Unreal Engine 3 on R420 at GDC

[PatrickL]

Exactly. None of the specs were posted by independant and reliable sites. For now they are at best speculations.

 

[Sabastian]

Dear god, this thread has degenerated horribly.

 

[Evildeus]

Anyway, at leat we will see in a few weeks how both wards behave on present games. 8)

I'm still waiting for the links where i can find games more "cinematic" on Nv cards than Ati cards, Doomtrooper. Till know, i thought it was the contrary :?

And if you fill like on Nvnews, then you can stay over there i think

PS: This thread is dead it seems

 

[surfhurleydude]

Exactly. None of the specs were posted by independant and reliable sites. For now they are at best speculations.

I'd say that some of the specs have been "confirmed" by some fairly reliable people, however (for both nVidia and ATi). As far as those UT2004 AA screen shots go, I think the pictures speak a thousand words and does confirm that nVidia has changed their AA methods for the better... And that by itself provides for a GOOD outlook on the nVidia side of things.

 

[{Sniping}Waste]

Thank you to all that had info about the two systems in Epics booth in GDC.

From the info given we have one system with a NV40 and another with a R420 in it. Both systems look like about the same specs using A64 in them with the same mem and SATA raid for hard drives. The demo ran at 1024x768 and ran choppy on the NV40 and ran smooth on the R420.

From the info given it looks like the R420 might be faster then the NV40 but take this in mind this is the first info of the two cards running the same thing at the same time and looks like ATI wins here.

We have to wait to get better benches to see what is faster but most of all what is better to get a better picture.

Mark Rein, thank you for posting here and hope if time allows that you will post here more offen.

 

[Doomtrooper]

As far as those UT2004 AA screen shots go, I think the pictures speak a thousand words and does confirm that nVidia has changed their AA methods for the better... And that by itself provides for a GOOD outlook on the nVidia side of things.

I think you overlooked probably the best member on this forum for AA modes:

http://www.beyond3d.com/forum/viewtopic.php?t=10946&postdays=0&postorder=asc&start=400

I don't think it's as simple as 4xOGMS 9tap either

It is ordered grid though, it does have a blur and on low angle edges it's worse than a 4xRGAA.

You'd probably get a result like that if you used quincux then shrunk the result by 75% or something (bilinear filter in photoshop or whatever)

Sometimes people see only what they want to see

 

[surfhurleydude]

I stated that nVidia has changed their AA modes for the better. The link you have provided does not in any way, shape or form render my statement invalid...

Edit:
After quickly re-glancing through a number of pages in that topic, I'd suggest you read after page 21 as well

 

[kemosabe]

Calling SpellSinger....

If this thread is to stay alive in any meaningful way, you're going to have to build up an extra ounce of courage and set the record straight. Were you at the GDC? Did you in fact see Epic's engine running on either NV40 or R420 or both? Did you actually get a gander at the R420 board layout? Was there an iota of truth to what you said, or do you retract all of your statements

 

[demalion]

Do you know that Nv paid more than Ati to be able to do so? That's how you do to have adds, in case you don't know.

As we all know, the TWIMTBP program invlolves more than 'ads' including not allowing cards certain visuals base off device ID detection..even though it is more than capable.

The TWIMTBP program can involve device ID detection. I hate it for that and other reasons. But that's its own topic.

Discussion like this has broken down into two extremes that I see:

"These observations being shot down prove the exact opposite and the R420 is slower than the NV40, and Epic can't be criticized so as to more conveniently jump to that conclusion."

"These observations are an inconvenient reality of the R420 being much faster than the NV40, and Epic is guilty of everything possible with TWIMTBP because that is more convenient for dismissing anything someone from Epic says when they show up to shoot this down."

Here's the thought this conversation avoids: "TWIMTBP has issues, you might have reason to criticize Epic for it" is a different topic than "someone from Epic showed up to shoot down these statements because they were misrepresentative", and can't be arbitrarily associated as proof of whichever you prefer to be the case of the above.
Whether the NV40 or R420 is faster, and when, is a unique discussion, and it simply seems like the answer isn't found here...no need to stretch things to such a massive degree to insist there is proof otherwise.

 

[Evildeus]

Thank you to all that had info about the two systems in Epics booth in GDC.

From the info given we have one system with a NV40 and another with a R420 in it. Both systems look like about the same specs using A64 in them with the same mem and SATA raid for hard drives. The demo ran at 1024x768 and ran choppy on the NV40 and ran smooth on the R420.

From the info given it looks like the R420 might be faster then the NV40 but take this in mind this is the first info of the two cards running the same thing at the same time and looks like ATI wins here.

I'm sorry, but appart from the first post, i didn't find any news on R420 demonstration . The only confirmation i've is:

The ATI board was running the demos much smoother than the NV board and at a higher resolution.

This is also complete bullshit.

Could you provide me with a link? Thank you.

 

[Dave Baumann]

My understand is that Epic was pleased with the engine performance on all the demo systems they had on display, at this early stage in the development.

I think this thread has run its course for the time being. If Spellsinger, Mark or Daniel has more to say please Private Messate or mail (wavey@beyond3d.com) me and we'll see about reopening the thread to present your case.