenecks and tweak the card to perform better at those situations.
However enabling such optimizations might actually reduce performance, or even worse cause bugs in other applications.
So detecting the game is a valid way to improve performance.

OTOH, if all an IHV do is optimize for a few popular games and their performance suck at everything else than it's very bad for the end user.
I'm happy that many sites use a wide variety of games to benchmark a card, so cards like DC and Volari can show their real "values".

 

[demalion]

I agree completely.

I'm also quite interested in looking at some CPU test results (3dmark) for DeltaChrome systems on various specification systems compared to other IHVs, and wondering if Rightmark's profiling benchmarks might be a useful tool for some insights.

 

[Ailuros]

Unreal 200x screams along at a very competitive pace, and I doubt the hardware is only capable of running the one game. It smacks of driver issues to me.

Application specific optimization, anyone?

No Mr. VolariNet (most likely a very weak/unoptimized OGL ICD).

 

[anaqer]

1. We are not guaranteed significant performance increases with future driver releases.
2. Who can say whether it is a driver-related issue or just an inefficient architecture?

1 ) You are not guaranteed ANYTHING regarding the DC. It's a pre-release board with pre-release drivers FFS.
2 ) I fail to see why the architecture happens to be relatively competitive in DX but at the same time so utterly flawed in OpenGL. This looks like a driver issue if I ever saw one... but if you can come up with an explanation for how an API change cutting back performance so badly could be a fault in the silicone itself - well, I'm listening.

 

[StealthHawk]

Yes, but the drivers may be optimized up the wazoo for UT200x. And S3 may or may not spend the same time optimizing for other applications.

I was more referring to his second point, to show that the hardware is actually capable of performing well. I do find it strange though that they haven't concentrated just as hard on Quake3, given how widespread that engine is.

That may or may not be something we want to assume. I'm not saying S3 is cheating, as we have no evidence or reason to believe they are, but it's not beyond possibility. Just because the drivers score well in one app doesn't mean that is a good reflection of the power of the hardware.

 

[ben6]

When is DC going to actually hit retail?

 

[vnet]

end of 2004