Sorry for another post, but I simply must say something about this occasional "false"-false-starting. Many are saying it's the people trying to predict the green light or they are not engaging brakes or they are "gunning" the engine. When it happened to me, it happened as soon as the race restarted. I didn't even get a chance to see the grid infront of me. I had not touched anything control related. It just through me into spectate with a false-start warning along with 6 or so other people.

I have tried gunning the engine other times. No false-start. I have even tried to false start on purpose in multiplayer and moved about 2 meters, false-start message appeared with a drive-through penalty. Great it worked perfect.

There just seems to be a little false-start glitch occasionally, no biggie - I'm sure scawen is working hard on it. Just because you have not experienced it yet, does not mean it does not happen. Thanks everyone.

Quote from Dygear :

No, I'm saying it could be the reason for the false false starts.

No, the false² starts also occur when you don't touch the controls in any way. They seem to be completely random.

Quote from Jakg :

Scawen, have you optimised this code at all? i usually get about 45 FPS with a full grid, and now with the extra 8 cars on track i only get 40 FPS, which which is a ~12.5% decrease in FPS for 40% more cars!

Great work Scawen, now go and get yerself a pint!

I would a bit disagree or there could some discussion.I sometimes feel - well I see it on my screen .Just tried yesterday to put 16AI on track.When the cars where driving on track and almost all AI were close to each other, I had from wheel view with 1st car around 250 FPS when game was paused(I guess when game is paused physic calculations are not happening) and with 2nd car I had around 150 FPS. Its not about I have this huge FPS and I am complaing but I think this "bug" affect huge grid start FPS drop or where there are many cars around.If my assume is right that when the games is paused that physic calculation is not done then it has nothing to do with CPU power.

It doesnt not make sense to me that with only 1 car drowed before me the FPS drop is so huge.I see this behavious in LFS 3D engine very often and it give a HUGE question why this is happening.Something wrong with utilising Vertex engine in some cases? I would like to really know the answer to it and if from Scawen this could be even better.

If this could be "fixed" somehow then we could have some nice FPS boost especially at grid start with many cars.Maybe if Scawen could try NVPerfKit for vertex peformance? I wanted to try it myself just for curiosity on LFS but LFS is not DX8.1 which is the minimal requirements.Its just DX8.

I think my PC spec are good so this should not happening.
AMD 3200+
2x512 MB RAM,CL2.5,Command rate 1T
NVIDIA GeForce 7900GT

SORRY IF THIS IS TOO MUCH OFF TOPIC

Am I right that the false start spectate (after 4m) is disabled when you're along on the grid (in multiplayer) or always in single player? Because I tested it offline, and never got the expected spectate. Didn't try it online because I'm on the grid to race not to purposely get spectated (and I haven't been online much recently anyway).

Quote from DEVIL 007 :

I had from wheel view with 1st car around 250 FPS when game was paused(I guess when game is paused physic calculations are not happening) and with 2nd car I had around 150 FPS.

If you're in first car, you have no other cars in front of you, thus saving a ton of polygons to be rendered. If you're in second car, there is the first one in front (duh) which of course takes some calculation power.

More important: Please realise that a drop from 250fps to 150fps is NOT huge. The rendering time for a frame simply changed from 4ms to 6.6ms. If you had 50fps (20ms), this would be equal to a drop to 44fps (22.6ms).

Quote from a n d 1 :

So when i start a Qualification i start in the Pits, that is actually fine but I have a message (False Start - sent to Spec).

I couldn't reproduce that locally but I could reproduce "False start - drive through penalty" in qualifying mode.

Quote from Ondrejko :

Single player and demo races now up to 16 cars

does it mean 16 cars in race + 1 spectator?
should i set 17 connections in setup.cfg?

I am going to change this soon, to 16 connections, but only 12 actually racing in the demo.

Quote from SpaceMarineITA :

A simple question :
Atm with 20 players limit + 4 spectator there is a command that allow you to set the server for 23 players and 1 spectator. (i have played some times ago into some 23 slots servers)

There will be the same command with the next patch? (allowing to set some servers with 47 players? )

No, that is not possible.

Quote from danowat :

I understand all that, but 20+ spec slots?.......

Driver changes in endurance races, etc. Absolutely vital. And it doesn't harm anyone if the server is up to it. Anyone hosting should use the bandwidth calculator in the client version of LFS to check the upload requirements and do not overload.

Quote from sinbad :

I've changed my mind about the "handicaps". If they're just changes to the car, then the cars should always be the same everywhere, single or multi. If they're truly handicaps, though, then why not have them work properly everywhere if it's possible to do so, which it is.

I can't change them in single player because you are not connected to the internet. And this would destroy the hotlap tables. The handicaps (or other changes like narrower tyres on the fxo) will become permanent in the next physics update.

Quote from Shotglass :

There´s a problem with the race setup screen in multiplayer. If I´m not mistaken (almost) all drivers on the server have to click ok the start the race which takes ages with 48 clients 20 of which are speccing and maybe AFK.

Only drivers have to click ready, not spectators, if I am wrong please test and tell me.

Quote from JasonJ :

There just seems to be a little false-start glitch occasionally, no biggie - I'm sure scawen is working hard on it. Just because you have not experienced it yet, does not mean it does not happen. Thanks everyone.

I guess some position packets are flying over from the previous race - that could cause the server to make you spectate (it thinks you are 4 metres away already before green light). And about the false start penalties, I can only guess at the moment it things the revving car twitch is actual motion. Pure guesses, I had a day off yesterday for the first time in while.

To the other person who was pressing me for an answer - obviously I cannot give answers without investigating. Just give me time. It's a test patch for finding issues, it's not the official version of LFS, and quite clearly I won't leave it with that glitch, and when I know the answer, of course I would tell you or release the fix.

Quote from AndroidXP :

If you're in first car, you have no other cars in front of you, thus saving a ton of polygons to be rendered. If you're in second car, there is the first one in front (duh) which of course takes some calculation power.

More important: Please realise that a drop from 250fps to 150fps is NOT huge. The rendering time for a frame simply changed from 4ms to 6.6ms. If you had 50fps (20ms), this would be equal to a drop to 44fps (22.6ms).

Hi AndroidXP,
I know quite good when there is no car in front me only track is renders but still something is rendered.I am not such a noob ....

Still the LFS cars has not so much polygons to make such a drop.And DROP is still huge even you saying the opposite.Its not about the "ms" staff you are talking about.60 percent drop with track+just 1 more car looks really odd and huge drop to me.When I am in 3rd car the drop is not so huge....etc.The drop seems very non-linear.I only saw this with games badly optimised in 3D engine. LFS seems overally very effective but sometimes these drops doesnt make really sense to me on very powerfull machines.

I really dont want to steal this thread by this 3D stuff so this is my last post regarding this but as I said something looks probably "wrong". I could make testing with examples if needed.

Overally I would like to thx Scawen for this patch.Looks realy promising in the multiplayer area and the clutch thing as well.I am looking forward for the next updates...

Quote from Scawen :

And about the false start penalties, I can only guess at the moment it things the revving car twitch is actual motion.

I don't think that's it. I am in 1st with clutch pressed on the grid and I rev the engine. Usually that works fine, no penalties. Some times I do get spectated for doing exactly the same thing. Those times I have got spectated I have been able to be on the grid for while (revving or not).
That packets from previous race sounds like the most reasonable explanation to me.

don't have time to test but just want to say :wow:

Quote from DEVIL 007 :

...

No, again, the drop is absolutely irrelevant. The higher you get with framerates, the less they mean. It's really the frame rendering time which is far more important. For example, I coded a little sprite based "spaceshooter", and while not rendering anything but the ship I got like 2000fps (0.5ms frame rendering time). Then I added a few enemies and suddenly I had a "huge" drop down to only 800fps (1.25ms per frame). The real difference was only 0.75ms, but the frame rate display showed a huge drop. Trust me, this is no bug -the way fps are calculated is deceiving you there.

Quote from Scawen :

I couldn't reproduce that locally but I could reproduce "False start - drive through penalty" in qualifying mode.

Just to let you know. Fresh of playing on a W9 server. I was only person on the server and when qual restarted got the False start and was jumped to pits every time I Shift+Q.

Preload has made huge difference to the cars. Has made things just that more natural feeling. The RA now only wants to kill you sometimes

Quote from Scawen :

I couldn't reproduce that locally but I could reproduce "False start - drive through penalty" in qualifying mode.

Hi Scawen,

I have today downloaded and installed the W9 test patch and I have also come across the bug where by joining a server picking your car and then leaving your garage I was booted off with a "false start - return to spectate" message as I drove out of the pits. I have been trying to reproduce it but with no luck. It only happened the first time I actually joined a server with this patch and after that all seems well

cheers for the patch btw, full of nice features

mad

Quote from geeman1 :

I don't think that's it. I am in 1st with clutch pressed on the grid and I rev the engine. Usually that works fine, no penalties. Some times I do get spectated for doing exactly the same thing. Those times I have got spectated I have been able to be on the grid for while (revving or not).
That packets from previous race sounds like the most reasonable explanation to me.

OK, I must point out there are two totally different things :

1) The host making you spectate (if it think you have driven more than 4 metres). This could be caused by delayed position packets from a previous race (or qualifying) - I have a solution for this, marking the packets.

2) If you receive a drive through or time penalty (but do not get forced spectate). This could possibly be engine revving + wind effect - it is worked out on your local machine but uses 3d speed. I have a reasonable solution for this as well using 2d speed. But I don't know yet if we have got the real cause. This would happen in single player as well - if someone can get this in a SPR then I will have some hard evidence to find out with 100% certainty why this is happening. The qualifying version of this is just a simple bug, an oversight.

By the way, the fixes are going to be incompatible again, W10 will be incompatible with W9.

Quote from amp88 :

I don't think this is the case. In single player, after a few test runs I managed to guess and time it perfectly so I was in 1st gear as close to the green as possible, but I didn't get a penalty. So, does this mean the multiplayer jump start detection is unreliable due to lag (or other complications), or is it just that the MPR file doesn't accurately portray startline events?

Scawen, can we please get some confirmation on this matter?

Nice start 0.01 secs

No person can react that fast in reality so there definitely needs to be some limit set on that.

Keep up the good work

already been mentioned about 30 times within this thread

hm.. i think its a good idea to level the performance of the cars in the different classes. But as far as i tried, the FXR is now as fast as the FZR, now uses even less fuel and is still much more stable (4WD). So before, there was the tradeoff between speed and stability when choosing your car.

Now there is only one reason to keep driving FZR and that is less tirewear (for me at least). In races the guys driving FXR just push themselves alongside you and stay there, not having to worry about being thrown off the track. A collision, even a little one is more likely to throw the FZR out than the FXR.

I would propose to put a little less weight on the FZR so that its still faster but not so much to leave the FXR and XRR drivers far behind.

Good comment. I've now made a special thread to discuss car balancing, perhaps you could post the same text over there.

http://www.lfsforum.net/showthread.php?p=393772#post393772

Other people please post there if you have something important to say about car balancing but please don't spam the thread, only post if you have researched it and know what you are talking about. That way we'll keep the post count down and to the point and avoid repeated postings.

Thanks!

Quote from Gabkicks :

already been mentioned about 30 times within this thread

I hope that wasnt aimed at me, and from what I can see it is mentioned three times after me and axus brought it up and amp88 did some testing.

Quote from Scawen :

Good comment. I've now made a special thread to discuss car balancing, perhaps you could post the same text over there.

http://www.lfsforum.net/showthread.php?p=393772#post393772

Scawen - I didn't want to sully your nice clean new thread with this, but: No plans to balance the LRF class at all? I know it's not as popular as the TBOs and GTRs but the LX6 thrashes the other two in most circumstances. It would be nice to see the other two cars being used more.

Just my 2p.

Quote from Scawen :

Only drivers have to click ready, not spectators, if I am wrong please test and tell me.

Kind of hard to test this without having a team behind me.
But shouldn´t /clearlist (I think thats the command to clear the list of drivers in the race) help to start the race instantly? All I know is it didn´t.

The command is /clear But I agree, it does not really help to force a race start, since you will also remove yourself from the list and any chance of hitting "ready" before anyone else joins the grid.

I suspect Franky can second this statement

Edit: ah, a temporary solution could be to create a script which executes following commands

Edit 2: no, the commands are executed too rapidly, I guess we will have to wait for the pause/sleep/wait command.

Edit 3: Executing two scripts with a (driver induced ) delay could work though:

script 1:

script 2:

Quote from ebola :

I hope that wasnt aimed at me, and from what I can see it is mentioned three times after me and axus brought it up and amp88 did some testing.

no it wasn't. It was about a post in between that later got deleted. I read that post and the other minute it was gone. It was about a small bug been mentioned too often I guess

Why can't you access lfs.net?

Anyway, here you are.

mirror: http://boost.am/58991/lfs_mp_test_w9.zip.html

because im posting already: love the patch great step forward again for this lovely sim
now we just need a disallow shift + s option in an ongoing race and i am happy

edit: was to slow there

Quote from ebola :

Nice start 0.01 secs

No person can react that fast in reality so there definitely needs to be some limit set on that.

Keep up the good work

Since we are at the point where human reaction and prediction is involved in the starts, would it be more realistic to make the lights more random? In drag racing here in the US, when the starting lights go on (from a human starter), the time from yellow to green in constant, but the time from both cars ready to the start sequence is random.

In LFS, we have the three light countdown that is at constant time intervals (red to red to red to green), giving people the opportunity to predict when the green light comes on, giving some people (that time it correctly) an advantage and others a larger disadvantage from lag.

Would it be difficult to change it so that the red light still come on at the same intervals, but they all stay on for some random amount of time (or at least one of a few different set intervals, chosen randomly)?


I have another question regarding patch versions: When this new incompatible patch comes out, can older patches (W) still communicate and log actions/laps/PB's with LFS World?