看了评测，G80开AA下降的幅度比R580大，原因？

linwu1

原帖由 来不及思考 于 2006-11-9 23:51 发表
楼上，请你首先搞清楚

8800GTX和GTS虽然实际运算能力要大大超过X1950XTX，GTX可以达到近2倍的领先

但显存带宽并没有翻倍
GTX相比X1900XTX仅仅增加了50%不到，GTS更加少

而对于游戏测试来说，在非极限 ...

按你的意思理解，G80的AA其实已经很强了。那么，我是不是可以推论，即便R600再强大，只要它是256bit显存，高分辨率AA就无法跟GTX抗衡？

linwu1

原帖由 来不及思考 于 2006-11-10 00:00 发表


老大，你们都不算频率的么

GDDR4最高可以到2.8GHz !!! :blink:

问题是R600推出时，就可以拿到2.8GHz？
而且G80也可以换GDDR4，
其实我想知道的是G80的AA够不够强，如果只是带宽的限制，那就是比成本，而不是比设计。毕竟我现在看的测试，G80的AA下降比X1950TX大，让我怀疑NV的功力

linwu1

原帖由 来不及思考 于 2006-11-10 00:09 发表


纯粹比AA比带宽的时代，已经过了
以后是FP HDR with AA
Alpha AA with FSAA
Softshadow with AA的时代

G80虽然纯粹的带宽没有提升多少，但在FP Tex filter,ROP和正交Framebuffer上的重大变化足以在 ...

那就没法比了，毕竟R580是上一代的产品，没考虑那么多，肯定不如G80。R600同样是DX10的设计，这些应该都会考虑

cmd

ATI的驱动似乎在2560X1600分辨率的OGL游戏中有bug，(CF模式中？),4XAA会自动降为2XAA？:unsure:


PC Perspective  的疑惑：   http://www.pcper.com/index.php#NewsID-3049

the standard X1950 XTX card falls back into the 2xAA mode even when 4xAA is requested and set in the game.

Whether or not this was an accidental bug or something else, there are probably numerous reviews online about the X1950 XTX cards that need to be re-evaluated

To answer some questions that have come up, the driver version I used to get the numbers above was Catalyst 6.10, default settings in the control panel.  I have tested 6.9 and 6.8 and saw the same issue on these three games using the original X1900 XTX and X1900 CrossFire Edition cards as well

AMD(ATI)的回应： http://www.rage3d.com/index.php?cat=75#newsid33872662

ATI Responds to OpenGL Performance Bug
Nov 7, 2006 at 03:49 PM by Ratchet14 Comments   
ATI has responded to the high-resolution OpenGL performance issues on Radeon graphics cards reported by PCPer last week:
Hi all,

As highlighted on page 27 of the GeForce 8800 reviewers guide, and reported on last week by PC Perspective, I wanted to provide everyone with an update on the topic of ATI Radeon X1900 XT, X1900 XTX, and X1950 XTX cards dropping to 2xAA when 4xAA is set at 25x16 in OpenGL games.

The driver team has been exploring this topic over the weekend and has discovered that there is a bug in the memory allocation for the X1900 XT and XTX and X1950 XTX that is not affecting the CrossFire Edition cards. The slave cards currently allocate memory for two monitors (i.e. dual display). The CrossFire Edition cards only allocate memory for one monitor. The bug in our driver is such that the total memory allocated is based on the CrossFire Edition (master) cards and thus higher AA resolution settings fail for slave cards and the driver brings the application to a lower AA level.

For background on the topic of memory allocation, driver teams at both AMD and our competitor have several options they’re presented with when a card runs out of memory in a given application. They can choose to use system memory rather than the on-board memory buffer and accept low frame rates, they can choose to have the driver drop to a lower AA setting to maintain playable frame rates, or they can choose to not run the application at all (i.e. the driver closes the application and provides an error dialogue box).

In essence, the second option is what you’re seeing with the slave cards today, although as demonstrated by the CrossFire Edition cards, there is sufficient memory to run at 25x16 4xAA and the AMD driver team is working on that fix. Another example of this, is looking at the X1900 XT 256MB card at 25x16 4xAA in “Call of Duty 2” vs. a competing 7900 GT. At this setting, the 7900 GT provides a video memory error and doesn’t allow the app to run.

Regarding the ATI slave card memory allocation error, we have identified the code in our driver that causes this bug and will be fixing this in an upcoming Catalyst release.

Thanks,
Will Willis
Senior PR Manager
Advanced Micro Devices