View Single Post
Old 07-10-08, 11:21 PM   #60
Registered User
Join Date: Apr 2006
Posts: 14
Default Re: The right price for the GTX200 family...


I happened to run into this thread. Haven't been here for a long time but this has been one of the most mysterious observations I've had over G92 (and to some extent G80).

When I got a HD 3850, believe it or not, I felt as if it was a better card than 8800 GT. Yes, 8800 GT can do a little more detailed settings in games, but the performance was nowhere near as consistent as HD 3850. I didn't purchase the HD 3850 for gaming, so I was pretty much shocked. (I've used 8800 GTX, 8600 GT, 8800 GTS 320, 8800 GT SLI)

That my personal experience and glowing reviews encouraged me to purchase a HD 4870, and I'm totally impressed. Having a 30" LCD I was very concerned with its memory size prior to purchase, but at least for the games I play the experience has been better than any NV setup that I've had.

I do understand what you're trying to say with regard to absolute amount of frame buffer. But the problem with NV cards is as if, for whatever reason, they hit their own memory limit before they get to that absolute amount of memory required. If you could explain to me, for example, why I can play Oblivion @2560x1600/8AA with a HD 4870 but 8800 GTX has a hard time handling (plus stuttering) 2560x1600/4AA even with x1.5 times of RAM.

Following your explanation, obviously GTX 280 will have a better chance to deal when a game/setting require more than 512MB of RAM than a HD 4870. But at the same time, looking at how HD 4870 handles many games, I'm thinking:

- Even today, there just aren't many games that needs such amount of RAM. (Exception: Crysis), or
- When the amount exceeds 512MB, AMD cards handles it a lot better than NV cards, or
- For some reason NV cards tend to run out of memory even when the *theoretical* amount of RAM needed doesn't exceed 512MB.

I've been genuinely curious about the dynamics of bandwidth vs frame buffer. I'd like to ask how the two interact. You yourself explained above that AMD's new architecture handles MSAA with fewer passes than NV's. Does that have anything to do with either bandwidth or frame buffer? If so, doesn't that explain why NV cards need more RAM or bandwidth than AMD cards given a same situation?
lopri is offline   Reply With Quote