The parking spots on the world map are just placeholders it does not put a plane on a geographical location. It’s just a visual representation of the dropdown. But yeah, it’s way off at some places. I fly only GA and sometimes have to click spots that would end up on a highway or outside the airport in the map check EHRD GA parkings for example, but i’m still correctly placed in the game world.
As far as i can see these tubeliners spots on airports should be corrected though, have had a couple of these misplacements some months ago when flying tubes.
It could have been a slight camera movement induced effect, but it wouldn’t show in this case like “sparkling” but as “smoothly” changing. And in any case, it never displayed these artifacts prior SU5. The videos are too short to really see this, but over a longer time period you really notice the cycling and repeating nature of this sparkling visual bug. As for the NVidia settings they also were never necessary prior, nor are they needed with any other game.
It is possible this is the case because in Lionel’s video he was saying how they changed the accumulation buffer time-spread or something related to this. But in this case, and given the other viewdir Maths bug in the Shaders (or at least what looks like it from the outside) I’d tend to think it is a rogue vector causing this, like when you compute a cross product but with the wrong source vector, or if you wrongly swizzle .xyz, and this makes the resulting vector something rotating or pointing in the wrong direction frame after frame.
Location: red dot
Directions: the two yellow arrows
Time of Day: early morning (sun on the back from the East and low on the horizon to exhibit the problem better)
Will check later today, currenly being vectored into i think a nice holiday spot in a mountain. I love how ATC cares about me not having to take a bus or cab from the airport.
CPU RAM USAGE DOWN FROM 16GB TO 4.7GB
Microsoft’s Flight Simulator used to be one of the most demanding PC games ever made, but a new update introduces significant engine optimizations that drastically reduce hardware usage.
“We have re-written a lot of the parts of the engine, the architecture, in order to get the maximum performance of the sim and minimize resource memory bandwidth footprint,” Asobo Studio CEO and co-founder Sebastian Wloch said.
Before the update, the game about 35-40FPS with 100% CPU and 16GB of RAM memory used in a congested NYC Manhattan scene on a PC with an Intel Core i7-9700K, NVIDIA GeForce RTX 2060 Super with Ultra graphics with upscaled 4K visuals.
After Update 5, FPS was practically doubled, RAM usage was down to just 4.7GB (a 70% reduction).
It should be also said if the video over Manhattan was showing any 70% reduction of RAM, it was also showing a 70% reduction of render resolution (were they running with 40% render scaling or something like than to prove 60fps over Manhattan with a 2060?)
I agree with you. We really do not know what they did and it seems like only little information is coming from OSOBO.
I have done some investigation as to shimmering.
If I was over downtown Shanghai, China I got a lot of the shimmer.
Switched to Hong Kong, China without changing anything the shimmer was gone.
What was the difference between the two cities the number of houses and structures.
It is my believe that in Shanghai, China it could not load them do to memory limitations.
@NineMile1 I’m sure they’ll find out and correct this bug, they shall have enough material now!
@scriptkid I believe they were, I was just pointing out the Q&A video was showing 60fps but maybe some didn’t notice they were also using a render resolution of 40% (or so) in the video (in both pre and post SU5 examples) in order to make the 2060 running this fast, and this helps. This doesn’t relate directly to the 70% reduction of RAM usage though, my post was more about highlighting the similar 70% figure (RAM and render res.). In other words, what they are showing and telling is great, but like any number, you also have to see what they are relating too. In the example of the 40% render res., you can search online and count the number of press outlets relaying and preaching how much better SU5 because FS2020 can run at 60fps on a 2060…
Yep, now also seeing it, it’s minimal, but, it’s there:
I marked where it was most noticable, other parts where really single pixel and barely noticable. But it definitely is there, these pictures taken with a drone cam which appears to be 100% stationary.
@scriptkid
Yes I have the the “Love VFR” mod installed. Otherwise you should try at an earlier time of the day with the sun lower. This will amplify the bug.
It is because of the risk of ruining the details with compression that I’ve shot the monitor close up (with iPhone) and cropped the 4K source image so that I can fit at least about 1.5 sec of material in 4MB in H264. I’ve made multiple conversions adjusting the compression rate (using HandBrake) until I can maximize filling 4MB. In my videos, since each physical pixel from the monitor is displaying as multiple pixels in the video, this helps eliminating video compression artifacts from the equation.
NB: my videos are “pulsing/shaking” but it is the “auto focus” working hard, and my heart beat holding the phone! If you look closely at the pixels near the mast against the background, you can see there isn’t much change in parallax and the camera is not moving (I selected clear weather no wind, outside view - not drone -, and these are from the VR headset acting as the camera and held in place).
Fully understand, it’s also clear you have individual pixels showing in the clip.
To your earlier response on the Asobo clip: Cheers
When i change the sun it’s a shimmer fest, when settled (around 11:15 utc september 5th 2021), it reduces to almost not there. but it’s more then when it’s lighter, slightly.
I think, in my case, i should consider it neglegible as i really need to try hard where objects are close to 1,2,3px? on screen. But in case of VR… i can imagine the frustration when facing this.
I also know these settings are not optimal for an RTX2060. But it suits me fine as it is ok for my play style (bit having a bit more powerfull card would be very welcome, i refuse to pay these scalper/inflated prices. i understand chip shortages and increased cost, but not by these margins).
Why Asobo make it tests with 40% of render scale?
I have everything in ultra and i have an average of 50 fps with a render sclae of 100
Of Course if I use 40 of render scale i can get 80 -90 frames but with not the same quality
Asobo should focus on getting good performance with a minimun of render scale of 80-90.
And another suggestion is to use more CPU. Its true that now ot uses more , but still can use more and more
What memory limitations? Are you suggesting that shimmering may be related to lower levels of physical system memory? Because if so, and if people complaining about shimmering have only 16GB of RAM (or even 32GB), that may actually help explain why my system with it’s 64GB of available physical RAM does not exhibit any shimmering.
If that is not where you’re going, then I’m afraid I’ve lost you. Or maybe you’ve lost me.
Asobo Studio CEO and co-founder Sebastian Wloch said. After Update 5, FPS was practically doubled, RAM usage was down to just 4.7GB (a 70% reduction).
So if you believe this statement how can you use all your ram on your system?
Unrelated to the shimmering (but I was reminded when looking at those videos!), to me it seems like the buildings in general reflect a bit too much light compared to the surrounding terrain, which in real life has many surrounding unmodelled 3 dimensional objects which would also be reflecting. I think it makes the buildings stand out as squares against the terrain and I guess would be an easy thing to tweak. However I am no photographer so interested to hear any other opinions ?
It is naturally most apparent when the sun is low (and in VR when everything larger and brighter!)
I do not think he in any way meant the RAM usage would ever exceed 4.7GB of RAM, because that statement makes no sense whatsoever. RAM usage can and will fluctuate based on a whole host of different factors, including (but not necessarily limited to) what aircraft you are in, where on the planet you are, whether or not you have dense or sparse scenery around you, whether or not your plane is moving, what view you may be using and a whole lot more.
I attempted an experiment a few days ago where I simply watched how much RAM the system used from right when MSFS was started, all the way to spawning on the ramp at Charles de Gaulle in an A32NX. I did not take note of which livery, but I suppose that, too, could use more or less RAM depending. It used 17GB.
Windows can will and does stuff that 17GB into some smaller amount for people who simply do not have 17GB to give to the sim, which obviously includes all people with only 16GB total, and could even include some with 32GB. However, I will note that during my test at no point did the total go above (or even close to) 32GB total in use, so I suspect 32GB is the “sweet spot”. I think windows using it’s magic to stuff that 17GB into something smaller simply MUST come with a downside somewhere, and it may be manifesting itself in the form of some of the graphical anomalies others report that me and my too much RAM do not have.
While I am not Seb, and I am incapable of reading his mind, there is simply no way he meant that under no circumstances would the sim ever use more than 4.7GB of RAM. During the scene they showed, perhaps, but it’s too big of a variable to put a hard limit like that on.
(were they running with 40% render scaling or something like than to prove 60fps over Manhattan with a 2060?)
