CPU Bug "Limited by Main Thread" is causing the bad FPS

Ok, what does your FPS counter look like? I‘ve seen a few posts up that someone seems to have a very long time on the manipulators.

1 Like

I will try this.
But the thing that MOST cut frames by far is big airports.(my case)
I go from 30 fps down to 15 fps or less, big cities with handcrafted airports , with the same aircraft.

How the hell do you get 45fps inside the a320?
We need to find out what is working in your computer that is not working in the rest of our pc’s, for some reason the game seems to be perfectly optimized in your pc, as it should.

1 Like

I feel sorry for you in that case because movies must be a nightmare for you at 24FPS. I do think its impossible for anybody to process movement at 500FPS though, that’s 2mS and no study has shown that anybody can process an image in less than 13mS.

1 Like

Oh for god’s sake, are we having this debate?

5 Likes

here are my tests

You are getting the same as me

1 Like

Depends on the movie. If there is a lot of horizontal panning with good lighting/contrast, yes, they are a nightmare. It just looks like the picture is skipping between the horizontal positions. If it’s a darker movie with no panning, it’s ok.

About the 13ms thing, you are right. But the eye has persistence. So you can notice flashes of light much shorter than that, if they are very intense, and you can even identify the image that was flashed. Yes, normally you can’t tell the difference between one flash, and two flashes 13ms apart, at the same spot. But if those are moving images, the larger the distance the image is moved between two frames, the more likely you are to notice that the image jumped/skipped.

Motion blur is very important in this perceptive scenario as well. Our eyes and brain are used to 24 FPS with rolling shutter in the movies (with usually a lot of “organic” motion blur) since we are born. That, plus a “passive” approach to movies helps in making most of the scenes acceptable as “fluid” (apart from the panning images with little motion blur, a cityscape pan in full daylight is usually like that).

In some console games locking to 30fps and adding strong motion blur is used as a technique for those games that can’t reach 60fps.
I always thought a realistic motion blur algorithm could help much more in gaming than RTX and raytracing. In my opinion fluidity (or at least the feeling of it) is much more important to our brain than “quality” to make us believe we are looking at the “real thing”.
But high FPS with G-Sync means a lot less latency, and this is an essential component of the gaming experience as well, in my opinion.
It’s a complex topic that could be digged for hours…

1 Like

Yes, but motion blur is just a trick to try to make you not notice the problem. Personally I’m not a big fan, looks ugly. The only way to fix the problem is not to make those panning shots too fast, or to increase the frame rate, which will allow you to pan faster, without creating the problem:

https://www.red.com/red-101/camera-panning-speed

So why did alpha/beta testers report good fps with prerelease and worse with this release version? It has to be some bug that was not existing in previous versions.

5 Likes

Unfortunately we cannot talk about it, we still are under NDA for some reason

2 Likes

The amount of trouble DX12 is giving developers makes DX11 the more ‘stable’ choice. Asobo made right choice here.

Just curious, at what airport did you run this test at? (it makes a difference) and at what settings? was this on ultra or high-end? thanks Guther

Just an FYI the gauges are programmed in C++ I’ve worked with that framework before it just allows you to render UI in HTML and Javascript l.

From what I can tell the visual aspect of the gauges and FMS is coded in Javascript and HTML the actual data it’s feeding is from C++ and the delay is caused by how bloody complex the Javascript is and how complex the C++ behind it is it’s taking a while to load. There’s defiantly a lot of room for tweaking here.

It’s an interesting decision to use Javascript and HTML for an in game avionic, this stuff is usually used for UI like leaderboards.

5 Likes

I have hopes because I loose 10 fps on the cold and dark XCUB just by popping out the G1000 window (alt-gr click).
A black empty window eats 10 fps so there is a lot of room for improvement…

I filed a zd ticket to push as performance is not listed yet in top issues if I’m not mistaken

3 Likes

Another interesting test is getting in the 787 Cockpit in the Profile > Hangar section of the game. Of course there is no world outside but with the plane completely off here I get constant 144fps (my monitor limit) and full GPU usage in Ultra.

I think this is another proof our GPUs are not suffering for the cockpit geometries or tessellation at all.

2 Likes

The issue is within the instruments as you well say, not the geometry or 3d model, that is why if you use planes with no virtual cockpits, but analogue, there is practically no difference in fps from inside to outside.

1 Like

Sure, but getting in the hangar for me was the only way to see high FPS in the airliners’ cockpit and exclude in a “scientific way” the possibility some geometry could make things worst (apart from the well known glass cockpit problem in all planes).

Doing a cold start with instruments off in any airport would not reduce the FPS problem, this is another thing to be considered in my opinion (probably the instruments code and related framework is running in background even if the instruments are completely off).

Exactly, that was my guess as well, probably everything is running in the background even if the plane is turned off when in-game, but not in the hangar module.
I guess they need tl find a way to optimize this, in the a320 I still get 30-35 fps (60-70 outside) but in the 747 and 787 it is abysmal