Jump to content

Sparker

Member
  • Content count

    35
  • Joined

  • Last visited

  • Medals

Everything posted by Sparker

  1. Thanks! You are very much right... I was looking for something like the profiling build does actually. Profiling, Performance, Arma Script Profiler, these things are so confusing!
  2. So, could we load it on a linux arma 3 server and remotely profile it? I'd like see what the engine is doing every frame. because we often have huge performance drops down to ~3 FPS and it would be very interesting to know what the hell is going on.
  3. AI Driving - Feedback topic

    Where is 2.? Having a list of known mission developers is a good argument. Why does 4. make you laugh? I'm not really aware of the structure of the company and who is responsible for what and who we should send it to, but I can imagine the collective message having more effect than our usual bumping of this thread.
  4. AI Driving - Feedback topic

    I suggest we do it like this: 1. Find scenarios and videos showing the evident failures of driving and convoy handling without any mods. 2. Collect (a sort of) signatures of people affected by this issue, most important is to find mission developers of popular scenarios. 3. Create a proper text which explains everything. 4. Send it to someone in the upper management of the company who can dispatch the resources to respond to the problems. If the company is known for its relations with the content creating community, they won't ignore this (I guess).
  5. AI Driving - Feedback topic

    You don't just say the magic word 'modular' and expect something to be magically better based on that. For instance, the issue of airplane issuing a turn-away maneuver 300 meters from target instead of 500, is it because it's not modular? Also I think there is no way to let modders access the AI code since it's written in C, not SQF. Unless they specially implement the AI code as a library, give us its source code and let us connect a different library to the game. "a way too low cost-benefit ratio (spend days, weeks fixing seemingly simple issue) " - the AI topic clearly affects mission creators like Barbolani. Upset mission creators -> less missions -> less players -> less money, I thought it should look like this from a company's point of view. I think this whole conversation is pretty pointless. Clearly, nobody is going to do anything with AI any more, ever. I have made a negative review on Arma to raise the awareness of other people about this problem. It's not something I'd like to do to my favourite game, but what else can we do?
  6. AI Driving - Feedback topic

    Sorry to post my workaround to 'convoys get stuck trying to depart' problem', but in case @barbolani needs it: if convoy is detected to get stuck by a script, try to select a new different leader for the convoy group. AI will do some replanning and eventually they will start moving... that's the best I know right now.
  7. Help on debugging FPS drop

    The BIS_fnc_safePos mentioned above, if massively called, could slow down the framerate because it asks the engine to do a search on nearby objects. However no idea how high the search radius should be to really cause a trouble.
  8. I thought that creating very many member variables in the missionNamespace might degrade the lookup time for accessing them. Though, storing the variables in sub-namespaces (engine's vehicles or other kinds of objects) might help it. But actually all these BIS_fnc functions and functions from our own missions already occupy the missionNamespace, there are about 2 thousand of them total IIRC, so probably storing member variables from 100 "objects" or so doesn't change anything much.
  9. Hi! Probably noone can tell you. Better contact the author himself at steam: https://steamcommunity.com/id/154578/ He's a cool guy and I'm sure he will help you, if he has time.
  10. I've had a look at oop.h. Naught and you probably have a degree in preprocessor magic and its applications! Really impressive. I'll give it a try some time soon! I was considering one thing though. Is it faster to store all object's methods in global namespace or can it be made faster if I create objects with createVehicle and store its member variables there? Does it make any sence to spend time doing this, what do you think? It appears to me at the first glance that manipulating variables in mission namespace is pretty fast.
  11. Hello, developers! As I see SQF having more and more functions added, this came to my mind. The title explains everything. Obviously the engine has the path finding algorithm built in, as it finds paths for AIs all the time. Other mission/mod creators might find it useful. Would it be possible to expose it to the SQF scripting language? Or are there huge problems, like it must be called asynchronously because it takes a lot of time? There are some solutions implemented already in SQF but the engine-based solution will work faster for obvious reasons. Thanks!
  12. I couldn't find any script/tool/addon which would provide such a basic functionality: how much is ARMA's SQF scheduler loaded right now? Can I run a check every X milliseconds, with the given other scripts running in background? When we are spawning many different SQF scripts with infinite loops we want to know how much the SQF scheduler is loaded with these scripts. Typically we would try to force the scheduler to execute a simple script and observe the delay between spawning the script and its output. For instance you could spawn a lightning with Zeus and estimate the delay. How it works: The script I suggest uses the same idea: it runs an infinite loop, where it tries to execute a piece of code as often as the scheduler lets and measuring time intervals between executions. If the scheduler is flooded with many heavy scripts then sequential executions of our tiny script will happen less often. The script also performs exponential averaging of measured intervals. What it shows: It shows the delay between executions described above. The delay can help us understand the maximum frequency at which your loops can run(average). It also outputs the overload, measured in percents, which is equal to: overload=100*game FPS*delay=100*game_FPS/SQF_FPS. So if the "overload" is around100%, SQF scripts are running at roughly the game's frame rate. If it is 200%, then spawned SQF scripts are running at half the game's frame rate, and so on. Why not just measure performance of single scripts with standard methods? Maybe you are running other people's scripts in the mission, or you want to get an overview, how much load all the scripts produce. How to use it: Run this locally: Measurements will be output to system chat. You can fine-tune _a to perform smaller or bigger averaging (smaller _a -> bigger averaging). [] spawn { sl_prev = time; private _i = 0; private _ds = 0; // Smoothed value of delta private _a = 0.009; // private _N = 30; while {true} do { sleep 0.001; private _t = time; // Get current time private _delta = _t - sl_prev; // Calculate time passed since previous measurement sl_prev = _t; // Store the new time back _ds = _a * _delta + (1-_a)*_ds; // Exponential averaging //Output value every N measurements _i = _i + 1; if (_i == _N) then { _i = 0; private _overloadPercent = round(100*diag_fps*_ds); systemChat ((format ["Scheduler delay: %1 ms, overload: %2", round(_ds*1000), _overloadPercent]) + "%"); // Calculate new value for N to produce outputs every 0.5 seconds _N = ceil(0.5/_ds); }; }; If you want to quickly test the script here's a dummy load you can use. Just run it a few times: for [{_i=0}, {_i<5}, {_i=_i+1}] do { [] spawn { sleep random 1; while {true} do { sleep 0.5; private _a = []; //6 ms @ Core i7 3770 _a set [70, 0]; _a = _a apply {2}; { private _sum = 0; { _sum = _sum + _x; }forEach _a; } forEach _a; }; }; }; Please share what you think about this!
  13. AI Driving - Feedback topic

    Yes indeed. Since they can't do this I was forced to make a simple controller for that, might be helpful for you. It checks the maximum separation between vehicles in convoy(that is, maximum distance between pairs of cars following each other). Then if the distance is lower or higher a threshold it changes the lead vehicle speed linearly. Those who know the control theory might implement a PID controller for that.
  14. AI Driving - Feedback topic

    If that vehicle is a part of multi-vehicle convoy group, try to select a new leading vehicle, if you notice that the convoy has stuck. It helped me a lot. Also I wouldn't do "forceFollowRoad true", because then they won't be able to drive around an obstacle in the middle of a road.
  15. AI Driving - Feedback topic

    Here's a script I made a while ago to prevent friendlies driving over friendlies to death: private _unit = _this select 0; //: Object - Object the event handler is assigned to. private _source = _this select 3; //: Object - The source unit that caused the damage. private _instigator = _this select 6; //: Object - Person who pulled the trigger if ((side _source == side _unit) && /*(_projectile == "") &&*/ (isNull _instigator)) then { 0 }; Add it to the "HandleDamage" event handler with addEventHandler of the unit you want to protect from vehicle collision dammage. Now, if you do want a specific soldier to be killed with a vehicle... just don't give him this event handler, or remove it from him.
  16. AI Driving - Feedback topic

    How exactly does the magic 'modular' design going to revive the AI? After all, the AI 'module' will have to do exactly the same things. Get data from same kinds of entities, analize what's going on around, and perform the same actions as now. Maybe the new engine's AI will be build not with FSMs but with something else? Of course if they redo the AI from 0 it might improve things, because, as you said, the core of AI has been there from OFP times. As I see it, arma's AI is pretty good, apart from some cases when it severely fails.
  17. Now it should be used like this in mission.sqm: briefingName="@STR_MISSION_NAME_SQM"; Actually now there is a hint in the mission editor saying how to use it, if you put the cursor over the mission name field.
  18. AI Driving - Feedback topic

    I've found another condition when a vehicle gets stuck: if you order an inf. group to board an assigned vehicle and assign a "MOVE" waypoint to the group at the same time. They won't even attempt to turn on the engine. But if you add the waypoint after they have boarded the car, they start moving instantly. Steps to reproduce: //m0, m1 are guys of group g0. //"car" is the vehicle m0 assignAsDriver car; m1 assignAsGunner car; [m0, m1] orderGetIn true; // If you execute the second part (below) after they get in, the car starts moving fine wp0 = g0 addWaypoint [[2700, 3800, 0], 0]; wp0 setWaypointType "MOVE"; g0 setCurrentWaypoint _wp0;
  19. AI Driving - Feedback topic

    Apart from all the things already mentioned, sometimes AI ground vehicles are stuck so much that they don't even switch on engine to move to the waypoint. Happens both whis wheeled and tracked vehicles. I have found out that assigning waypoints without index (groupName addWaypoint [center, radius] instead of ... addWaypoint [center, radius, index, name]) reduces the chance of such event a lot. Anyway they still might get stuck. Also I've found that if they get stuck, assigning a new leader usually solves the problem. I guess that their brain gets reset or something like that. If this is not going to be repaired any time soon, maybe we can get at least some commands to debug what's going on in their brain? Like why he doesn't want to move: because his pathfinding has failed or something else. Or even better, some commands to reset their FSMs, redo the planning, etc. Really, my attemts to make them move sometimes remind me of something like this:
  20. Thank you for making this! I've been using your plugin for a while and I must say it does a great job!
  21. Hello! As ARMA's scripts spawned inside the scheduler behave as separate threads, it seems logical to me to have a way to estimate how much percents of total scheduler run time is taken by every spawned script, similar to what a task manager in any OS shows for started processes. Am I missing something, or is there really no such feature in ARMA? Thanks!
  22. How do you check if they are disturbed or not? You can check their behavior with this command: https://community.bistudio.com/wiki/behaviour Their default behaviour is usually AWARE. After they have spotted an enemy, it changes to COMBAT. Once they forget about their enemies(~ 3 minutes when the enemy is out of sight, must be fastrer if they just kill all enemies) the behaviour switches back to AWARE and you can order them to get into their vehicles.
  23. Yes, unless the script has been suspended with waitUntil or sleep. Now, If I have the majority of scripts suspended and a single one running an intense piece of code, the scheduler will mostly spend time executing this script while quickly checking and moving other suspended scripts to the top of the queue. So, I could say that 99% of scheduler's resource is utilized by this single script. Did I make a mistake anywhere? Of course a different way to estimate scheduler load is to measure execution time of everything between suspension statements, but it would still be hard to see the whole picture of how multiple scripts run concurrently. What If I have scripts that are made by other authors and I want to see their impact on performance? Another method I'm aware of is to spawn some script that does something and see how many seconds it takes the script to reach the top of scheduler queue. It is very strange that such a feature is not built into the game engine, considering that we have some diagnostics functions. I hope it will be implemented in the future.
  24. I've tried the script and I like it very much! Well done!
  25. Vcom AI V2.0 - AI Overhaul

    I think there is a Github repository for VCOM: https://github.com/genesis92x/VCOMAI
×