PWCG Server Overload with Good Specs and 'reasonable' settings.

[kingstreet73]

I just saw a YouTube video uploaded a few days ago where a guy shows how to setup PWCG and he launches his game with all HIGH simple settings, and does a test run perfectly fine.
However, anything other than default low/medium settings on PWCG simple settings causes me to get server overload settings, followed by rubberbanding and maybe even slowdown's over time, whether it be by myself or playing with a friend.
Any idea how to fix / what the culprit is?
This is also I /think/ a new issue after I updated PWCG because I used to play on better settings in the past and didn't even know server overload was an issue so I'm not sure what's causing this now.

I know server overload is when SPS/Tick Rate dips due to too many units, which is why I assumed it's normal to get bad performance with too high settings, even with a good computer - after a certain point hardware doesn't matter and the game will just not be able to run well.

But, as stated previously, this guy made a mission with all simple settings at high (and said to lower if needed), and showcased the mission with 0 issues. I have also seen other people claiming to be able to run at high settings and only when they put it absurdly high (using the advanced settings and altering min/max planes allowed per side) do they get issues.
Objectively my specs should be more than enough to run the game relative to the youtuber, so I don't understand how he can run a coop campaign flawlessly and I can't without gimping my server settings.

His specs: GTX 2070, i7-7700k, 16gb ram   ------------ My specs: GTX 3080, 5900x, 16gb ram, also games on a fast NVMe SSD, 480download/25upload speed so surely bandwidth isn't an issue for solo/duo coop.

Again, I know after a threshold, specs are irrelevant and too many units will cause issue - but surely there's something wrong if two people using the same settings get conflicting performance - as in Person A gets better performance than B when B has better hardware. I'm just struggling to figure out what's causing the disparity.

[PatrickAWlson]

PWCG has never run smoothly on high with most systems.  My system is a an I9-9900, 32 GB RAM, 2TB SSD, 2080 Ti.  It doesn't stink.   I do not run on high air activity.  Never have.  

 

[71st_AH_Rob_XR-R]

I suspect that the biggest difference is your CPU. The i7-7700k Turbo speed is 4.50 GHz out of the box before tinkering.  The AMD gets better performance but to does it by using a lot of cores and iL2 doesn't take advantage of lots of cores.

 

Try running with performance monitor on and I think you will see that the bottle neck is your CPU.

 

If it is, try running it using DServer on an old PC and I think your performance will improve.

 

The other thing to investigate, you may have the same RAM but he might have a larger page file, if your CPU is not the bottleneck and the RAM is.

Edited October 23, 2021 by 71st_AH_Rob_XR-R

[kingstreet73]

9 hours ago, 71st_AH_Rob_XR-R said:

I suspect that the biggest difference is your CPU. The i7-7700k Turbo speed is 4.50 GHz out of the box before tinkering.  The AMD gets better performance but to does it by using a lot of cores and iL2 doesn't take advantage of lots of cores...

It shouldn't be - out of box my CPU boosts to 4.8 on most of my cores - and with the new PBO settings I reach 4.9-5.1ghz single core speed so even though the game doesn't use all my cores, my single core speed should be more than enough. 
I tried running a Dserver and setting it to only use a specific set of cores to make sure it wasn't being overloaded with running the game and server on same cores/threads and it still had the same issue.

I'll do research on paging file but I currently have 26gb's set currently which seems normal so I don't know if that is it but I'll figure it out. Also his motherboard (MSI Z270, sorry for not including before), can only support a max of 3600Mhz ram, which is the same speed as mine. Latencies are unknown but I really don't think there would be a massive difference.

Anyways - with what PatrickAWlson said, I had the same notion from my own experience and almost everyone else: PWCG is just hard to run on high settings. Only reason I'm writing the post now is because of the video I saw where that was disproven so I was curious if there is actually a way to run it on high.
For most of the people that claimed they can run PWCG on high, I disregarded and took them as either exaggerating or not telling the whole truth, but seeing video of him running everything on high with 0 issues made me question if truly there was just something wrong that is making people get mixed performances.

I guess the only way to prove this is to have someone else show they can run on high simple settings with no issue besides that video I saw, but until then...

Edited October 24, 2021 by kingstreet73

[Nocke]

I think if you really want to compare, you need to use always exactly the same mission. There is a lot of randomness in the PWCG missions. You can generate a mission on all high, and be just lucky in your flight not to meet all the stuff you set up, and all will run fine. Or you just as well may run into it and get a slideshow.

Edited October 24, 2021 by 216th_Nocke
edit: slow motion, not a slideshow

[kingstreet73]

Hmm that's actually one thing I didn't consider, thanks for this point. Since there's no way for me to prove this situation in particular, I guess no reason for me to be upset. I'll  just stick with playing low settings, and if I find another case I'll try and setup a proper benchmark with the same mission.

[PatrickAWlson]

PWCG on low should produce plenty of contact.  I generally play on medium. Medium does strain my system if everything happens to get spawned in the same time period but more often than not it's OK.  I find high is not only more than my machine can handle but also more than I can handle.

 

Some things to consider:

Use Medium Air with Low CPU Allowance.  That will put flights in the air than low while limiting bomber flights, which suck up a lot of CPU. 

If you are flying a fighter campaign you might want to set ground activity to low to save CPU for more planes.

If you are flying a ground attack campaign, do the opposite - set air to low.

 

In the end it is planes that take up the majority of CPU cycles with bombers being the worst offenders.  Some savings can be had by reducing ground units, but a truck is not nearly as AI intensive as a plane.