Slow motion syndrome in multiplayer, on Advance and Secure server

[stupor-mundi]

I had stopped playing for many months, due to what I call Slo-Mo syndrome. This generally happened to me on Advance and Secure, but pretty much killed the game for me, since EFront is usually empty and Finnish is quite boring.

Now I have a different monitor and tried again, 2 days ago, and to my surprise the tank behaved normally, so I thought, yay, either it had something to do with monitor sync, or maybe a new version had fixed the issue. But now within those 2 days, it's back, and unplayable again.

I'll first describe what SloMo syndrome is, and what it's not:

It's basically slow motion of the tank's actions, as they have relevance in a multiplayer context.

I.e. moving, turning, turret motion, gun movement, hatch movement, everything like that.

What's not affected is actions like looking around, that works in the normal way/speed.

What I also don't mean by SloMo syndrome is low FPS. When I get low FPS because my hardware isn't so hot, I don't blame the game, that's fine and unavoidable.

What's part of SloMo syndrome also, is that the tank is slow to react to user input, and really hard to control on a hill. Basically like having a heavier tank with a lot less horsepower that you have to rev really high in order to do anything. And then it just slides around like a supermarket trolley.

Essentially it feels like running on the beach, feet sinking into the sand, tits bobbing in slow motion, in Baywatch.

Usually when I mentioned this SloMo syndrome in chat, people seemed unfamiliar with it, and thought I was referring to FPS. From that I deduce that most people don't experience this.

 

So now that it was fine for 2 days, and then back to SloMo syndrome, my curiosity is peaked.

Clearly it can't be something inherent, either in the game in multiplayer, or in the setup of Advance and Secure server.

Hence this thread, to see if other people experience this, and how it can be that it's gone for a short time, and then be back...

 

 

Edited December 20, 2022 by stupor-mundi

[stupor-mundi]

Update: I just tried a Normandy mission on Advance and Secure, on an empty server with some AI around, and my tank had normal performance.
Earlier, when I had SloMo at Engelskirchen, there were around 8 or so players.
This reinforces my earlier perception that server load may have something to do with it.
However, 2 days ago or so, where I had no SloMo effect on Engelskirchen, there were some players online, maybe 5 or 6.

So certainly this is not proportional to server load.
Even if server load is involved, I think it's unacceptable for a multiplayer game, but especially for a sim, that vehicle perfomance should be affected by such factors.
 

[[SN]_Reaper_]

I have this manifests itself when there may be too many players on the server.

Play becomes uncomfortable. You're like wading through thick water. FPS at the same time sags, but not critically.

[dgiatr]

guys your are not alone, the same happens in Finnish while flying with plane over crowded places ( planes, tanks, fires, explosions , spawning points,e.t.c)  when there are many players in the game. Sometimes it happens while keeping high fps. In Vr mode its more obvious.

[moustache]

it looks a bit like the "slo-mo" phenomenon in solo when there's too much unity and all, right?

[super-truite]

not sure what this slow motion bug is.
I know that AAS can cause problems  due to the concentration of people at the same spot.
My guess is that the max number of players is around 50-60. But for numbers below 15 this should not be an issue.
The server hardware  is not a problem I think, because  Dserver (the executable used to launch a server) shows nice performances.
Did you try to switch between full screen and windowed mode ? I have some performance  issues sometimes when Alt tabbing and going back to the game, but not slow mo I would say. I also have some performance issues when  using the brave along side the game (would probably be the same with any chrome-based browser)
 

[ickylevel]

I think it is linked to the number of supply trucks present (among other things). The more flags are captured , the more it increase. I believe I addressed that in the vernon and Dieppe maps (but we need to test it with more players).

Edited December 21, 2022 by ickylevel

[stupor-mundi]

Well the explanations of it being related to player numbers and supply trucks and such things seem to make sense, and I'm glad there are other players who have experienced this, but I'd like to point out that I had this effect countless times, when other players in the vicinity clearly weren't affected.

So it can't be a server-side reason on its own, it has to combine with something client-side, where the clients are different in some respect.
 

A detail I haven't mentioned:

While currently on AAS I don't get a km/h display in the HUD, but some months ago that was still visible, and the km/h that was shown was the speed that the tank would have if it wasn't suffering from the SloMo.

So for example you might want to drive through a forest at 12 km/h in order not be wrecked by invisible trees. So if you're driving at shown 12 km/h you might actually be doing 3 km/h. Or you might drive what's 12 km/h in reality, with the slo/mo, but what's shown is much faster. When you then hit something, you get wrecked based on the shown, 'virtual' speed, not the actual.

 

To me, this shows that it would be not very hard for the game to detect the discrepancy between the actual speed based on the computer clock, and the screwed up 'simulated' speed which it shows. I.e. it should be possible to calculate and display the discrepancy, as a factor.

 

What really bothers me is that clearly we have a game/sim engine where you get a vehicle performance degradation, on a gradient, determined by vaguely guessed at factors, and no clear metric shown anywhere how much of this is happening at any given point of time. I can only detect this in a T34 because I'm used to it's perfomance, as it should be, within the game. But what about a new player who is just trying out some vehicles? He would have no idea that this is going on.

 

What a sim should do, if there is some kind of bottleneck, either on the server, on the client, or the client-server interaction: to degrade something, typically FPS, but also LOD, in order to get enough compute as necessary to overcome the bottleneck. And then, if all that doesn't cut it, show/display clearly that it is not accurately simulating anymore. This whole affair with underperforming vehicles is a joke in my view.

 

[stupor-mundi]

6 hours ago, super-truite said:

not sure what this slow motion bug is.
I know that AAS can cause problems  due to the concentration of people at the same spot.
My guess is that the max number of players is around 50-60. But for numbers below 15 this should not be an issue.
The server hardware  is not a problem I think, because  Dserver (the executable used to launch a server) shows nice performances.
Did you try to switch between full screen and windowed mode ? I have some performance  issues sometimes when Alt tabbing and going back to the game, but not slow mo I would say. I also have some performance issues when  using the brave along side the game (would probably be the same with any chrome-based browser)
 

 

 

21 hours ago, [SN]_Reaper_ said:

I have this manifests itself when there may be too many players on the server.

Play becomes uncomfortable. You're like wading through thick water. FPS at the same time sags, but not critically.

 

 I would agree that the SloMo was generally more prevalent when there are many people on the server. But clearly, some people like super-truite never experience it, even when there are many people on the server. If you have encountered the effect, you would recognize it from the description straight away.
So I find it quite strange that Dserver shows nice perfomance in such a situation, since it would only make sense that more people on the server cause this, if the server actually approaches some limit. Weird.

I've also suspected some connection with alt-tab. I've had many situations where I started a mission without much SloMo, had to alt-tab for some reason, and after alt-tabbing back into the game, the SloMo was there. Almost as if the game wants to punish you for alt-tabbing.

What's also curious regarding alt-tab, is that the vehicle's performance seems unnaturally enhanced while the game is not in foreground. By which I mean, not just that the SloMo is gone, but that the vehicle performance is actually better than the reference, i.e. offline performance.

 

[stupor-mundi]

Update:
I've since been on a bunch of missions without noticeable SloMo, with larger numbers of players, for instance Prelestnoye with about 15 players, and some Normandie missions with 10+ players.

So some other factor seems to play a bigger role.

Just now I tried Engelskirchen with 9 players, and noticed SloMo and quit. It seems something about the Engelskirchen mission tends to produce the SloMo effect at relatively low player numbers.

 

 

[super-truite]

interesting, I remember there was some problems before on that map. Could it be linked to the weather ? The weather is changing randomly from one mission to another

[dgiatr]

I think its a combination of the following factors :

1. Number of objects (planes, tanks, airfields, smokes, explosions, clouds, e.t.c., ) that appear in a certain place of the map during a specific time

2. Number of players at that place or around there

3. How strong is the P.C of the player who suffers from slow motion. For high tech PCs you suffer less from slow motion. if you are CPU bottlenecked you will probably suffer more. By the time i overclocked my cpu from 4.5 ghz to 5 ghz, i suffer less from that. But either way there must be many players inside the server for me to happen.

[stupor-mundi]

On 12/27/2022 at 7:19 AM, dgiatr said:

I think its a combination of the following factors :

1. Number of objects (planes, tanks, airfields, smokes, explosions, clouds, e.t.c., ) that appear in a certain place of the map during a specific time

2. Number of players at that place or around there

3. How strong is the P.C of the player who suffers from slow motion. For high tech PCs you suffer less from slow motion. if you are CPU bottlenecked you will probably suffer more. By the time i overclocked my cpu from 4.5 ghz to 5 ghz, i suffer less from that. But either way there must be many players inside the server for me to happen.

 

All those factors are certainly valid.

I would add that low network speed also has very bad impact. 1 1/2 yrs ago I lived in a place with wifi-only, no ethernet. And there I got some SloMo effect even on Finnish, which doesn't happen to me today, with the same laptop.

 

But since I can often play on AAS without SloMo, even with lots of players in the near vicinity, it's also clear that there are other, more hidden, and seemingly random, factors.

 

I just tried Engelskirchen again with 10ish players and heavy SloMo. Whereas in recent days I didn't get noticeable SloMo in Yuznaya. And those missions seem quite similar to me with the narrow valley and steep slopes. I had hypothesized that the absent or weak SloMo in Yuznaya could be explained in that way, that the space segmentation due to the hills might help the game reduce cpu. But clearly this doesn't hold for Engelskirchen.

 

What I don't get is why people, if this phenomenon is indeed no surprise, and known, to many players, why they are not bothered by it more. In an online, multiplayer, sim.

I assume the same thing must be going on with the planes. And I can only clearly detect this in a tank, since tanks are so slow to begin with, and you have plenty of visual controls and accoustic feedback from the engine. So you can detect SloMo without having to rely on the instruments, which would lie to you.

OTOH in a plane I'm not confident I would be able to detect if my perfomance was degraded by 10, 15 %, whilst ignoring the misleading instruments.

 

On 12/27/2022 at 1:10 AM, super-truite said:

interesting, I remember there was some problems before on that map. Could it be linked to the weather ? The weather is changing randomly from one mission to another

The recent times I was getting lots of SloMo in Engelskirchen, the weather was quite clear, with a few clouds only, like most other missions.

[[SN]_Reaper_]

21 минуту назад, stupor-mundi сказал:

What I don't get is why people, if this phenomenon is indeed no surprise, and known, to many players, why they are not bothered by it more. In an online, multiplayer, sim.

It's disturbing. Both the tankers and the pilots. Recently there was an event on the Russian server and some of the players complained about slo-mo (Russian for kissel). In discord there was a discussion. I can give you the link, try to read through the translator, copying messages. 

[dgiatr]

2 hours ago, [SN]_Reaper_ said:

It's disturbing. Both the tankers and the pilots. Recently there was an event on the Russian server and some of the players complained about slo-mo (Russian for kissel). In discord there was a discussion. I can give you the link, try to read through the translator, copying messages. 

Intresting about the discussion in discord, please send us the link whenever you can. Thank you

[stupor-mundi]

'nother update:

I rejoined the Engelskirchen mission later, when player count had dropped to around 5.

At that level, I had no SloMo for a while, but later a phenomenon which I had seen before sometimes: Jumpy speed changes, very hard to control the tank and drive through the forest without hitting something.

It was as if the game engine detected it was making my tank too slow, and then overcompensating, back and forth, a few seconds too slow, then a few seconds too fast.

[[SN]_Reaper_]

Ok. Here's a link to the beginning of the discord discussion.

https://discord.com/channels/1030018584556097576/1030077482835128370/1055568210738548806

 

From the post Zommer - 12/22/2022 22:31 and on down. Pay special attention to the comments of -DED-Rapidus.

In brief, this phenomenon can depend on many parameters. It can be a bottleneck on both the server side and the client side. It doesn't depend on the power of your computer. It can be mission related, map related, having a lot of objects, etc.

[stupor-mundi]

'nother update:

I was just on Prelestnoye with around 15 players and initially no SloMo, for a long time. Then the flag moved, the enemy got it, which left me quite far behind enemy lines.

At that point very noticeable SloMo started, and didn't go away until I quit.

Because the flag move left me so far behind enemy lines, the number of players and AIs anywhere near me actually decreased a lot at that point. There were some plane AIs around but they had been there before.

So it felt as if the flag move had put the server into a different 'mode' or something. As if the server was doing lots of other stuff and couldn't be bothered to spend much time communicating with tank clients. Just speculating.

Anyways I thought it was interesting that the slomo set in at a point in time where I was suddenly all alone in a large patch of terrain.

 

 

 

 

[stupor-mundi]

20 hours ago, [SN]_Reaper_ said:

Ok. Here's a link to the beginning of the discord discussion.

https://discord.com/channels/1030018584556097576/1030077482835128370/1055568210738548806

 

From the post Zommer - 12/22/2022 22:31 and on down. Pay special attention to the comments of -DED-Rapidus.

In brief, this phenomenon can depend on many parameters. It can be a bottleneck on both the server side and the client side. It doesn't depend on the power of your computer. It can be mission related, map related, having a lot of objects, etc.

 

I'm no good at Discord, though I've got an account. Somehow for me the link just shows an empty page.

At any rate, it's interesting that people in that discussion came to the conclusion that the client computer's power is not what it depends on.

I had thought that the client's performance must be involved, since some players are not even familiar with the topic, i.e. it never happened to them, and that must be explained somehow...

[super-truite]

23 hours ago, [SN]_Reaper_ said:

Ok. Here's a link to the beginning of the discord discussion.

https://discord.com/channels/1030018584556097576/1030077482835128370/1055568210738548806

 

From the post Zommer - 12/22/2022 22:31 and on down. Pay special attention to the comments of -DED-Rapidus.

In brief, this phenomenon can depend on many parameters. It can be a bottleneck on both the server side and the client side. It doesn't depend on the power of your computer. It can be mission related, map related, having a lot of objects, etc.

in which discord is this ? the link does not work for me. I would like to take a look to see if there is some recipe to avoid it, because again, the server performance is not bad ( tick delay and SPS most of the time in check)

[stupor-mundi]

'nother update:

was just on Yuznaya at around 11 players, with some SloMo, and clear skies.
I think I had been on Yuznaya recently with around that number of players and no SloMo. Pointing towards other, mystery factors.

[[SN]_Reaper_]

@super-truite, @stupor-mundi

there is indeed a problem with access.
I had to leave the Russian-language IL-2 server to check it out. First, you need to connect to this server. 

https://discord.gg/gtzU3x8RN6

 

But at the moment I can not pass the verification

It may be easier to post links directly in the AAS channel on discord.

[stupor-mundi]

'nother update:
earlier I was on Yuznaya with medium player numbers and no SloMo for a long time.  Then more players joined, at around 15, and SloMo became pretty bad.

So I think, other things being equal, player numbers certainly have to do with it.
I considered the theory that computer performance doesn't play into it, but I don't agree on that. In 2020 for a while I had a proper, fairly fast pc, and I think experienced no SloMo in that time.
A few things I forgot to mention when describing the SloMo:
Not only are your vehicle's motions affected, but also things like shooting, reload times, and repair times. Basically everything that plays a role in player to player interaction.
Another thing: At the time when the bumpy ground, or rather, the bumpy tank riding on the bumpy ground, was introduced, that made the SloMo a lot worse. It feels as if there is some bad synergy between the two, causing the tank to shift into lower gears than it should have to. It's as if the cpu can't keep up with doing those bumpy riding calculations, causing the tank to go even slower. But that of course doesn't cut it as an explanation, since it only happens online, and only on certain online missions.
 

[stupor-mundi]

'nother update:

earlier I was on Engelskirchen where initially we were only 5, no SloMo, all friendlies.
Plenty of AIs of course, and those did not seem to cause any SloMo. Then more players joined, all enemy (mostly tanks), to a total of 10, at which point, heavy SloMo.

This gave me the idea, could it be that it's not so much about the total number of tanks, but about the number of enemy tanks?

Although from a theoretical point of view it doesn't make that much sense, after all the friendlies also need to get all the state updates.

Then I went on Finnish which had all of SCG on it, it seemed, so, an unusually large amount of tanks, and enemies from my point of view since I was red.

Finnish was nearly maxed out player wise, and there was heavy SloMo. I hadn't experienced any SloMo on Finnish in many many months, even when it's maxed out, so that was interesting. Blue had way more tanks in there than we did, so maybe there is something to that hypothesis.

I could imagine for instance that the engine prioritizes enemy state updates over friendly updates, that could maybe explain it.

There were many on red that were complaining about 'lag'. I suppose people don't distinguish between lag and slomo.

Edited January 1, 2023 by stupor-mundi

[stupor-mundi]

On 12/28/2022 at 8:52 PM, [SN]_Reaper_ said:

It's disturbing. Both the tankers and the pilots. Recently there was an event on the Russian server and some of the players complained about slo-mo (Russian for kissel). In discord there was a discussion. I can give you the link, try to read through the translator, copying messages. 

 

I managed now to read the discussion, but understanding only fragments, since a lot gets lost in google translate.

I gather what I call SloMo is translated by google from Russian as 'jelly'.

There seemed to be a consensus that 'complex triggers' were a big factor. Maybe that can be some help to server admins.

DED Rapidus mentioned about some (client) settings in startup.cfg, such as client_download_traflimit.

But I didn't get the details about how they should be calculated. Mine are all '1000' for this group of values, can't recall ever changing them, are probably default.

 

I just had really weird SloMo related outcomes on different missions.

First, one of the Normandie missions, with close to 10 players, and no SloMo.

Then Yuznaya, which is usually relatively good for not having SloMo, at only 4 players, gave me lots of SloMo. Super strange.

Just for the hell of it, I quit IL-2, turned on a VPN and tried again, in the meantime the players had changed a bit, but were actually more, and SloMo was much reduced.

You would expect a VPN to make SloMo worse, so I found that pretty sketchy.

 

 Remembering Rapidus' statement about settings, I began to wonder, could it be not just number of players, but differences between player's client setup, the combination of how fast their connectivity is, combined with what network config they have? So that strong SloMo can occur if some player is connected who has some extreme config?

 

The next time I get SloMo  again I'll try the VPN again. It doesn't make any sense to me that the outcome with VPN should be better, wonder if the fact that in this way I had changed my IP, is a sign of something fishy going on.

 

[stupor-mundi]

Just now again, I was on Tessy, and then Yuznaya for a bit, interesting comparison maybe.

Tessy at 5/6 players had no SloMo, at all, tank moved perfectly, and as soon as Yuznaya started, at only 2 players, slight SloMo, and at 5 players, quite pronounced SloMo for me. Seeing as Tessy has quite a lot AIs, as does Yuznaya, the AI numbers don't seem to be a deciding factor.

So Tessy and Yuznaya don't seem radically different from a players point of view, but maybe there are intersting differences regarding triggers or something.

Anyways, from a player's point of view, it makes the game engine look a bit dodgy, when, given the same client hardware, same connectivity, same player numbers, two seemingly similar missions give radically different tank perfomance and input responsiveness.

While my FPS aren't high, I get the same FPS on those missions, so it doesn't seem as if the game client would try to lower the FPS in order to get the same, simulated, tank performance.

In this comparison between Tessy and Yuznaya, but also something that tends to be the case often with SloMo, in Yuznaya the 'Outdated data hit received from .. xyz seconds.' or similar, message, is displayed a lot. So this message is maybe misleading on some level, since everything else was the same, and only the mission changed, and this message appears which seems to be saying something about connectivity, but the cause must be something else.

 

Edited January 6, 2023 by stupor-mundi

[stupor-mundi]

After attempting some client fixes and comparing various missions on different servers, I'll sum up a bit:

 

• VPN didn't turn out to help that much, it did seem to make the steering less mushy (delayed), but without fixing the actual tank performance. Still strange because VPN should just have a negative effect and somehow doesn't.
• The total absence of slo-mo when the game client is backgrounded (alt-tab) is intriguing, but possibly due to some necessary client-server state updates being skipped. No idea, but no way to get to the bottom of this as far as I can see.

 

As a model for what's going on, I suggest the following:

 

2 Aspects to be considered, each of which can become the slo-mo cause:

 

1) Client power (probably single thread cpu performance, gpu not so much, network connectivity)
vs
Number of other tank players in the greater area.

I.e. even on powerful servers with 'ideal' maps, slo-mo will eventually occur on my (weak) laptop, once tank player count becomes real high, let's say approaching 30 players. Meaning, even if everything server-side is as ideal as can be, the IL-2 netcode will create slo-mo problems for at least the players with weak hardware, when approaching 30 players in the area.

 

2) Client power
vs
Number of other tank players
vs
Server distractedness by features of the map/mission
On AAS, the total (tank) player count at which slo-mo sets in for me, is very different from mission to mission. On some missions, things can get quite bad at 3 players, whereas other missions were still ok at 8 players. Other players, with presumably strong hardware, seem to be able to carry on without slo-mo at those player numbers where it becomes unplayable for me.

 

Regarding, which aspects of maps/missions will distract the server, it doesn't seem to be something obvious/visible such as number-of-ais. Only the mission maker will be able to know which might be the crucial factors within the maps.

 

Based on this model, the problem arises: the mission designers will typically be on fairly strong hardware and not notice causing slo-mo in their mission/map making.

 

What would urgently be needed is a measure of simulation faithfulness, basically a factor which ideally should be 1, and which, at halved tank performance, would be 0.5. This factor could be displayed together with FPS, and could also be transmitted to the server, so that the map/mission maker could take that info into account.

 

The current situation where you effectively have players driving around in vehicles with radically different performance, determined by the power of their computer hardware, is absolutely appalling for what is called an online mupliplayer 'sim'.

Edited January 13, 2023 by stupor-mundi