[poncho]

With some mods, users report loss of frame rates. When I am assembling a mod I try to test it's effect on frames, but I never see that impact that others report. What I would like to know is what are the characteristics of a mod that influence frame rates? I assume that TGA size is the major factor, but do the number/size of mesh files have a large impact also. I ask because the easiest solution to repositioning cockpit components is to create many smaller meshes to replace one large one. Does anyone have any insight into the frame rate effect of doing this.


Also I have been trying to economize on the number of new tgas by using one tga for many new meshes, does this help with processing speed? My thought is that this will speed up the process on slower computers, is this a valid assumption?

[{HVY-E}Jinxx]

The one BIGGEST factor is the higher resolution objects and cockpits. Stock are 128 x 128 and 512 x 512. Most of the high res stuff the guys have been putting out are 1024 x 1024.

Also, the larger maps and denser population areas are a factor as well.

Even with my machine, if I run everything at max and fly over a densely populated area, I drop from 204fps to about 22fps.

[Neil Lowe]

I often wonder if the large texture cockpit updates are a bit overkill for the overall effect achieved. When you think that mapmakers can use a 1024x1024 tex to map to an area of a city, these same sized tex are used for cockpits. Having said that I have a few high res cockpits in my install and you can see the difference. But if texture memory is on the slim side a well drawn, smaller tex looks great too

Too many mods seems to affect overall performance too. I had a clean out recently and the game was noticably smoother.

Cheers, Neil

[JAMF]

To give another option to the question:

Would 4 separate instruments meshes textured individually at 512x512 have a bigger hit on framerate than a single "block" mesh of 4 instruments with one 1024x1024 texture cut in 4 parts?

[poncho]

Yes, that is a good example of what I am talking about. When I release the next version of P47 cockpits, I was thinking of making 3 versions. A low res, medium res. and high res. And I want to understand what is needed for frame rate friendly mod. Altering texture size is the only factor that I am aware of to do this.

I add a lot of bits and pieces of mesh to my mods.
I know when Oleg was accepting models for the game, he had set a limit on the number of polygons in the 3D model. Clearly, this must have been done for frame rate reasons. I wonder what the impact of adding my bits of mesh to cockpit models may have on frame rates.

If I knew what the bottlenecks were in the game system, it would be useful.

[bolox]

in my experience tga size has only a minor effect on fps- this is with fx57 1800xt gpu with 512mb and 2gb ram,1600*1200res- anecdotal evidence suggests vid cards of 128mb struggle with hi res cockpits.

on my rig:-

JU88- barely 1 fps drop despite having texture size of ~0.5gb from a 'normal'~43fps. no difference when i took the size to ~950mb as an experiment hock: hock: :twisted: - this causes v long load times and crashes/freezes with large maps


most other pits give no noticeable drop with a couple of exceptions, notably Freddy's P51 pit which causes~10 fps drop and P47 and Mosquito giving a very slight ~2 fps drop

to me this indicates mesh/polys are responsible for fps loss more than tga size- but others experience may be different :wink:

on a side note, i get a consistent 3:1 ratio of d/l's between hires and lores versions of my repaints

[poncho]

Thanks Bolox,

That has been my experience also. I have had 256 MB and 512 Mb vid cards, and have only seen a slight FPS loss with hi res cockpits. I am trying to finish up some P47 pits and I am trying to assemble a frame rate friendly version. I don't know if I am succeeding though, I can't see much difference between the versions on my rig.

[PA_Willy]

Yesterday I was doing a test with FRAPS:

-- Single GPU + SLIAA 1600*1200 x16AF SLI8x (4AA + 2SS) / Driver 182.06 (with MODS)
- Time: 106448ms - Avg: 100.143 - Min: 33 - Max: 240

-- Single GPU + SLIAA 1600*1200 x16AF SLI8x (4AA + 2SS) / Driver 182.06 (without MODS)
- Time: 106472ms - Avg: 108.714 - Min: 36 - Max: 263

As you can see, the difference is low. A few frames lost (from 108 to 100 in average). But I have to recognize I got a powerful machine: C2D E8400@4.0Ghz + GTX295 OC + WD Velociraptor 300 + 4Gb Gskill 1100, etc.

Regards.