What FSX settings when testing for Q3 2008 ?

Gerard

Distinguished
Apr 25, 2004
38
0
18,530
When looking at the graphics cards chart Q3 2008 I was surprised that my 8800 card was a top of the bill card for FSX.

I expected that the 260 and 280 cards would perform much better.
Faster core speed, much faster memory speed.

As I have a 26 inch monitor ( 1920 x 1200 ) and a high end computer , I have set almost all sliders to max.

At a flightsimulator forum there was a discussion about this.
Some people stated that probably medium settings where used in FSX and therefor the true performance of a faster card would not show.

Could someone tell me what settings where used when doing the tests.

 

CompuTronix

Intel Master
Moderator
Gerard,

Ultra Quality was used for each resolution / AA / AF combination. It's right up front. If you missed it, then you're grounded because you're too blind to read the check list :lol: - http://www.tomshardware.com/charts/gaming-graphics-charts-q3-2008/benchmarks,30.html

As long as you have at least a mid-range 512 MB nVidia card, don't get stuck on graphics. FSX is on the extreme end of the CPU bound spectrum, and as a global view simulation, frame rates do not follow the "more GPU is better" rule to which gamers are typically accustomed. The highest frame rates are achieved by highly overclocked quad cores, cache and memory.

Comp :sol:
 

CompuTronix

Intel Master
Moderator
zipzoomflyhigh,

There's nothing wrong with the charts. If you study the topic, you'll eventually understand that FSX doesn't behave at all like a first person shooter limited view game; it's a global view simulation, so you must rethink the program dynamics.

SLI runs slower than a single card because while the CPU cores are maxed out doing all the physics crunching required to prepare global views for rendering, the GPU interrupts from 2 graphics cards require more CPU clock cycles, which has the effect of slowing the CPU cores down more than 1 graphics card. Just overclocking 1 graphics card can have a negative effect on frame rate. If you overclock or underclock the graphics card, it has little noticable effect on frame rates, but if you overclock or underclock the CPU, it has an extreme effect on frame rates.

There are no games which are 100% GPU bound or 100% CPU bound. Certain game titles typically have a certain balance between GPU and CPU bindings, however, FSX is definately on the extreme end of the CPU bound spectrum at perhaps 85% CPU / 15% GPU.

Comp :sol:
 

Gerard

Distinguished
Apr 25, 2004
38
0
18,530
Thanks for responding , but my question was how the FSX settings where :

- scenery complexity
- autogen complexity
- ai traffic
- clouds
etc etc


And then I would also be curious about the testing rig.

If , I don't know, it is a mid end I can understand that a faster card won't make any difference.
On the other hand with a high end system it might be the other way round.
 

CompuTronix

Intel Master
Moderator
Gerard,

I answered your question in my first post and provided a link for proof. Ultra Quality settings were used in each instance.

I've built many different FSX rigs of varying horsepower with many different CPU / GPU configurations, and test flew them till I was bleary eyed time and time again, and I'm here to tell you that Tom's charts are correct. Because FSX is so extremely CPU bound, this is why the frame rate curve is so flat in the charts from high-end nVidia cards to mid-range nVidia cards, as compared to other "games". I can pull my 8800 GTS 512 out, pop in an obsolete PCIE 7600 GT 512, and it's very difficult to notice the difference. Conversely, if I drop the overclock on my Q9650 from 4.2 Ghz down to 2.4 Ghz, the frame rate goes right down the toilet.

Shadow,

No, FSX does not support AGEA hardware or driver level physics support.
 

CompuTronix

Intel Master
Moderator
It means whatever the "Ultra Quality" settings cause the sub-settings and sliders to default to, accross the board, all display tabs. Utra Quality settings do not, by default, set AI traffic to "full Maxed".

EDIT: I remember reading the "Test Setup" specifications for the rig which was used for Tom's graphics testing. I've searched and can't find it again, however, to the best of my knowlegde, the processor was a 3.0 Ghz stock clock E8400.