Installed the beta, and had a flight. On triple screen, 4K, i can keep bertween 40 and 60 FPS, very smooth. But I've had a flight with the fenix, and randomly I'm getting very big stutters for couple of seconds. then back on track but the fenix it's like bouncing/falling before getting back to normal flight. What may it be?

Trying to update. Stuck decompressing an Asobo airport (heca cairo). Been 15 minutes. 

51 minutes ago, AnkH said:

The question is: do you really still need the framerate lock if you are able to get more than those 30 FPS when using the AutoFPS tool?

Yes, because my monitor is not G-sync or Freesync, it's fixed 60hz, so I must lock either to 60 or to 30, otherwise I lose smoothness: 40 fps, in my case, looks worse than 30.

7 minutes ago, virtuali said:

otherwise I lose smoothness: 40 fps, in my case, looks worse than 30.

Even with the AutoFPS app? 

Edited March 13, 20242 yr by AnkH

3 minutes ago, AnkH said:

Even with the AutoFPS app? 

Yes, because even if I set a 30 fps target, if I don't lock it, AutoFPS doesn't keep it *exactly* to 30 but, instead, it's always a bit more, in the 36-45 range, which is perfectly fine for GSync/Freesync monitors (it's better, actually), but not with fixed refresh ones.

Edited March 13, 20242 yr by virtuali

5 hours ago, MarcG said:

Well this doesn't bode well, as per ever all benchmarks run identically as possible, I thought there was a discrepancy with the first SU15beta benchmark so ran it again....same result 😐

Well done for doing the CapFrameX comparisons! It's pretty much in line with what I've noticed. Initially I thought I had a tad better performance in first beta build, but I could very well have been influenced about all the hype when a new beta drops. 

I did get very decent performance in SU14 though. IMHO at least the performance in SU15 is not worse. But again, I haven't done any scientific comparisons like you did.

Just did some comprehensive testing low-flying over Manhattan at 180kts 1500ft w/ 4K HDR DLSS ultra quality most settings ultra and TLOD around 500, OLOD 1000. I'll share the settings here, could be interesting for some, dunno...

 

Note: Those aren't the best FTV values I've ever seen, but for me, in this type of situation, it's more than good enough. 

And btw I got some issues with buildings not being fully rendered today that I didn't see yesterday. Or perhaps "popping-in" - don't know exactly how to describe but it should be easy to see in the video. Could be cause of lower altitude, server connection etc etc etc I truly dunno.

I think someone (in this thread?) made a comment re: VPN (EDIT: @Nemo). I honestly don't know if it makes a difference. Some MSFS-days are good, some are worse... I do notice a huge difference though when downloading e.g. WUs and such when using VPN connected to the "right" servers though. HUGE difference (2-3Mbps vs up to 300...)

EDIT: Definitely something up with "server connection" today, just got a "bandwidth too low for photogrammetry" message. Also, I just now noticed that Ethernet usage (NET1 in the first screenshot) was quite high - 10.4 MB/s = 83.2 Mbps. Which is quite high. Way lower than my home fiber-optic broadband though. But somehow the connection to the particular MSFS server *might* have been not good enough to render the thousands of objects within a TLOD of 4-500. Again I dunno - these things are above my head. 

Currently running MSFS in the background (Vertigo in a holding pattern over lower Manhattan). HWiNFO reports around 9.5 MB/s average DL rate. Not downloading anything else at the moment. 

Edited March 13, 20242 yr by Cpt_Piett

1 hour ago, Cpt_Piett said:

Definitely something up with "server connection" today, just got a "bandwidth too low for photogrammetry" message.

if you wanted to do completely repeatable tests, you could pre-download the scenery and re-fly the same tour from your locally stored cache.

36 minutes ago, turbomax said:

if you wanted to do completely repeatable tests, you could pre-download the scenery and re-fly the same tour from your locally stored cache.

You're right, that would be the best option. Come to think of it, don't think I've used the non-rolling cache option (forget what it's called) for ages.