Resource Utilization, LOMAC vs SF2E

[+Fubar512]

I recently ran a little experiment using the resource utilization logging tool built into Riva Tuner. It was an eye opening experience.....

 

The is admittedly a worst-case scenario for SF2E. Every video option maxed-out (except horizon distance, that was set to "normal"), CAP off the runway with JSF_Aggie's textures, a whole flight of the Mirage Factory's MiG-29As, Stary's 4-Season GermanyCE terrain, on high detail, Wide-sky with broken clouds at dusk, etc. These are the results.... With a 512 MB 9600GT, SF2E had to "borrow" an additional 260 MB of RAM to make up for a deficit.....notice that the processor utilization was around 60% during game-play (the maxed out portion occurred during the loading process)

 

 

Now, compare that to a title that many consider a resource hog, LOMAC-FC... (Default A-10 mission)

 

 

Holy Crap! Other than a bit more CPU-utilization, LOMAC was amazingly easier on the graphics sub-system.

[Spinners]

What have we learned from that? You are not comparing apples with apples!

[+Fubar512]

What have we learned from that? You are not comparing apples with apples!

 

Gee, I was hoping that you'd tell me....

[+Brain32]

Holy Crap! Other than a bit more CPU-utilization, LOMAC was amazingly easier on the graphics sub-system.

 

A simple fact I've been talking about for a while now

 

So ok the recent dll fix shut me up for a while(still waiting for DX10 shader fixes btw.), but by default this game is incredible resource hog, you'd need an octal SLI with hexa-core CPU to run it smoothly I guess...

[jomni]

All I care about is that SF2E works in my low end laptop while LOMAC does not.

Useless talking about maximising graphics when it won't work on certain rigs in the first place.

[JG7_X_MAN]

Well I don't think we have learned anything we didn't know before: LOMAC has more eye candy than SF2E and thus it will tax your system more.

[jomni]

Well I don't think we have learned anything we didn't know before: LOMAC has more eye candy than SF2E and thus it will tax your system more.

 

The irony is that when maxed out, it is SF2E who taxes your system more than LOMAC.

We are seeing two different design philosophies from the developers.

 

SF2E caters to more systems. Even those with weaker gaphics card. That's because most of the load is in the CPU.

 

LOMAC caters mostly to gaming PCs. Won't work on low end machines but will "wow" you if you have a decent graphics card.

 

Nothing new here as well.

[Aintry]

Well, I didn't know any of this stuff, so I'm glad Fubar posted the comparison.

[+ST0RM]

I'm one of the people who have to see it, versus read it. Nice visual presentation.

 

I've noticed on my low-end laptop, I can run SF2V on medium-high settings at the highest res and I get about 15FPS. Smooth enough for game play, until Stary's effects kick in when an AI flight lets go of a stick of Mk.82s. I've got Green Hell 2 trees added as well.

 

Now with SF2E, I've got to turn it way down. I've got the forest/trees add-on. Medium settings with med. res and I still get more stuttering.

 

Each game runs things a bit differently with each release.

[+Gr.Viper]

Fubar, can you run the same test in Nam with a mission involving a massive B-52 strike?

[Helmut_AUT]

So that means SF2E (and likely the other games on similar engine) are bottlenecking on the graphics card and memory, NOT on CPU?

 

I actually thought graphic effects were more CPU bound and thus high graphics would max out the CPU before the card.

[+Fubar512]

The SF series (1 & 2), can be waaaay more taxing than LOMAC. And I believe the reason behind this is due to model and terrain complexity, coupled with rendering distances. You see, LOMAC-FC achieves what it does, with less. It's native models are simpler, and the actual draw distances are lower (thanks in part to the creative use of fog). Many of the SF models are more complex, SF's textures tend to be larger (which is especially the case with third party add ons), and SF's normal and far horizon draw distance are well known frame rate killers.

 

Conclusion? If you want to run with the big dogs, and be able to play SF with everything maxed out, and loaded with third-party models and textures, then you'll need a video adapter with at least 896 MB of VRAM. Not that that's an expensive a proposition..... Here in the 'States, one can still pick up a GeForce 9600GT armed with 1024 MB of VRAM for under $100 US. Even though it's considered nothing more than last year's mid-range card, it should still easily run either SF1 or SF2, even with all the eye candy and third party add ons.

[+ST0RM]

Here in the 'States, one can still pick up a GeForce 9600GT armed with 1024 MB of VRAM for under $100 US. Even though it's considered nothing more than last year's mid-range card, it should still easily run either SF1 or SF2, even with all the eye candy and third party add ons.

 

Where? I'm still running a 512Mb card from two years ago. I'd like to give myself a welcome home present in a couple of weeks.

[+Fubar512]

Where? I'm still running a 512Mb card from two years ago. I'd like to give myself a welcome home present in a couple of weeks.

 

 

Right here:

 

http://www.newegg.com/Product/Product.aspx...N82E16814500043

 

Of course, I'd rather see you get this:

 

http://www.newegg.com/Product/Product.aspx...N82E16814130469

 

[+Fubar512]

If anyone's curious as to how to perform this experiment, go here: http://forums.legitreviews.com/about8247.html

[Helmut_AUT]

Word of advise: The GTS250 is a card built on the previous 9-Series Nvidia chip, not on the newer g200 chip. Still better than a 9600GT, but If you are looking for a new one, and not the absolutely cheapest around, start with the 260GTX and up, best performance for value currently being with the 275GTX (which is kind of like the 9600GT back in it's generation a surprising fast mid-range price card).

[+Dave]

I have the 285GTX, man does it run this sim well.

[+Fubar512]

Word of advise: The GTS250 is a card built on the previous 9-Series Nvidia chip, not on the newer g200 chip. Still better than a 9600GT, but If you are looking for a new one, and not the absolutely cheapest around, start with the 260GTX and up, best performance for value currently being with the 275GTX (which is kind of like the 9600GT back in it's generation a surprising fast mid-range price card).

 

Actually, the GTX260/275/280/etc is turning to be a bust in the flight sim world. The GTS250 actually delivers higher FPS in Miscrost FSX than even the GTX280.

[+streakeagle]

TK's last round of sims ups the ante on bitmap resolutions and with addons that have even higher detail, what did you expect?

Cap the game to poly numbers and texture resoutions comparable with LOMAC to try to detect efficiency differences between the game engines.

All I have learned is that when a game developer allows uncapped poly counts and textures sizes that someone will find the limits of their system

TK's willingness to do so is great, otherwise the original series of games would have been obsolete years ago in terms of graphics.

[+Julhelm]

The main problem is that TK's engine clearly isn't intended to match even LOMAC's 2005 visuals since the engine is still essentially 2002 technology that has no provision at all for quick lodding of and procedural generation of terrain objects such as trees, nor any useful implementation of graphics standards such as normalmapping of objects.

 

The latter is funny because he does have normalmapping on the water and terrain, but just won't bother linking the libraries to the other shaders and justifies this that he "doesn't have the budget".