Mainly with new x30

I have used 0.80, but as I tested now with 20 AI cars on new SO6 track, starting back of grid, I got only 14fps.

Then changed that to 0.30 (mirrors 0.20, dust 0.20) and got already around 20fps.

I'm running with Athlon 64 3200+, 1GB ram, X850XT gfx card, no AA no AF, 1280x1024 native resolution of 19" tft.

So it won't run, I prefer constant 60fps to LFS to be playable, so what is your LOD settings?

Intel E6550, 3GB RAM, GeForce 8600 GT, 1680x1050x32

I have everything on full blast, dither disabled, and set up a manual profile for LFS to have 4X AA, texture filtering, etc. etc. (basically followed the 'how to make LFS look really good' guide). Also, i have the high res pack from lfs-database.

With 31 other cars on the track in the most taxing graphical environments (cars running off track, etc.) i've seen it dip into the 50FPS range, but most of the time it will run in the 70s... i cap it at 60 because i'd prefer it to be consistent.

edit: should also note that turning off texture filtering in the nvidia profiler lets the game run over 100FPS, and i've never removed the cap to see how high it would really go.

Quote from JTbo :

I'm running with Athlon 64 3200+, 1GB ram, X850XT gfx card, no AA no AF, 1280x1024 native resolution of 19" tft.

That doesn't make any sense.

I have a similar system (P4 3.2GHz, 1GB RAM, X800XL running as 1280x1024) and I run 4xAA / 8xAF with good FPS.

Hes talking about running with a full grid of AI...

Quote from sil3ntwar :

Hes talking about running with a full grid of AI...

Indeed and SO6 track back of the grid.

But most importantly I'm curious to know which LOD you are using?

i use 0,0,0 with x30

I've got the same hardware as JTbo, but a bit older graphics card (x800). I am, and I've always been running full level of detail BUT with the dynamic LOD reduction set to 1 (ie very dynamic). These settings have proven to be working just fine for me previously.

But with X30 I am experiencing a drop in FPS. Haven't investigated it though. Nor am I whining. I'm just pointing it out.

E6750, 2Gb DDR800, GayForce 8600GTS 256Mb.

LFS running on full tilt without LOD reductions at resolution 1650x1050 and AA/AF both 8x. FPS capped to steady 58 frames to keep things... steady.

I haven't tried with the AI, though.

Oh and anyone else suddenly having strange slowdowns with newer Nvidia drivers?

Athalon 64 3500+ (what is the + signify? :shrug ATI x1650, 1 gig RAM and I haven't changed a thing. Everything full in LFS and 4x AA 6x AF (maxed) and see solid 50 FPS because I lock it there in LFS. Same settings as X10 and same FPS. I think I see 75 FPS if I don't lock at 50.

With everything else in the patch, I completely forget about the AI. I haven't tried the AI yet.

+ is AMD ego booster, they imagine it could match that fast cpu

With 0.80 LOD, which is default I think, there was no problems when online and alnoe on track as it was rarely I went to server that had 20 racers and even more rare that I would been starting from last place ( only on combos I don't know of ) but with so many AI cars there seem to be need to reduce LOD, also it seem to help somewhat with online too, there was some hiccups earlier but now it is again silk smooth, can't tell difference in graphics, though.

In bl y So all in minimun , i dont know what , but the tracks are all texture or dss in High resolution

i have 15 fps in x30

and in x12 i have more that 25 Fsp
________
Mywebcamshookup

AI will make those FPS dissapear fast with a full Grid. Comparrisons should between online and single player should be seperated i think and shouldnt be compared.

I setup a single player AI Race 1 lap and recorded the replay here
would make a good benchmark for comparing results.


Race condition
Single player or Online: Single player
Track: BL1
Amount of cars & type: 20 AI FXO
Conditions: no wind, clear day

Game settings
View Settings: Cockpit
Resolution:1650x1080
Mip Settings: 0.0
LOD settings: User 1.0 Mirror 0.3 Dust 1.0
Mutliplayer Speedup: no
Dynamic LOD: 0.5
Skin Settings: default 512

Hardware Settings
CPU type: AMD X2-5000BE(@3.2ghz)
Memory: 2048mb DDR800
OS Type: Windows XP-SP3
Graphics Card: 8800GTS320mb
Driver Version: 169.21
AA AF Setting: 4AA 8 AF

FPS Results

Average: 48
Low:36
High:80+
Car viewed From:AI 20
Start Of Race: 43

Not so bad for a full grid of AI i think.

Quote from Gener_AL (UK) :

AI will make those FPS dissapear fast with a full Grid. Comparrisons should between online and single player should be seperated i think and shouldnt be compared.

I setup a single player AI Race 1 lap and recorded the replay here
would make a good benchmark for comparing results.

Mine with my system changes compared to yours in bold. I didn't use Fraps, just watched the FPS counter myself. The differences is my resolution and AF setting (Changed them to my maximum available)

Race condition
Single player or Online: Single player
Track: BL1
Amount of cars & type: 20 AI FXO
Conditions: no wind, clear day

Game settings
View Settings: Cockpit
Resolution:1152x864
Mip Settings: 0.0
LOD settings: User 1.0 Mirror 0.3 Dust 1.0
Mutliplayer Speedup: no
Dynamic LOD: 0.5
Skin Settings: default 512

Hardware Settings
CPU type: AMD Athalon 64 3500+
Memory: 1024mb
OS Type: Windows XP
Graphics Card: x1650 256mb
Driver Version: ?
AA AF Setting: 4AA 6 AF

FPS Results

Average: ?
Low:19
High:29
Car viewed From:AI 20
Start Of Race: 25

I used my eyes also (years of 3dmark) there was some LFSbench app around, ill have to search for it cant remember who when etc.

Do you know what I don't understand? I tried it with 20AIs on BL1, with 8xAA (that's how I play LFS) and 0xAA. There was no difference in fps! From that I would conclude that my gfx card does almost no work, in other words it has heaps of surplus, the cpu is what limits the fps here (which is btw a core2duo @ 1,8Ghz, please Scawen, implement multithreading).

But: then I lowered the LOD from maximum to minimum, and I gained some ~10 fps! This is totally contradicts to the fact stated above. Ie: If the gfx card has surplus, how do you gain fps from lowering the gfx details?

Hm, wait a second, computing gfx involves the cpu too, even hardware T&L can't leave out that step entirely. So maybe that's the reason? Computing AA requires only/mostly the card, but computing verteces requires more power from the cpu too, and that's already on its max computing AIs and physics.

Opinions?

I also had an fps loss with X30 but reinstalling fixed it. I think the three multiplayer settings in the misc section of the options make the most difference. I haven't tested with AIs but I imagine they will put quite a load on the CPU because of the physics of their cars.

As Iron says LFS does most of the work on the CPU. So a real fps improvement could be achieved only by adding multi core support but I'm not sure it would be worth it for our lone coder (I think he once said it would be about 1 month of work), considering how CPUs are getting faster and faster all the time.

Better wait for a real rewrite of the graphics engine in S3 to get those GPUs a bit warmer.

Quote from Apoc112 :

Intel E6550, 3GB RAM, GeForce 8600 GT, 1680x1050x32

I have everything on full blast, dither disabled, and set up a manual profile for LFS to have 4X AA, texture filtering, etc. etc. (basically followed the 'how to make LFS look really good' guide). Also, i have the high res pack from lfs-database.

With 31 other cars on the track in the most taxing graphical environments (cars running off track, etc.) i've seen it dip into the 50FPS range, but most of the time it will run in the 70s... i cap it at 60 because i'd prefer it to be consistent.

edit: should also note that turning off texture filtering in the nvidia profiler lets the game run over 100FPS, and i've never removed the cap to see how high it would really go.

can you post a link to that guide?

Quote from Gener_AL (UK) :

I used my eyes also (years of 3dmark) there was some LFSbench app around, ill have to search for it cant remember who when etc.

All the benchmark was, is what you and I just did there. It was a replay someone made with AI cars, but had a more easily aquired 1024x768 resolution and no AA/AF so everyone could use it and compare at an even level.

There was a set replay and 2 cfg.cfg (whatever the file name is), one max LFS settings and one min LFS settings. The procedure used Fraps to give a log of your max/min/avg FPS. I used it extensively when I had a lesser spec computer to test various PC configs for framerate back around patch P.

Quote from joeynuggetz :

can you post a link to that guide?

The search button can bring you miracles!
http://www.lfsforum.net/showth ... p;highlight=make+LFS+good

Default, whatever that was (I'm not at home to check right now).

I followed Jakg's guide, so I assume everything is up full I also use a bit of AA, although not much.

I do drop to about 15fps when looking at more than maybe 10 CPU drivers at once, but that doesn't happen often. I'm using a single gfx card (7900GS-E) and a single-core CPU (64-bit 3800+) so I know what's knocking it down.
Because of our connection being off to move house tomorrow I've not had a chance to try X30+ online yet, but I can't imagine the frame rates being that much different to what they were before - about 30 on the starting grid. I limit it to 100fps since the 150 I was climbing to seemed a bit pointless. I might even drop the limit further since 100 is still more than my monitor will display