Found a pretty annoyng bug on skidpad, was just playng around. Made a little replay for this, just a short one to see because i dont know how to describe this. It gets stuck.
E8400/4G/GTX260+/XP32-SP3. Res is 1920x1200, FSAA ranging from none to 16xQ (via the driver), AF from none to x16, both high and low res shadows, FPS changes from ~58 to ~75 over these different settings with max AI cars, paused, external view from the front backwards that enables seeing most of the grid (Fern Bay Green).
It's a bit hard to see the exact FPS since it changes so fast, but I'd say about 5% slower with the test shadows.
Interesting FPS observation: When looking from the last car forward (either cockpit view or behind-view), FPS is almost double (70-90% increase) than when looking from the front backwards toward the grid... even though on both cases the LOD setting is such that I can see all cars...
is it possible to talk about the GTR/TBO balancing (Patch Z) in a new spinoff thread? i think there is a need for clarification before the next patch....
wanted to create one but dont have the permission
you probely missed the last 2 times that we have the disustions.
These two were both at a spin off (theard)
And i do agree, i think there are some things that need to change in that,
anyhow i got the feeling that scawen and special victor know that the tbo needs to have a better balance, because of the TBO cup where victor drove a couple of times.
So i think we just need to ask what they are going to do about it
If they know it : the reason why it is not in a test patch yet that is because its not online compatible anymore if you do that
To see constant FPS number use FRAPS. Its more accurate in my view then LFS.
I have seen a lot of disprepencies of FPS in LFS which are puzzle to me and does not make a logic much unless there is something behind we dont know. For example looking at Start grid of AI in BL from TV view (LFS paused) the FPS are like 1/3 rd of when looking from first AI wheel view but the GPU usage is still the same and around 50 percent between both views. Not a much logic in this unfortunately.
Another example would be when LFS is paused and I am in the corner before the bridge at Westhill (offline game) the FPS goes higher when I turn virtual mirror on. Logically it should be opposite or its like that as the hardware must draw more information on screen. I found this odd behaviour in other places as well.
/the above is not based on the issue I meantioned in the other thread regarding differences of FPS between 2 OS systems and performance gain/loose between old and new hardware/
Other thing why LFS sometimes only stress some CPU&GPU only to 50 percent and the FPS are still low. I would be expecting the FPS drop due to the hardware not possible to cope with it not to be fully unused.
I am going to test the shadows performance gain/looses once I will be back at home.
CPU: Phenom II X3 720BE
Graphic: 4850 DDR5 512MB
Driver: 8.56.1.13
WIN 7 RC1 - practically same a the version which will be sold
1680*1050;4xAA + 16AF
USER LOD: 1.0
Screenshots always showing Standard shadows
- In the first screenshot from TV view the difference between test shadow and standard shadow was 1-2FPS max (less with test shadows)
- In the second picture from wheel view
Standard shadows: 110FPS
Test shadows:100 FPS (around 10 percent descreas in performance)
- In the 3rd pictureChase cam
Standard shadows: 150 FPS
Test shadows: 148
I tried also different tracks and positions and found that flucations is around 1-10 percent descrease with test shadows. However found some interesting pattern. In all views the difference between standard and test shadows is only 1 FPS hower from wheel view with mirror turned on its often 5-10FPS.
The GPU usuage was only all the time around 45 percent with fluctuation 2-3 percent up and down in all views except TV where the fluctuation is like 43-60 percent with sinusoidal shape.
I have theory, maybe stupid...
LFS might have too much pixels overdraw or not so good vertex cache performance.Maybe I am totally off.
Also heard somewhere when someone modified CAR models that its problem to use many polygons otherwise the performance decrease a lot. Sorry for mentioned about modding of LFS but wanted to report it.
I think i found a bug (may be repost though - not sure) as you can see on attached pictures the wheels (disc brakes and spokes to be exact) are visible only from one side.. the far side wheels are missing disc brakes and spokes. I attached images from 0.5z (works fine here) and 0.5z15 (missing elements) but i also checked in z13 - missing too.
I have big issues since patch Z13.
The screen freezes, anythign else (chatbox, car, menu) doesnt.
Today i tried minimizing and re-opening.
and it gave me "Warning: Shadow Error (1)"
this happens with normal AND test shadows.
also made a video of it, will be uploaded later.
i will try using .exes rom earlier patches and see if it still happens.
In real life: the closer the surface is to the car the sharper the shadow it is. Diffuse light is bouncing back and forth causing shadows at a greater distance to fade.
Regarding the test patch:
Z15 works fine for me. Haven't come across any bugs. Windows XP Professional SP3. Great to finally be able to see the gears in MRT! Still waiting for orange (ca. #FF5500) text colour
EDIT:
Not Z15 related, but you can't see the gears (especially 5 and 6 are hard to recognise) in some other cars aswell from the default driving position (UF1, FXR, FZR).
something weird has happened to me this morning
I tried to watch few of my old hotlaps, done in Z13 with Z15 and just when it starts my best lap in t1 it shows OOS (or whatever is the name) error