tfm

V quick download for me: 3.6Gb at 300-770Mb/s. Quick flight in the Fenix A320: nothing seems broken, settings retained, performance similar without 3rd party traffic, perhaps slightly improved when adding FSLTL traffic, switching to external view now has no “lag” before being able to zoom around smoothly.

Some further tips, assuming ILS approach: 1. FSIPanel is useful (invaluable!) for landing practice: it can set you up on short finals in a few seconds so you make good use of time. Best settings: 2nm out, 1000ft. It exacts high / unforgiving standards of precision, which of course is great for training. 2. Disconnect autopilot at 500ft. Retain auto throttle. 3. Look mainly at instruments down to c100ft. Thereafter mainly / wholly visual. 4. Aiming point = adjacent to PAPI. 5. The aeroplane will be stable after disconnecting autopilot, under most conditions. Expect to make small corrections, not large changes. 6. Initiate flare at 30-40’. Very gentle back pressure on the side stick: probably less than seems plausible! 7. Throttles smoothly and swiftly to idle at 15-25’. 8. if you do all of that and crossed the runway threshold at 50’ then it’s probably safe to “take what you get” rather than chasing a “butter” landing.

I also have a fair few WinWings products: about £1000 worth in total so far, and I think the quality is quite high at a relatively affordable price. But as an organisation they do seem pretty chaotic: maybe struggling to scale up and trying to handle an international market without incurring the expense of using third party logistics? Having pre-ordered this throttle back in October, I was recently sent a link to make payment, but it took me to a page saying it thought I was in the USA with some wordy gibberish about trademarks and needing a different website … which also didn’t work. Very frustrating. I think I may eventually have worked my way around it but only time will tell!

Am I just imagining it, or has the big performance boost with the early builds in SU4 been eroded with successive builds? I’m fairly sure the old FPS counter has taken a bit of a dive but maybe I’m hallucinating.

Oh sorry to hear that, still not getting anything here that I would call a stutter but your standards might be higher than mine.

I have also found that this build brings better performance: in my case up from c80fps to c108fps (using Framegen, ofc) under like conditions in the Fenix A320 at Inibuilds EGLL. Also in today’s nasty weather I again got a “windshear” warning in the Fenix.

Sounds logical! But on the other hand, I like using the WinWings FCU and MCDU to minimise use of the mouse: presumably this is foreclosed with VR?

A few tips from someone who struggles to get consistent landings with this model, gleaned from when I’ve nailed it: - concentrated practice helps: commit an hour a day to doing a dozen or so, one after the other. FSI Panel makes this achievable and also provides a readout of how far the centreline you land, touchdown FPM and how much runway you’ve consumed. - mentally, you need to focus super-hard from about 100ft AGL and be meticulous in observing the closure rate, as manifested by the ground coming towards you, scanning the full length of the runway and making the most of peripheral vision. - Don’t laugh but with enough practice you find that those fleeting seconds before touchdown seem to expand: you realise that there IS enough time to NOTICE the closure rate and relate it to the inputs you apply to the sidestick. Being able to make “time (seem to) slow down” in this way only comes with experience but it’s crucial imo. - until you develop that level of awareness, it’s hard/ impossible to judge how to “squeeze” the stick from the point of view of timing and also amount of travel in the stick - for this to work best, you do indeed need a linear or near-linear sensitivity curve and minimal dead zone : you are looking for a predictable, smooth response from your inputs in the final seconds - a big monitor also helps: 55” is a different league from 32” and helps massively in training your brain to recognise closure rate/ distance from centreline / approach course - usually, DO NOT commence the flare until you have heard the 30 feet call out - as exceptions to that rule, you might do a tiny “pre-flare” at about 40 feet if you have a tailwind and / or are coming in a bit hot (glance at your descent rate as you cross the threshold: if it’s over 750 fpm you might need a pre-flare) and/or if you’re coming in heavy (eg, I’ve found that if I’m above c61000kg, I’m more likely to want to do a tiny pre-flare) - throttles to idle comes AFTER initiating the flare, never before. As a rule of thumb, expect to snap the throttles to idle as you hear the 20 feet call out - but experience will let you refine this (and everything else!)

@Tony P Hi - I use FSI Panel to practice landings. I found it salutary : I always thought my landings were ok, turns out they’ve been execrable. With the help of FSI Panel I’ve become much more self-aware especially of the importance of precision. My landings have improved but tbh they still suck 🤪.

I haven’t noticed this but neither have I looked for it 🙂 Others have reported aircraft sometimes taxiing with “gear up”. Could that be the phenomenon you’ve observed?

In my completely unscientific testing, this build seems regressive in two respects: - the spawning of traffic and other small items seems to be more noticeable than before. - Also, quite noticeable stutters / jerkiness in finals.

I don’t usually watch YT videos but this comment intrigued me because I don’t recall observing stuttering that might count as “terrible” or “desperate” since about FS9. Being as dispassionate as I can, I couldn’t see any stuttering at or after 4:25. Watching a second time - maybe there was a tiny jerk at one point, but so fleeting that it might equally have been a blink of the eye. This tells me that some of the discussions on here, as elsewhere sadly, are like ships passing in the night! People have completely different perceptions of the same thing, and there is no way acceptable to all participants of arbitrating who is “right” and who is “wrong”.

When you eat toast, how do you tolerate the ghastly crumbs?

Only by inference from the observed framerate delta between v3 and v4b: I don’t run monitoring software to indicate those metrics in real time.

Pretty much all I do in FS is make short flights in the Fenix A320, usually between airports for which I have installed detailed add-ons such as EGLL and LFMN. In my particular use case, the only downside I’ve noticed with SU4 is that not all the airbridges are depicted, or operable, in the airports. Apart from that, I have not noticed any attenuation in visual quality or detail, or any sacrifice in traffic density or LOD or the like. In fact, if anything, my impression is that there is more traffic on the roads and more ships. Performance was fine with SU3 but with SU4 it has improved appreciably as measured by the built-in FPS counter, from ~80fps in SU3 to over 100fps in SU4 at my “stress test” location, 4k but using DLSS and FG of course. In the cruise I have seen the framerate exceeding 150fps: unprecedented in my experience. Under stressful conditions, takeoff and approach and finals, the framerate fluctuates less, and within a smaller range, than SU3, meaning that the overall effect is more smooth. I have not noticed any “stutters”: I’m not claiming there are zero stutters, but if there are any stutters then I haven’t found them to be a distraction (unlike some previous versions).

Yes! An aural warning (a male voice saying “windshear, windshear”) plus “WINDSHEAR” in red text on the PFD. Very disconcerting!

Practising my useless landings with FSI Panel at EGLL 27R using “live” weather (it’s super windy here right now). For reference this is c1330h on 15/9/2025. I didn’t know that MSFS could simulate windshear. In c40 years of flight sim, I’ve never before had a “windshear” alert and truthfully I assumed it was non-functional in the Fenix - but it works! Thank goodness it was only in a simulator because it was genuinely scary. I definitely do not have what it takes: there were two alerts on short finals. I recovered from the first at c680ft AGL and was feeling smug when the second one hit at c200ft. I reacted too slowly 🪦 - but I question whether it was recoverable at all: it was like walking off a cliff.

That’s also my assessment of what’s changed: it explains why we’re not (or not all!) seeing huge increases in FPS across the board but, rather, less of a “hit” in the most demanding situations. The load has been distributed more evenly across available cores, like forcing traffic to use all the lanes on the motorway instead of travelling in single file: same speed limit + less congestion = higher throughput.

My first time with a v24 beta. EGLL to LFMN in the Fenix A320 latest update, traffic injected by FSLTL. Uneventful, smooth flight. I didn’t get the huge performance improvement reported by some. But the framerate seemed to take less of a hit than before when adding FSLTL traffic, which I guess is consistent with there being more “headroom” in the new version: maybe better use of multi-threading?

Oops I meant vPilot not FSPilot 👴

(Sorry - misunderstood the question with my initial reply). FSPilot injects Vatsim traffic independently of MSFS’s online service. You need to disable native / online traffic as fully as possible, to ensure that the only traffic you see is what FSPilot injects. For the same reason, do not use the FSLTL injector when using Vatsim.

Yes that is the tendency of all of us! But if you’re a getting a crash in situations which cause no issues for others, then there is something specific to your case. Sometimes it turns out that you’re running software alongside FS, which is causing hiccups. But - one would hope - most people with persistent issues will have excluded this, by using a fresh installation on a wiped disk. If your issues continue after this measure, then it’s time to confront the disagreeable truth that there is something amiss with your kit.

I know, you’d think so, right? And yet again and again over the years/decades, FS in its various forms has always managed to stress the components in ways that the tests can sometimes miss … 😵‍💫

One of the better-known stress test programmes.

Infuriating! The fact that others don’t seem to be getting this experience indicates, sadly, that the latest beta has exposed an instability in your particular hardware: not something any of us wants to hear! 😖 FS has always had the potential to reach parts that OCCT etc cannot reach 😕