Very Weird Client Behavior

[Coug4r]

Yeah sure. CTLD uses MIST but MIST does a lot of background stuff, if you get fps drop just by running MIST it's more likely to be a MIST issue or DCS+MIST than CTLD.

[VAF [136] Striker]

OK, here's my update to the test this morning.

The problems with stuttering and planes flying in weird positions appears to be MIST related because the mission I originally posted about has MIST only.

When I run the mission with no clients the FPS swings wildly from 10 FPS up to 45 FPS.

You can also see the counter actually stop and start so it looks like the mission is running very erratically.

When I run a mission with no scripts in either 1.5.7 or 2.1 I get very clean FPS, very steady, around 60 FPS (default server setting).

When I run a mission with MIST and CTLD the frame rate drops to a steady 5 or 6 FPS and almost never moves from that.

So it looks like in combination, MIST and CTLD make the mission unplayable for now.

Just using MIST slows the mission as well, not as bad but still slows the missions enough to cause the weird client behavior or at least that's what it looks like to me.

Edited August 30, 2017 by 75th-VFS-Striker

[Coug4r]

Ok seems more like a MIST issue to start with, i don't know who did the latest work on that. Maybe you can draw Grimes/Speed/mrSkortch attention to this thread.

[Pikey]

if it's a scripting issue then it's CPU. You say it's not, but this is exactly the behaviour you get under load. The CPU doesn't neccessarily flatline at 100% when you have a problem processing, it can be 50% or 100% divided by the amount of cores you have.

You cannot see the "per core processing" since Windows10. On two cores if you have one CPU maxed it shows 50%.

No one is saying your CPU is bad, it's just getting used by something you aren't expecting. Scripting is a classic overhead that can be hard to find an issue with.

[VAF [136] Striker]

if it's a scripting issue then it's CPU. You say it's not, but this is exactly the behaviour you get under load. The CPU doesn't neccessarily flatline at 100% when you have a problem processing, it can be 50% or 100% divided by the amount of cores you have.

You cannot see the "per core processing" since Windows10. On two cores if you have one CPU maxed it shows 50%.

No one is saying your CPU is bad, it's just getting used by something you aren't expecting. Scripting is a classic overhead that can be hard to find an issue with.

I'm not a scripting person but I don't agree with you saying this is a CPU processing problem. This problem did not exist with 1.5.6 or the most recent update to 2.1. We were running these missions perfectly fine until then. Also, I'm not on Windows 10, I'm on Windows7 Pro 64 bit so I can see all the cores and what they're doing. I'm not an expert by any means but I can see that the maximum combined processor usage is 9% and the individual cores only spike (not all at the same time) at 80% occasionally. They then drop down to a flat amount around 10%. This is telling me that the processor is functioning properly. This is definitely scripting related since it doesn't happen when I remove MIST and CTLD.

[Coug4r]

Sure you can see per core usage under Windows 10, apart from existing tools that work i wrote something for it myself.

I'm also running CTLD with a stripped version of MIST and i have no performance issues, although that's on 1.5.

[Pikey]

Does it help you more if I say I had the exact same experience with MIST recently?

 

My processor was sitting at 40-50% and the network was lagging and the planes flying sideways and pointing up and updating every second.

 

The CPU wasn't flatlined or even half at capacity, but the game itself was running like a dog.

 

Removed scripts, it was like night and day.

 

Processing is occuring inside the frames with the scripts, it's not getting enough time to deal with the network traffic. CPU cannot process whatever is looping or getting stuck, thus wastes time and smashes the frame per frame delivery of the sim.

 

We can call it Jim if you like :)

I'm not a scripting person but I don't agree with you saying this is a CPU processing problem. This problem did not exist with 1.5.6 or the most recent update to 2.1. We were running these missions perfectly fine until then. Also, I'm not on Windows 10, I'm on Windows7 Pro 64 bit so I can see all the cores and what they're doing. I'm not an expert by any means but I can see that the maximum combined processor usage is 9% and the individual cores only spike (not all at the same time) at 80% occasionally. They then drop down to a flat amount around 10%. This is telling me that the processor is functioning properly. This is definitely scripting related since it doesn't happen when I remove MIST and CTLD.

[VAF [136] Striker]

Does it help you more if I say I had the exact same experience with MIST recently?

I get what you're saying now. You don't mean it's my processor specifically but more a processor issue related to using some scripts.

 

Has there been any investigation by the MIST and CTLD or ED people into what's causing this yet?

[VAF [136] Striker]

Sure you can see per core usage under Windows 10, apart from existing tools that work i wrote something for it myself.

I'm also running CTLD with a stripped version of MIST and i have no performance issues, although that's on 1.5.

I've found that CTLD does function on 1.5.7 still although it seems to have performance related issues. It's completely non functional on the recent version of 2.1 or at least on my setup.

[Pikey]

I get what you're saying now. You don't mean it's my processor specifically but more a processor issue related to using some scripts.

 

Has there been any investigation by the MIST and CTLD or ED people into what's causing this yet?

 

Yeah sorry, I meant, what you are seeing is a symptom exactly the same as CPU bottlenecking, rather than what I actually said which sounded like it's a problem with your CPU or there isn't enough or similar. IE comparing a symptom, not a cause.

 

I've been testing this myself and i'm not even sure it is scripting now, I have to be honest. I see no comparable differences with or without, but I've experienced decent performance and really knackered performance on a server with 3 connected clients.

 

I'm gonna shut my mouth a bit harder until I've properly been able to reproduce and isolate because right now none of my previous hypothesis withstood testing. I'm more inclined to believe it's generic DCS right now, but my personal jury is out.

 

I did the server testing with an identical mission that was terrible with one conencted client and the server FPS as reported by the ingame counter, with rendering off, didn;t budge from 30, it was incredibly consistent, unlike the client that had from 45-120 depending on what you were looking at.

 

I need more testing people, and unfortunately they don't want to test any more than I do.

[Johnny_Rico]

I help admin a pretty busy DCS server (20 - 30 people) on it most of the time

Before last weeks 1.5 update things were running pretty well on it (apart from no sling load due to crashing)

 

After last weds patch performance took a nose dive, with warping and planes flying sideways happening

[Pikey]

I help admin a pretty busy DCS server (20 - 30 people) on it most of the time

Before last weeks 1.5 update things were running pretty well on it (apart from no sling load due to crashing)

 

After last weds patch performance took a nose dive, with warping and planes flying sideways happening

 

Again, perfectly valid information that doesn't really conicide with a common story, as folks reported the 1.57 patch 3 weeks back doing something similar and 2.1's turn was last week (according to some).

 

Finding it hard to draw conclusions. :/

[Johnny_Rico]

I have a simple conclusion

1.5.6 was a good version of DCS

1.5.7 is not a good version of DCS