Stutters on 9800X3D 96 RAM in MP

[MoleUK]

54 minutes ago, SpecteRED said:

I also noticed that the stutter appears as someone disconnects from the server or joins the server.

That's an old bug/performance issue, you can't fix that client side as it's not your hardware that is the problem. ED need to fix it.

It can also occur when players die and are automatically slotted into new aircraft, or when they manually change slots.

Doesn't happen every time, seems to start happening when a server starts getting overwhelmed. Could also be down to server specific slot blocker stuff as well, but i'm skeptical about that.

I took this footage over a year ago (timestamped to show the join/dc/reslot events lining up with stutters), and it was far from the first time that the join/dc/reslot stutter issue cropped up for me. 

Servers that only use dynamic slots seem to suffer from this issue less often, presumably because the game/server is not refreshing hundreds or thousands of slots every time someone joins/dc's/reslots.

Theoretically the game should be more than capable of refreshing a list of slots even if it's thousands of them without a performance hiccup. But whatever way ED implemented it (before dynamic slots were introduced) wasn't designed to handle that many slots in MP, at least at a guess.

Edited Sunday at 10:04 AM by MoleUK

[MAXsenna]

25 minutes ago, MoleUK said:

Theoretically the game should be more than capable of refreshing a list of slots even if it's thousands of them without a performance hiccup. But whatever way ED implemented it (before dynamic slots were introduced) wasn't designed to handle that many slots in MP, at least at a guess.

Guess we won't see improvements until the server is "completely" multithreaded for these types of scenarios. Like a split between the mission running, the AIs, bombs, missiles and various aspects of the game. And then "the multiplayer technology" that binds everything together. 

[MoleUK]

6 hours ago, MAXsenna said:

Guess we won't see improvements until the server is "completely" multithreaded for these types of scenarios. Like a split between the mission running, the AIs, bombs, missiles and various aspects of the game. And then "the multiplayer technology" that binds everything together. 

Possibly.

But even 1 CPU core shouldn't be struggling with stuff like this, I suspect it's more a poor implementation than it is a single threaded problem.

You can also see the game struggle to update the list of MP servers, and when initially choosing a role on joining a server with lots of slots. And even when you aren't getting stutters when players join/disconnect during general gameplay, if you keep the score tab/player list or role select open you will start getting them when players join/dc. You can see it struggle to update/refresh the player list.

Something causes the CPU to come to a standstill or hiccup if only momentarily. They could all be unrelated to eachother, but they all feel symptomatic of a single problem imo.

Edited Sunday at 05:17 PM by MoleUK

[MAXsenna]

42 minutes ago, MoleUK said:

But even 1 CPU core shouldn't be struggling with stuff like this, I suspect it's more a poor implementation than it is a single threaded problem.

Sure, but when it struggles to handle 1K units in single player, I assume the already poor implementation doesn't exactly thrive in it. That's my point. I believe optimisations are needed all over. 

[Mr_sukebe]

Please take a scan here:

https://forum.dcs.world/topic/371717-vr-performance-issues-start-here/

[SpecteRED]

I think my only option is to switch to the 14700k. It was my first AMD and it looks like it will be my last.

[Raven (Elysian Angel)]

Just now, SpecteRED said:

I think my only option is to switch to the 14700k. It was my first AMD and it looks like it will be my last.

That's entirely up to you, but that would be a serious downgrade (X3D is hugely beneficial for flight sims). My guess is that there's something going on that doesn't have anything to do with hardware. If there were issues with the 9800X3D itself, I'm sure you'd see a lot more people complaining about it here on the forum  

[rob10]

5 minutes ago, Raven (Elysian Angel) said:

That's entirely up to you, but that would be a serious downgrade (X3D is hugely beneficial for flight sims). My guess is that there's something going on that doesn't have anything to do with hardware. If there were issues with the 9800X3D itself, I'm sure you'd see a lot more people complaining about it here on the forum  

THIS!  The 9800X3D is probably the most recommended CPU for DCS currently so clearly it's not having widespread issues.

[Raven (Elysian Angel)]

Stuttering is one of the most common problems experienced in DCS, and one of the most difficult to properly diagnose. Unless you get lucky, it's a problem that simply throwing money at won't fix.
Did you do a search here on the forum and read all other threads about people complaining about stuttering, and how they managed to fix it on their end? It doesn't necessarily mean that their solution will work for you too, but it's worth a try.

[SpecteRED]

It's just that I've already tried everything. as soon as I bought this CPU, my statters started. Before that, I had 11700 and there were no statters. I thought that the statters were caused by the fact that I did not reinstall Windows after changing the CPU. After I reinstalled Windows (I tried 10 and 11) - nothing has changed.

All the advice from this forum, redit and GPT chat did not help me. Tech support has acknowledged that there are unexplained lags on some AMD systems. So I just don't know what to do except take the proven option.

again, I repeat that my processor passes all the error tests.

I really want to keep this processor and not spend more money.

[Raven (Elysian Angel)]

As far as I know, the only real problems (on a hardware level) that you could have with a 9800X3D is when you have an Asrock motherboard. Which brand is yours?
Indeed I saw in your original post that you already tried a lot. Are you using the latest BIOS version (there are regular stability updates with newer AGESA versions), and the latest chipset drivers? Are you sure all BIOS settings are correct, and not causing desynchronisation issues?
Did you try disabling the Power service in Windows? I myself don't need to do that, but a lot of people are recommending it as it can potentially fix stuttering.

[SpecteRED]

I have a Gigabyte X870 Gaming X WIFI 7, I tried to roll back to the bios version F6, now it's F7 and I also tried the beta F8d. I tried default settings, tried overclocking. I tried to turn off the power service, but it didn't help

[SpecteRED]

38 минут назад, rob10 сказал:

THIS!  The 9800X3D is probably the most recommended CPU for DCS currently so clearly it's not having widespread issues.

I look at many benchmarks and many people don't show the low 1%/0.1%. It's important to me. I also saw a lot of articles on reddit where everyone praises AMD for their high FPS, but scolds them for statters in various games. One guy changed the motherboard from msi to asus and the sutters didn't go away, and when he changed 9800x3d to 7800x3d, the statters stopped (but that's all in other games)

[MAXsenna]

@SpecteRED I read through the thread. May I ask you why you upgraded?

DCS can be such a diva so it might be some obscure combination of peripherals and what not in your setup. 

[SpecteRED]

5 минут назад, MAXsenna сказал:

@SpecteRED I read through the thread. May I ask you why you upgraded?

DCS can be such a diva so it might be some obscure combination of peripherals and what not in your setup. 

I had a pretty low fps (40-50) on the loaded servers on the Germany map on the F-4. Everything was perfect on all other maps (100+ fps)...

I bought it in June and since then I have not been playing, but just testing bios versions, different overclocking, and different versions of chipset drivers. And I want to finally play

Edited Sunday at 08:59 PM by SpecteRED

[Raven (Elysian Angel)]

9 minutes ago, SpecteRED said:

It's important to me

Funnily enough that’s exactly why X3D is that good: when I upgraded from a 5900X to 9800X3D, my minimum framerate in various games went up by as much as 50-60%.

I also play ARPGs sometimes, and that’s another genre that X3D is particularly suited for: the increase in performance was eyewatering (and I kept my graphics card so in all cases the increase in framerate is 100% due to the CPU).

I personally don’t have stutters in any game. DCS can sometimes do weird stuff but that’s when ED messes something up. But they also fix it

[SpecteRED]

@Raven (Elysian Angel)I'm just wondering what temperatures you have in the game and in stress tests? and what is curve value?

[Raven (Elysian Angel)]

8 minutes ago, SpecteRED said:

@Raven (Elysian Angel)I'm just wondering what temperatures you have in the game and in stress tests? and what is curve value?

I haven’t checked in a while. While gaming CPU temperature is around 45-50 degrees, and GPU 70-75. I haven’t done any stress tests since I built the system, and I don’t remember temperatures. I use a 420mm AIO and my case is packed with 140mm fans so temperatures stay low even at low fan speeds, so I can keep the noise down.

What do you mean by curve value? Fan curve? I keep my fans between 300 and 600rpm. More is not needed.

Edited Sunday at 09:21 PM by Raven (Elysian Angel)
Clarification

[Raven (Elysian Angel)]

I take it you also added exclusions for all DCS folders (both main folder and saved games) in your antivirus software, and are making sure no other processes interfere while gaming/simming?

Have you debloated Windows and made sure the amount of processes running stays at a minimum required level, without stuff running in the background that’s not needed?

[MoleUK]

1 hour ago, SpecteRED said:

I look at many benchmarks and many people don't show the low 1%/0.1%. It's important to me. I also saw a lot of articles on reddit where everyone praises AMD for their high FPS, but scolds them for statters in various games. One guy changed the motherboard from msi to asus and the sutters didn't go away, and when he changed 9800x3d to 7800x3d, the statters stopped (but that's all in other games)

The X3D's also perform much better at 1% and 0.1% lows.

Unfortunately stutters in DCS in VR are notoriously tricky to track down. It can be memory, vram, third party monitoring overlays/apps (CPU temp monitoring, RTSS, MSI afterburner etc), antivirus software, the CPU itself, BIOS settings (or bad BIOS versions), the GPU or specific GPU driver versions, issues in the VR software itself, wireless problems if playing in wireless VR, usb problems if over USB. And then there are the DCS bugs and server side performance issues. That's without even getting started on in-game settings.

Just got to go through it all to track them down. It does get frustrating.

The stutters when players join/disconnect you can at least rule out as something you can't control. It happens to every user connected to the server when it's happening, not just you.

Edited Sunday at 10:22 PM by MoleUK

[MoleUK]

Oh, and there is one other MP specific stutter bug to be aware of. It was partially fixed earlier this year/last year (it was VERY bad for a while), but it's not gone entirely by any means. Sometimes ground units taking damage from certain munitions (or certain units being damaged at all) would also cause server wide stutters. Mavericks in particular were bad for triggering them.

These stutters when ground and naval units were taking fire would also show up in track replays in a repeatable fashion, so it's possible they affected SP just as badly. But generally the more complex the MP mission and the more ground units there are the more the odd performance issues start to stack up.

Keep an eye on the chat feed for AI units being killed that coincide with the stutters, as well as players joining/disconnecting/dying. You might see some of them match up.

Edited Sunday at 10:21 PM by MoleUK

[SpecteRED]

39 минут назад, Raven (Elysian Angel) сказал:

I take it you also added exclusions for all DCS folders (both main folder and saved games) in your antivirus software, and are making sure no other processes interfere while gaming/simming?

Have you debloated Windows and made sure the amount of processes running stays at a minimum required level, without stuff running in the background that’s not needed?

Yes, I did it. I'm pretty sure it's not about the Windows settings, because with the same Windows settings, I used 2 different processors and there were statters on one, but not on the other. 

I didn't have any statters. Then I changed the processor and the motherboard and the statters appeared. everything else remained unchanged (antivirus, USB devices, Windows settings, video card)

1 час назад, Raven (Elysian Angel) сказал:

I haven’t checked in a while. While gaming CPU temperature is around 45-50 degrees, and GPU 70-75. I haven’t done any stress tests since I built the system, and I don’t remember temperatures. I use a 420mm AIO and my case is packed with 140mm fans so temperatures stay low even at low fan speeds, so I can keep the noise down.

What do you mean by curve value? Fan curve? I keep my fans between 300 and 600rpm. More is not needed.

I'm talking about the voltage curve. I have -25 and a stress test temperature of 72 degrees. game 62 degrees

[MAXsenna]

According to he's OP, he's not using VR but TrackIR. Just mentioning it.

Sent from my SM-A536B using Tapatalk

[SpecteRED]

34 минуты назад, MoleUK сказал:

The X3D's also perform much better at 1% and 0.1% lows.

Unfortunately stutters in DCS in VR are notoriously tricky to track down. It can be memory, vram, third party monitoring overlays/apps (CPU temp monitoring, RTSS, MSI afterburner etc), antivirus software, the CPU itself, BIOS settings (or bad BIOS versions), the GPU or specific GPU driver versions, issues in the VR software itself, wireless problems if playing in wireless VR, usb problems if over USB. And then there are the DCS bugs and server side performance issues. That's without even getting started on in-game settings.

Just got to go through it all to track them down. It does get frustrating.

The stutters when players join/disconnect you can at least rule out as something you can't control. It happens to every user connected to the server when it's happening, not just you.

now I really only have statturs when another player join/disconnect, but they didn't happen before. also, there are no 13700 of them either. So this is a solvable problem.

moreover, I see such a problem in star citizen. And I saw a similar situation on Reddit in Final Fantasy (online).

[MoleUK]

11 minutes ago, SpecteRED said:

now I really only have statturs when another player join/disconnect, but they didn't happen before. also, there are no 13700 of them either. So this is a solvable problem.

moreover, I see such a problem in star citizen. And I saw a similar situation on Reddit in Final Fantasy (online).

The stutters when players join/disconnect is very intermittent. It will only happen on some servers, and only some of the time. You mentioned using your wifes machine earlier, you might want to try having both machines logged into the same server at the same time if possible. You will likely see them both stutter at the exact same moments.

A server like growling sidewinder for example is far less likely to have any of these join/disconnect stutters, but a more busy server like 4YA will have them more often. Particularly if 4YA are still using non dynamic slots, I haven't checked in a while if they're still on the old slots or not.

Edited Sunday at 10:48 PM by MoleUK