Curse

moriz · Aug 22, 2007, 07:44 PM // 19:44

so, today i tested out the performance between the two... the results are a little odd.

i've been testing them against GW with these graphics settings:
terrain: low
reflections: off
textures: high
shadows: off
shaders: low

GW intro screen:
the GMA beat the FX5500 by a good 8fps. if i turn shaders on high, the GMA wins by more than 10fps. GMA: 31-33fps. FX5500: 21-22fps

GW character screen:
here's where it gets really weird. the FX5500 tops the GMA by 10fps. GMA: 15-17fps. FX5500: 23-25fps

GW guild hall:
the trend continues, with the FX5500 consistenly 10fps higher. GMA: 15-25fps. FX5500: 25-40fps

GW alliance battle:
well, you can probably predict what happens here. GMA: 9-14fps. FX5500: 15-22fps

------

so... why the performance differences? it's not like the FX5500 is better always, just under certain conditions.

lordpwn · Aug 22, 2007, 09:03 PM // 21:03

The Geforce FX series is listed as the game's minimum recommended Nvidia card because it was the first series from the company that supported Shader Model 2.0 pixel and vertex shaders.

However, the card's performance actually drawing something using those shader effects is poor; it was heavily criticized for it back in 2003 even though very few games used those back then.

My guess is - the new intro screen uses shaders so heavily the FX5500's "Achilles' Heel" kills the performance. Other in-game situations involve fewer shader effects and the generally slow integrated junk from Intel can't compete with a proper dedicated graphics card there.

moriz · Aug 22, 2007, 09:24 PM // 21:24

yep, the aurora effect on the intro kills my card.

however, why would the GMA perform better on the intro screen? isn't it even worse in terms of shader support?

or is it simply because the GMA has a higher clock... 333mhz vs 270mhz

lordpwn · Aug 22, 2007, 09:48 PM // 21:48

The GMA performs better because the complex shader performance of GeForce FX cards was a total fiasco for Nvidia back in 2003, and the newer GMA chips are at least designed with modern shaders (if not the highest performance) in mind, that's all.

Comparing the clock rates of completely different pieces of hardware is meaningless. Don't do that.