kevinfirth

Nah, only paid £1800 for mine last year, best upgrade I ever made in terms of performance (and value compared to now!). all luck though, can't claim any massive prescience 🤣

Update from a few days ago on Majestic forums is they have a modeler now, work has been done in the exterior, and will move on to the VC when that's complete

To be fair, any decent atc app is going to have to take a chunk of performance away from other stuff. Everything costs compute, whichever way you slice it.

I'll retract my earlier comment - I was watching on my phone, but just having rewatched in 4k, I can see many differences and improvements. Whether it's worth it as an update I'll leave to others to decide individually....

I'm just not convinced, looks to me like pretty much all of those comparison shots were taken under different lighting conditions. So how much of any visual differences are due to the better lighting capabilities of the sim engine itself as opposed to the modelling? I wonder...

Ok, but I wasn't going into the politics of it, no point, and I absolutely don't want to litigate the actual referendum again(!), just curious as to whether it was a semi self inflicted outcome...

Unintended but quite foreseeable. Given the age profile of people I saw at the show, it would be interesting to learn how many of them voted leave...

Thankyou. I've set up the oasis driver, couldnt get the controllers to connect via my onboard bluetooth, but fortunately there's just been a beta driver release last week that allows connection via the headset itself - so have that all working great. I havent played with the frequencies yet, but will do. I see what you mean about the sweet spot, just have to live with that I guess! Cheers!

Just making a foray into VR, wondering if anyone had any tips/tricks to share that will cut my learning curve please? 😁 Thanks in advance!

..apart from the fact you need 2x 5090's to run it

Asobo's advice to combat contribution culling seems to be to group objects, so they are evaluated as more significant by the sim engine and pass the render threshold. It will be interesting to hear some feedback from @Kaiii3 as to whether their internal testing has shown this can mitigate the problems?

For the benefit of others, the landing gear issue is apparently a function of 'contribution culling." Here's a summary produced through Claude AI. Contribution culling is one of the more fascinating rendering optimisations in MSFS 2024, and it operates at several different levels. Here's a breakdown of what it is and how it works in that engine. What "contribution" means In the sim's rendering pipeline, every object — a building, a tree, a terrain patch, a cloud — is assigned a contribution value: an estimate of how much it actually affects the final pixel output. This is a function of several things combined: the object's apparent screen size, its distance from the camera, lighting conditions, and whether it's occluded by other geometry. Objects with low contribution aren't worth paying the GPU cost to render. The culling hierarchy MSFS 2024 applies contribution culling in a tiered way: How the contribution score is built The score isn't just distance. MSFS 2024 factors in: Projected screen area — how many pixels the object would actually cover. A large hangar at 2 km scores much higher than a lamppost at 500 m. Luminance contribution — objects in shadow or behind haze have their score reduced. A dark tree in an overcast scene contributes less than a sunlit building. Object priority class — the sim assigns category weights. Aircraft, runways, and navaids have protected high-priority classes; ambient vegetation and distant traffic are low-priority and get culled first under frame budget pressure. Temporal history — if an object was culled last frame, it carries a small hysteresis penalty to avoid flickering (the "popping" artefact where objects flash in and out). The render budget dimension This is what makes MSFS 2024's approach different from older LOD-only systems. Rather than a fixed distance cutoff, the engine runs a budget-aware cull pass. Each frame has an estimated GPU time budget, and the system ranks surviving draw calls by their contribution score. If the scene is cheap (empty ocean, no weather), almost everything passes. If the scene is expensive (dense photogrammetry city, heavy cloud layer, rain effects), the threshold rises and lower-contribution objects get cut — even ones that would have passed under normal conditions. Shadow pass separation Shadow rendering runs its own contribution pass independently of the main view. An object can be rendered in the primary scene but culled from the shadow map (or vice versa, though that's rare). Ground vehicle shadows, for example, are often culled at distances where the vehicle itself is still visible, since the shadow's pixel contribution is negligible from altitude. World streaming interaction Contribution culling feeds back into the streaming priority system. Tiles whose objects would be immediately culled on load don't get promoted in the streaming queue — the system avoids spending bandwidth on assets that won't make it past the cull threshold anyway. This is why you'll notice different pop-in behaviour depending on viewing angle and altitude: the engine is continuously recalculating what's worth pulling down. Where you feel it as a sim pilot At cruise altitude, dense cities resolve more slowly not just because tiles are far away, but because individual building contribution scores drop as screen area shrinks. Fast panning camera moves can cause brief popping because the contribution scores haven't stabilised yet for the new view direction. Heavy weather scenes cull more aggressively to keep frame rates stable, which is why distant scenery can look sparser inside a storm than outside one.

some active sceneries in add-on linker were detected, and some werent

It only partially worked for me, had quite a lot of scenery added through addon linker which was just not picked up as even existing

I'm accessing Avsim on a PC and on my mobile phone, using Edge on PC and Chrome on mobile, and still not seeing ANY ads like those mentioned by other people...Go figure...

I'm on mobile chrome and not seeing any of these ads...

Chris, what have I told you about visiting those dodgy OF sites? 🕵️‍♂️

I'm not getting anything at all...

I upgraded from 10850K to 9800X3D. Runs at 70C flat out, undervolted and overclocked. Great performance, very happy with it

For me, it seem its the nose gear only that really is apparent as culled. so I'm not sure if this explanation is responsible for what I see?

This is my experience as well, not all models seem to perform in exactly the same way. I'm not enough of an expert to know why, or how to correct them, nor have I documented which perform and display differently (yet). If I can pin any info down I'll send it to AIG to see if that helps them any.

I still think it's an issue at the individual model level, as I've said from the start....

Its possibly related to broadband contention ratio? If so, little you can do about it other than change to a different provider. Although, I've found connection to MS servers have sometimes been problematic from my geographical location, and using a VPN to pretend to be in a different part of the world suddenly made a massive difference. Go figure! A broadband contention ratio represents the number of users sharing the same data capacity or bandwidth from an ISP, directly affecting connection speeds. A lower ratio (e.g., 20:1) offers higher, more consistent speeds, while a high ratio (e.g., 50:1 or 100:1) often causes significant slowdowns, especially during peak usage hours. Key Aspects of Broadband Contention Ratio Typical Ratios: Residential: Often around 50:1 (50 users sharing 1 line). Business: Usually lower, such as 20:1, for better performance. Leased Lines: Often have a 1:1 ratio, providing dedicated bandwidth. Peak Times: High contention causes the most noticeable slowdowns during high-usage hours (e.g., 9 am–11 am or evening, depending on if it is business or residential). Cost vs. Quality: Highly contended services are cheaper but offer lower, variable speeds, whereas low or 1:1 contention offers high, consistent performance at a higher price. Definition: The number of premises/users sharing the same internet line or bandwidth. Impact on Speed: When many users are online simultaneously (high contention), bandwidth is shared, reducing individual speeds and potentially causing congestion, similar to traffic on a road . Contention Ratio Examples 50:1: 50 users share the same bandwidth. If everyone is online, your speed could drop to a small fraction of the advertised maximum. 1:1 (Uncontended): You have exclusive access to the bandwidth, ensuring consistent speeds, common with leased lines

Agree with this, given you've tested throughput at 435 Mbps download, that's unlikely to be the limiting factor. I had a old powerline connection last year at barely 100Mbps which was enough. Most likely the GPU I'm afraid 😞

The certificate isn't the problem, its that danged memo on the left insisting on names in posts, or you'll be considered a failed airman...