To MT or not MT

I’ve created the shortcuts for VR on and off to the MT executable, but haven’t been able to leave work just yet to try it. Impressions?


On my box, it’s led to a massive bout of tweaking. It works, in that there were no showstopping bugs or crashes on my first hour or so. Then I went looking for that perfect setting. On my rig (9600, 3080ti, 32gb) I’m GPU locked acording to the openXR tweaker menu.

However, even if my framerate is hardly improved at all, it is far more consistent. Rock solid,. And that makes it feel fluid as… beer. Yeah beer :angel:


:beer: :beers: :beer:


Very Similar experience to yours. very brief tests of the following settings:

DCS shortcuts

No crashes as yet, but I have done zero tweaking so far.

Using my current VR settings as a baseline I’m not seeing a dramatic improvement in FPS, but I am GPU bound. I am yet to try a busy mission that would really tank my frame rate and see what FPS I can maintain.

Some preliminary data for y’all. Multi-thread seems to at the moment only mean two**?

Standard DCS:
Task manager details lists DCS.exe at 76 threads, ~17% CPU usage. You can see the primary core DCS is using in the core graph. We’re hitting only about 80FPS up high altitude in the open in the busy Offensive Posture mission on Hollo Pointe North.

Multi-Thread DCS:
Task manager details lists DCS.exe at 77 threads**, ~20% CPU usage. You can see the two cores DCS is using heavily in the core graph. We’re hitting only about 100FPS up high altitude in the open in the busy Offensive Posture mission on Hollo Pointe North.


  • In MT-DCS, I see the thread count vary between 76 and 78.
  • I normally run capped 60fps in autoexec.cfg & Vsync because I only have a 60Hz monitor and don’t want wild framerate swings which induce stutter and time-warp effects. For this test I have removed both of those restrictions.
  • Lower down and closer to the action, framerates varied from 40-60s generally in both cases.
  • Frametime chart in MT mode is much smoother.

While my first screenshot showed CPU limited by the main thread, it changed out of nowhere to GPU bound. The usage shown for my 3070TI is not near the limit, although DCS is using all the VRAM.


I’m still on the road from a week of work. So I’ll update tomorrow.
TBH, I’m a little disappointed to hear how much tweaking you still have to do, in order to get solid results. I truly thought this was going to be more use friendly and plug and play. Hopefully, I’m wrong.


Completely different on my end. I get 8 threads (out of 24) that get hit pretty hard, in singleplayer, DCS usage is mostly at around 20% with spikes to 40% over large cities like Damaskus. The only busy server I flew today was the 4YA Project Overlord WWII server with about 50 people online and when I was flying at high altitude in contrails I was sitting comfortably at 80-90fps in a large furball and I get around 70fps at low altitude (I was around 45fps on the server before this patch). I looked at usage every now and then and DCS was always around 25-40% CPU usage.

All the visibility sliders have zero impact on performance for me. They’re all maxed out now xD

Edit: VR of course. Reverb G2, PD at 1.0 using DCS’s OpenXR runtime


This is a Good Day for DCS.

Keep them reports coming.

It will be a bit before I can it give it a go. I said to Wags (YT reply) that, in VR anyway, I expect 25%-ish FPS improvement (non-trivial scenarios) – but much smoother. A Good Thing. Lot of variables there. Looking forward to throwing more units into the mix.

My 5900X will have something to do now [with the 12? cores lying around]

Interesting comments about being Gpu bound now…if I understand this correctly then this is good news for nVidia/AMD; being Cpu bound I felt justified not upgrading my video card. Where’s my wallet?

I can only begin to imagine how complex it was/is to rebuild this beast for MT.



My results were solid out of the box, using my previous VR settings. The (little opportunity to) tweaking has all been turning eye-candy on/up in order to see what I could do. I didn’t hit the limit with clouds ultra, shadows on, vis range turned up and heat blur on. Even bumped up my OpenXR Toolkit FSR Res % without losing useful FPS (repro on). No noticable repro artifacts, and smooth framerate. Like it’s not even DCS anymore, but some other game; if that makes sense?

I tried repro off and turbo mode on briefly, but decided with the Reverb G2 I still need repro in the cockpit or else the text gets a bit blurred when moving my head. Will definitely experiment more when I return home.


I identify with this statement. Feels like you could justify a 4090 now, right? :rofl: :joy: :man_facepalming:

Not sure if that has to do with other tasks or what CPU you have to compare to.
I ran the same mission in singleplayer and I seem to only have one partially busy core more than the multiplayer result. Call it 2.5 cores.

It appears the other threads ramp up when you have a more complex scene on hand. Switching to a more complex mission I found the same results as above until I went down on the ground - then the cores got busy:


Busy busy cores doing busy busy things! :joy:

This is an awesome time to DCS!


For some reason I am having real issues with the menus seizing up when using the MT version with OpenXR… anyone else having this problem? If I can get to the cockpit everything works great while I am in the 3D environment.

I got a 30% boost in VR in a hot Tomcat on a crowded Supercarrier deck at night, but that’s still 25fps :frowning:

In 2D I got a 59% boost which is juicy for filming.


In MT I am seeing a clipped view, like looking at a screen covering about 60% of the field of view and in both I’m needing to keep my head still or I get some tearing. I’ll experiment more today. I’m pleased a lot of people are seeing real improvements but not yet for me.
EDIT: these issues now fixed.

1 Like

I noticed i am CPU bound so its probably time to ramp up to 5Ghz overclock again
Not very environmental but needs must. It will save on lighting the fire too

1 Like

I usually have my 9900K at standard turbo, so 5Ghz single core and 4.7 all cores. Decided to try an overclock** and let my ASUS board do it’s auto magic. It gave me 5Ghz up to 4 cores and 4.9Ghz 5-all cores. So I told it to round up to 5Ghz all cores.

While my Cinebench R23 score went from 12,300pts to 12,900pts (5% more) - DCS didn’t really change at all for me.

Maybe that means it is time to just buy a 13900K for a 59% single core uplift and a 324% multi-core uplift. :grin:

**I know manual overlocks can yield more potentially but I do not have the patience for that. Stability please!


Ok, this is wild. Since I’m CPU-bound, I tried reducing or turning off CPU-heavy settings and boosting GPU-heavy settings. I still dip into the 20s, but on average I’m getting 30-36 (half refresh rate) and sometimes up to 40. With MSAAx2, shadows high, default terrain shadows, and high textures. It looks mouth-watering.

I also noticed clouds are a huge CPU hog. If I go ultra, my framerates will go from 36 to 18 when entering or skimming a cloud. Standard reduces the impact from 30 to 26-ish. Might give low settings a go.


Yes, clouds are an FPS killer! I’m using both clouds low and terrain textures low, I think for at least the terrain textures I don’t have enough GPU RAM (8GB). This is in VR of course.


More than before MT? I can’t test it for a bit.