[SP] Pilgrimage

[Rydygier 1309]

Boxes are the only thing spawned dynamically, when player approaches buildings. Thus my attention was brought to them, when turned out, CPU may be overloaded with gameplay progress, as this is the only thing, that accumulates with time. Rest is there from the beginning anyway. So the only reason I'm trying to remove them is performance. But now, when you are saying, that problems start with 1.5, and whole island "boxed" in 1.4 made no issues of such kind, then I'll gladly leave them persistent for same reasons you want them not disappearing. Same for bodies.

Still, maybe I'll be able to implement properly working additional despawn/respawn caching to gain more FPS and relieve CPU. Main obstacle there is strange group behaviour after respawning - often tends to go immediatelly somewhere, sometimes far, sometimes at near ally group position, then goes back to normal routine. Very odd, as there is no waypoints nor doMove involved. If I solve that, we could from all this mess take performance improvement - code for that is almost ready after all. If not - oh, well. Primary goal here is to fix problems.

It is also very valuable information, as now I can focus on changes between 1.4 and 1.5. I feel, we are closing to the something particular, I could solve. I'll be back on topic in the morning.

Boat marker - I can make it deleted, when you choose no item options as part of that feature. Same for hideout flag(s).

[khugan 24]

Thank you Rydygier. The mission you created allows a lot of tools that Arma provides to actually be used and needed. FYI I am playing on old Core 2 Duo (no hyperthreading) cpu I build in 2003. The gpu is new though, a Geforce GTX 660. I will upgrade sooner or later.

[doc. caliban 34]

Performance observations:

I apologize for not noticing this earlier, but your mission is just about the only one I've spent any time with since buying the game.

I've been telling you in this thread that I am not having any performance issues, and what I've meant is that the FPS that I get during the mission is pretty steady, other than slight differences based on the map itself.

However, I just tested by starting a new mission in 1.51, walking into the woods, and noting the FPS. (Right around 30 FPS) I then created an empty mission and placed myself in the exact same location and tested again: steady at 60 FPS, so double the FPS in the empty mission.

I don't know what is expected for FPS change in a mission, so I don't know if that's a big impact, or normal.

My CPU is an i7 970 running @ 4.2GHz.

I hope that info is of use! I have not tried the test build yet as I am still at work. (just snuck a quick test during a break)

I will try the same test with 1.41 as well.

- Doc

Edited March 10, 2014 by Doc. Caliban

[doc. caliban 34]

Performance testing:

I understand that there is a massive difference between these missions, but this is what I thought of to test. I am only presenting the info in hopes that it's useful!

Hardware:

i7 970 @ 4.2GHz

2xGTX580 SLI

2560x1600

Software:

ARMA III - Current Steam version

Mods - None

Pilgrimage versions tested: 1.4 & 1.51 (results were more or less the same with both)

Test 1

- Mission: Infantry mission from Showcases - FPS at starting point: 53

- FPS in empty mission at same starting point: 60

Test 2

- Mission: Pilgrimage 1.4 & 1.51 - FPS at starting point: 30

- FPS in empty mission at same starting point: 60 (no mist)

Test 3

Mission: PilgrimageX

- FPS with settings at 100%: 37

- FPS with settings at 50%: 45

- FPS in empty mission at same starting point: 74 (no mist)

I'm not able to do any real play testing tonight though, so I don't know how things will look over time.

- Doc

Unrelated note about my past playing:

Regarding loot boxes, I never left any behind - I would sell all of them as I moved along, so I think they were not accumulating which may explain why my performance never worsens over time in the mission.

[Rydygier 1309]

OK. Thanks a lot Doc. How about these two sidechat values in PilgrimageX - Delay and cycle time, I asked to follow? These are most important to me.

30 FPS while on empty is 60 is expected. Cached, or not - 1000 AIs + hunderds of additional objects must have an impact. SQF scripts however by itself do not impact FPS, thus FPS is secondary in PilgrimageX testing.

[Rydygier 1309]

Continuing our little "dev branch", prepared second "X" version, I'm testing now, so anybody could test with me, if he likes (same test rules, as for first "X" - primary thing is to observe, if these starnge issues are still present):

PilgrimageX2

It is named same, as first X, so just overwrite. New things:

- as for loot, determied, there are problems with backpacks spawned rarely instead of boxes. For some reason backpacks refuse to be placed inside building and sometimes are placed on weird height. No more loot backpacks. Also added small code to ensure, box is positioned precisely in attempt to avoid flying/on roof/burning boxes.

- as for caching - this time enabled experimental, new caching, with despawn/respawn system. Should give at least 10-15 FPS more, but easily maty prove to be unreliable, so intersting for me here is also any odd things happening to encountered AI.

Noticed, that deleting units is considered by group as taking losses, thus observed earlier "I was cached, so now I'm going somewhere far just like that" effect. To avoid it I was forced to set allowFleeing 0 at caching. Unfortunatelly, there is no way to get default fleeing parameter, so I was forced to set it to taken from the head 0.1 value at decaching. Results not tested, so curious also, how much group will after decaching tend to flee. Another sideffects: damaged units and vehicles will be "healed", ammo renewed. May be assumed, that in the meantime they repaired vehicles, took more ammo and treated wounded...

- as for checking 1.4 to 1.5 changes: Found only one possible cause: was added second, parallel onMapSingleClick stacked event handler. So now code is rewritten, so only single such EH loop is needed.

- other things: added removing of boat marker and hideout markers when chosen no item feature, which now is named "hardcore".

Edited March 10, 2014 by Rydygier

[Rydygier 1309]

So far noted one of expected problems with new caching - TL/something at respawn makes subordinates to disembark vehicles. For checkpoint this mean empty vehicle and not manned static weaponry. Not sure, what to do about that, in revous test there wasn't such problems, I'll see more checkpoints, then I'll decide if switch back to previous caching.

[doc. caliban 34]

I hadn't noted the loging info as I thought those were only expected to change over time in the mission, and I didn't have time to play at all last night. Sorry about that! Are you wanting to just keep an eye on the average for both stats? I can probably put a couple hours or so in tonight.

Thanks for confirming that the FPS drop is expected (to that extent) in missions. I figured there would be some, but had no idea of how much of a drop to expect.

-Doc

[Rydygier 1309]

Thanks Doc, if you want to test more - use new version of "X" linked today. Of course, I do not need every value second by second, just roughly typical average and typical changing trends (especially, when these issues show up) and anomalies. If typical cycle time was so far eg 1.5, then constant level above 5-10 would be worhty of notification. Depends, what will be typical for you. Similar for "delay". However in new version I did, what I could to fix any possible problems, so perhaps even there will be not any issues anymore? Hopefully... So far all except these problems with cached checkpoint runs smoothly for me. Very pleasant testing, I must say. :) Using new version check also FPS difference, should be better, I hope.

[doc. caliban 34]

Excellent! I work until late this evening, but will try to get a new game started and put 2-3 hours into it. I'll post my results!

Thanks again,

- Doc

[Alex72 1]

First off, great mission my friend! :)

Through hours of play in this mission ive seen parts of Altis i never seen before which has been superb. The feeling of the mission is great through never knowing where enemies are and what loot to find, and you get so happy when you find useful stuff or manage to take out some enemies. Last night i had to sleep to heal up after a clash and ended up at night time. Luckily i looted NVG earlier that day so i could see. But when searching some churches i saw an armored enemy car with big guns and had to sneak around it for a long time - it was great. Great idea to make use of all the churches as well. :)

Hitch-hiking problem

I do have some issues though that i want to bring up. First, after aquiring a car, everytime i hitch hike i am moved into it no matter where it is on the map. And in my case its busted and got no more ammo so i left it (dont know how to get vehicle ammo, and its totally ok if we cant. Makes it funnier). But the "weird" thing is that i end up inside it even though its somewhere else on the map (i hitch hike to the right place and the car is teleported to that place and im placed inside it). Bit weird so to speak.

Color/light effects?

The other issue im having personally is that i think you use either BIS effect module for contrast/light? Or some effect regarding light/color? Because in your mission the clouds are super white, almost like smudges and when i load up the editor they look normal. I dont know if this affects night time NVG as well, but it might be BIS problem actually, but at pitch black nights without moon when using NVG the sky is super bright and it darkens your vision a lot. I think this is BIS problem actually but maybe if you have an effect on it might make it worse. IF there is an effect in this mission i would love for a version without it.

Disapearing AI

Third, ive seen enemies disapear in front of my eyes. A walking patrol walking away from me maybe around 50-100 meters. Maybe you could make that distance further or make them not disapear as long as they are around the player. At the same time i can see enemies really far away so i dont know if this was a patrol that is specially spawned AI.

Thanks for a great calm and atmospherical mission. Please keep enhance it. :bounce3: Especially if you in the future maybe do this for 2 players also. I understand that might be a whole other beast to make work, but i have been dying to play this with a friend and experience it togehter.

Regards

Alex

[Rydygier 1309]

Hi Alex, thanks for feedback.

(i hitch hike to the right place and the car is teleported to that place and im placed inside it). Bit weird so to speak.

Yet it is intended this way. I assumed, that player would like to fast travel along with his vehicle, not leaving it behind (I would like that way at least...). Simpliest way was to put him inside and teleport the vehicle. Assumed, that situation, when player will be far from his vehicle will be rare, so no problem here. If there is problem though - well, I can eg make distance check, if player is close enough to his vehicle. If not, vehicle stays, player moves.

he other issue im having personally is that i think you use either BIS effect module for contrast/light?

No. Not touched such things here. I only manipulated with daytime and weather conditions. Whatever you see, it is vanilla.

Third, ive seen enemies disapear in front of my eyes.

Probably part of wider range of CPU-overload-like-effect issues I'm struggling past few days. Caching distance for AIs is set to 2000 meters, and that way it works... when it works. Hopefully some of my last changes would help. Difficult to solve, cause many experienced such things, but not me, so I'm acting blindly...

As for coop - making optional second character playable is simple, but I'm focused on SP all the time, so frankly I do not know too well MP scripting. All this locality issues... When mission will be in the final state, then I possibly could think about such extravagance. :)

Edited March 10, 2014 by Rydygier

[Alex72 1]

Alright bud, fair points. Weird on the clouds issue. Its really bright in the sky and the ground is hard to see as it gets darker. its like the HDR is back to ARMA2. Maybe it came in the latest patch, i dont know.

Keep up the great work. ;)

[khugan 24]

I haven't gotten to play much. I started about an hour ago and all seems stable except for one big thing. I found a campsite, cleared it, and when I chose to establish it at the fire, I got message spamming that never stops.

Something like this:

Alpha 1-1 [Alexander Laskanis]: "Delay 16 {Average: 15.0157]"

and:

"Cycle Time: 0.47951"

This continues even if I save, exit, and resume. However, I have no problems with churches updating or loot spawns out of reach. Good fps (40s-50). I'm going to start a new game and avoid establishing camps so I can test further.

EDIT:

The message spamming happened at the campsite even though I didn't interact with the fire.

Edited March 11, 2014 by Khugan

[Alex72 1]

Maybe also make it save when sleeping at camp? Was a bit of a bummer to go back to camp, heal and sleep, move really far and then die - back to the church miles away from camp and having to move to camp again just to sleep and heal again.

Just a suggestion. ;)

@Khugan, i never seen that bug and i played many hours. Maybe you are on dev branch? Not sure if its completed yet. Honestly i havent read this whole thread. Or could be a mod. Or maybe even a mission bug, but i never seen it and i cleared lots of camp.

Which brings me to my next point that i forgot above. Seems there are no more camps spawning for me now. I logged a whole bunch of hours, slept 2 times for healing, done maybe around 200 churches. And since a whole real life day i havent seen any more camps although in the beginning they were plentiful.

[Rydygier 1309]

The message spamming happened at the campsite even though I didn't interact with the fire.

Heh, man, it is not a bug. When I told, I need to know average values of delay and cycle time, I was talking just about these constant sidechats. These was described at first X and this differs test "X" version from normal. Read carefully, please. :) Good at least, you pasted two values. Are these typical? Was any lasting changes with time? Anomalies? 14 of delay and 0.4 of cycle time is quite nice result, if it was typical.

Maybe also make it save when sleeping at camp?

Maybe... But it would be irritating for me, when I eg want to sleep 18 hours (3x rest) and got break for saving each 6. And you can always save manually. I must think about.

Maybe you are on dev branch?

Yes, he is. I thought, I explained this clear enough... I was mistaken apparently. :)

Seems there are no more camps spawning for me now. I logged a whole bunch of hours, slept 2 times for healing, done maybe around 200 churches. And since a whole real life day i havent seen any more camps although in the beginning they were plentiful.

It is quite possible. There are two dozens of hidden camps preplaced at init. And "hidden" means, these caps will be located rather far from urban areas and rather near forests. Now see the map, where you started, and where is most of forests on the island. There are big parts of Altis, where presence of any camp is very unlikely. Most of them usually are located in NW part of Altis, same area, where are most densely set starting points, as there is more good areas for discrete "desant". More - some of you may noted, that sometimes two camps are very close to each other. It is removable sideffect, but I decided to keep that way - this gives slight chance to enounter an bigger encampment composed by two small camps. Makes things a bit more diversed/interesting IMO.

Edited March 11, 2014 by Rydygier

[khugan 24]

Sorry, I never tested first X since it deleted loot, or so I thought. So I never really read. I am going back to read now though. I'm playing stable branch, and yes those are typical values.

EDIT:

Yes your posts were quite clear.

Edited March 11, 2014 by Khugan

[Rydygier 1309]

I see... Not a problem. If it is typical - very good then. It looks optimistic, as there wasn't earlier issues in "X" reported even once so far. That is most important part. Ahead of me only last tests of new caching to determine, if use it, or switch back to previous. More new things to come:

- at init Alex' weapon (and his companion) will be randomized from 3 available SMGs. Randomized will be also uniform from 10 chosen.

- will be added distance condition to determine, if player should fast travel (hh or return to hideout) with or without assigned vehicle (250 meters or closer to the assigned vehicle needed to move it along with player. If farther - vehicle stays, player moves on foot...).

Saves at rest for now not implemented, as mentioned, may be annoying and there is manual save always available.

Edited March 11, 2014 by Rydygier

[aseliot 2]

Does dying and then pressing end delete savegames?

[Rydygier 1309]

I think so. Arma is terrible regaridng to saved games management. No idea, why. Autodeleting saves, no control which save will be loaded at mission resume...

BTW I rewrote new caching and looks promising so far.

[tortuosit 486]

Hmm, my AI mate and I wanted to kill enemies at a campsite. Well, my mate died, we killed two. I thought there was nobody left and had a look. Suddenly I realized there was an AI enemy proning, but doing nothing. I killed him. Then I saw someone enemy was standing pretty close, aiming at me, but not shooting. Strange. My mods: "@JSRS2;@sthud_a3;@status_hud;@tao_foldmap_a3;@tmr;@tpw_mods;@asr_ai3".

[khugan 24]

Ok, I've been playing for around 6 hours and I can report no problems. The delay average moved between 11.5 and 12.5 and the cycle time 0.3 to 0.4. In the last large city I was in, going in and out of lots of buildings, the delay reached 13.3 average and cycle reached 0.5. I'm playing on stable with JSRS, bcombat, ASRAI, Blastcore, VTS Weapon resting, and of course CBA.

[Rydygier 1309]

Thanks Khugan. Good to hear. :)

Hmm, my AI mate and I wanted to kill enemies at a campsite. Well, my mate died, we killed two. I thought there was nobody left and had a look. Suddenly I realized there was an AI enemy proning, but doing nothing. I killed him. Then I saw someone enemy was standing pretty close, aiming at me, but not shooting. Strange. My mods: "@JSRS2;@sthud_a3;@status_hud;@tao_foldmap_a3;@tmr ;@tpw_mods;@asr_ai3".

Hm. This time, I think, this isn't me. Not saw such thing. Is this X2 perhaps? There was some glitches in caching, removed now, that possible may cause some strange things.

New version is close, I hope. I think, I'll make optional, which caching system to use, new one as default, also I'll try to add some hostile mortars run by my FFE script, as option (disabled by default) for people, that like to die from time to time without any warning... :)

[tortuosit 486]

I think the non reacting enemies is something about the 2 AI mods I use. Maybe I'll replace them with bCombat again, if it works performance wise.

Haha, seems like you are really obsessed by dying from a mortar attack...

[Rydygier 1309]

Obsessed? Nah, just little death wish maybe... :) Not that bad in fact. Artillery should be set at low accuracy (not very skilled gunners), and, unless you play hardcore mode, there will be FFE's markers on map, so after first shot you'll know what fires, from where, when and where next impact is expected. Not very realistic this way, but otherwise it could be too frustrating, I guess similar, as being under shelling in RL is, and, as we now, realistic war isn't funny at all. :) Still, if someone needs more adrenaline rush, this feature should provide it.

Edited March 12, 2014 by Rydygier