CaptainLHKLM Posted January 23, 2022 Report Posted January 23, 2022 (edited) Hi all! I was wondering whether more people have pretty big fps fluctuations in the challenger as opposed to other aircraft, or whether I have a problem somewhere in my system that I have to sort out. This happens with the CPU-core fix too. The two screenshots were taken with the same weather, at the same time (with a quick x plane reboot in between) and in approximately the same place (LOPI2G out of EHAM). The fluctuations are definitely noticeable to they eye; is this what Cameron and HotStart have found with the way x-plane deals with textures (the problem that was found after we thought it was gizmo)? Specs are: i7 10700KF @ 4.7 GHz, RTX3080, 32GB RAM and X-Plane on an SSD. Best wishes and thanks for the hard work on the plane! Edited January 23, 2022 by CaptainLHKLM Quote
Cameron Posted January 23, 2022 Report Posted January 23, 2022 Yes, that appears to be the same issue. 1 Quote
CaptainLHKLM Posted January 23, 2022 Author Report Posted January 23, 2022 Ok, thanks! Looking forward to what LR can come up with to fix it Quote
Cameron Posted January 23, 2022 Report Posted January 23, 2022 1 minute ago, CaptainLHKLM said: Ok, thanks! Looking forward to what LR can come up with to fix it I want to tread lightly here. I don't want to call this a "bug", but more a shortsighted situation for the day Laminar didn't account for coming as soon as it did. The workaround we gave is essentially what you can expect to be mostly baked into X-Plane 12, where we'll be able to tell X-Plane ourselves how many CPU cores to reserve for us. Right now you're having to manually do that with the info we gave you. In other words, it will put the responsibility back on us to tell X-Plane what to do once Laminar allows that. I can't speak to performance of 12 as a whole, or if the texture paging system is more refined in 12 either. It's possible it is, or it's possible it's the same as now, in which case you should not assume a miracle change. Just setting the proper expectation here, and setting the record straight that calling this a bug is also not correct. Your CPU is a 16 threaded machine, so you really wouldn't be in the camp of people we would expect to use the guidance we provided either. Keep in mind that we also aren't personally auditing everyone's configuration. Ortho scenery, object scenery, rendering settings, etc all bring this stuff into play. I'd be interested to see your X-Plane rendering settings just for the sake of knowing. Any ortho scenery here too that the paging system is playing around with? 1 Quote
CaptainLHKLM Posted January 23, 2022 Author Report Posted January 23, 2022 In terms of ortho, I have mainly ZL16 around the main airports I fly into (one of which is EHAM). The total GPU load was never above 8.2 GB though, out of 10, so I think it should be possible. I have also attached my settings. Things like proper CPU cooling, creating an AV exclusion for X-Plane, making sure game mode is not on, etc. have all been taken care of, so I also wonder why I'm seeing this performance (flightdeck2sim's stream looks a lot smoother). Best wishes! Quote
Ben Russell Posted January 23, 2022 Report Posted January 23, 2022 You need to take into account the fact that the os runs its gui on the gpu. You probably had a Web browser open, also using gpu resources. Possibly Spotify, probably gpu again. And then theres memory fragmentation. Turn the texture quality down and you should see improvement. There is also a beta gizmo update brewing that should lead to less dropped frames but it won't fix drasticly overloaded and struggling system resource contention that drags on for multiple seconds. 1 Quote
CaptainLHKLM Posted January 27, 2022 Author Report Posted January 27, 2022 (edited) Thanks for the suggestions, but unfortunately dropping the textures down didn't help (still have the same issue where the fps drops substantially every 10 seconds, either external or internal, either moving or stationary). Do you want me to run some tests for you (logs, video recording, etc.) or should I just wait till another update addresses the issue (I think Toto said that you've found the issue 95% for sure?)? Best wishes! Edited January 27, 2022 by CaptainLHKLM Quote
Cameron Posted January 27, 2022 Report Posted January 27, 2022 15 minutes ago, CaptainLHKLM said: (I think Toto said that you've found the issue 95% for sure?) I think you're referring to something that was thought prior to our discussions with Laminar about the CPU/GPU texture paging situation. Gizmo is not at fault, as has been explained. There was a topic elsewhere here where someone suggested some other plugins can cause these stutter drops as well, like the Skunkcrafts updater. May be worth doing an audit on your plugins just to see. FlyWithLua and anything SASL based would be a good start. 1 Quote
CaptainLHKLM Posted January 27, 2022 Author Report Posted January 27, 2022 Hi Cameron! I'm referring to a post made by Toto on the 24th of January: "As a bonus, we've narrowed down the "stutter" or "slowdown" issue and we're pretty well 95% confident we understand what's causing that." As for the plugins I'll definitely give it a try tonight/tomorrow but as it is I run a really small group (xE, headshake, ground handling + better pushback, universal de ice and that's about it). I'll report soon! Quote
Cameron Posted January 27, 2022 Report Posted January 27, 2022 5 minutes ago, CaptainLHKLM said: Hi Cameron! I'm referring to a post made by Toto on the 24th of January: That was in reference to some tests we did and discussions with Laminar about the use of DDS and PNG files, and how the texture pager differently interacts with them. PNGs are slower to process, and they must be re-fetched from disk. This slows things down and causes a stutter. It's a piece of the puzzle for objects that get "killed" when you're not looking at certain angles (like facing forward in the cockpit, but not at the engines). When you pan the camera in these situations, there is a momentary pause while X-Plane has to go all the way back to disk and fetch the files again. We solved that situation by providing DDS files, which are already compressed and contain mipmaps. In other words, no, he was not referring to Gizmo. Quote
CaptainLHKLM Posted January 27, 2022 Author Report Posted January 27, 2022 (edited) Ah sorry if I ever came across sounding as if I was blaming Gizmo. I didn't try to imply that, my bad if it came across as such! I tested a little myself and Gizmo never took up any extra computing power while I had stutters; and the stutters went together with increased GPU "Copy" usage so I completely believe that the issue is texture-paging based and not Gizmo based! Edited January 27, 2022 by CaptainLHKLM Quote
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.