Constant 20s long freezes + every time I save

• Jan 20, 2024 - 16:38

It seems far more frequent than the 5 minutes of autosave, and it's too long anyway. It's particularly bothersome given that I need to frequently save to close and restart the program due to bugs that make it arbitrarily not playback some notes, and then there's the long "loading samples" time... and 2-3 seconds of lag upon note input etc.
I've checked the forum (by the way, why can't we order search results in the forum by date? Or by version? Or by anything) and users with similar issues seemed to have their saving folder synced up to some cloud service. This isn't my case.


Comments

Your free account here on .org can also be used on the .com site. You can save to the cloud from within MU4. doing so will save your score to the .com site. However, I always feel it safer and easier to only save to my computer.
What are your computer specs?

In reply to by Nicola Rulli

You might need to make a few audio system adjustments. Starting with (in MU4) Edit>Preferences>I/O and change the Buffer to 4096. And make sure your actual playback device is chosen. I.E. if you have headphones plugged in, headphones (not default) should be selected. There are more settings to check. These are listed in a "How To" found in the Support page.

In reply to by bobjp

Buffer is already on max, sample rate is already 41k instead of 48k, reducing bit amplitude just makes audio saturate. The built-in speakers and headphone jack socket are the same "device", and they're already selected. The mixer is already turned down. Exclusive mode is already unchecked. The only thing I have left to try is keeping the mixer open...

In reply to by bobjp

oh, right, that's why I hadn't used the "keep the mixer open" method thus far: it stays in the foreground. As if there wasn't enough clutter. They really should make it able to go behind the main window, if keeping it open is going to be THE way to make the program at all usable...

In reply to by Nicola Rulli

Yes, if having the mixer open makes a difference, then undock it and drag the right side to the left to make it out of the way. The down side is that you have to re-size it every time you want to use it.
On of the computers I use is similar to yours but only two cores. It doesn't run MU4 perfectly, but it doesn't have as many sever problems as you do. There must be something that can be done.

In reply to by bobjp

That musescore takes 770mb of ram and that slightly climbs during saving. During saving, its CPU usage climbs upwards of 15%, then as musescore winds down, the antimalware climbs to 13% from its usual 0 to 0.5%. During laggy note input, the CPU usage of musescore climbs to 19%. Ram usage seems to steadily increase over time during a session, and that might be why at some point some notes stop playing back - EDIT: no, it just went down when I deleted some notes?

I can confirm that it isn't just the autosave's fault - I had such a freeze as a result of trying to switch tabs in the left panel right after opening the program (and after it was done loading), so wayy before the 5 minutes timer of autosave.
Also, the RAM use which inflates over time during a session isn't fully reset upon a new load, the new minimum is still higher than before, and by a considerable enough margin that I don't think it can be due to the few extra notes I inserted in the meantime. I'll check if that's still true after a reboot of the PC or if it resets and thus we can assume it's due to data lingering in the RAM when Musescore is shut down.

In reply to by Nicola Rulli

So, it isn't a resource issue. There are dozens of services running on your computer at any one time. And they change all the time. Just sitting, readings will change. All I can say is that on my Surface Go, which has lower specs than your system, I don't have the delays that you do. Playback tends to studder after I stop, in a big score. But is fine otherwise. Sometimes a slight delay selecting a note. I only have it for testing. I don't do any serious work on it. And it only started being as good as it is after I did some serious system cleaning. Junk files and registry.

In reply to by Marc Sabatella

Here it is. I don't think the score's the problem.
1. Saving > ultra long freeze ("does not respond")
2. Autosave > ultra long freeze ("does not respond")
3. Inserting notes with the computer keyboard > a few seconds of lag, from 1s to 6s approx. Gets longer if you input more notes in a row
4. Selecting an item > slightly under 1s of lag, sometimes more
5. Mass edits (show/hide empty staves, copypaste 30+ measures, change page margins, I think also toggle multimeasure rests) > ultra long freeze ("does not respond")
6. Toggle slash notation on something written on voice 1 > noteheads remain normal, position and cue size values sometimes mimick the last slash notation item to have been modified (without having saved those edits as default)
7. Select-shift/ctrl-select > 1/3rd of the time, the collective selection will appear then disappear and only the last clicked item will remain selected
8. Select one or several measures > the box will sometimes not reach all the elements, nearly always with invisible ones but often also visible ones (chord symbols, dynamics, articulations)
9. Select one or a partial measure > the box will sometimes expand indefinitely below or above the intended selection
10. Navigate between notes with left/right arrow keys > slightly over 1s of lag, increases dramatically when passing through a pause
11. Insert scoop/slide in > the target note's playback ends earlier than it's supposed to even begin (I had to individually set them all to not play)
12. Insert glissando between two notes > the "gliss." text will appear, even though if you check in properties the default text is set (by me) to empty, so they spawn in already out-of-default
13. Insert invisible tempo text and right click it > often the selection will be immediately lost before the drop down menu has time to appear, and the tempo text will become unclickable
14. Click and drag the timeline > it will move in an extremely jerky manner
15. (with italian keyboard) shift+0 as "=" doesn't work for inserting natural accidental, even if we change the default shortcut and move it onto the "toggle" action (like sharp and flat are by default) instead of the "add" action. Likewise, hairpin decrescendo ("shift+.", which is ":" on italian keyboard) works whereas hairpin crescendo ("shift+,", which is ";" on italian keyboard does not
16. (on a measure that had by mistake been filled with slashes twice, on voice 1 and 2, and had extra material written on voice 3) switching voice 2 and 3 then deleting the slashes, then resulting rests of voice 3 resulted in MS being unable to save, claiming "found: 2/4, expected:3/4" despite the bar having visual contents that correctly amount to 3/4. The only fix that worked was deleting and rewriting the bar. A nearby bar of identical before-and-later contents, despite having been edited in shared selection with the problematic one, did not present the same problem.
17. Upon opening, the docked timeline will be empty, apart from a small vertical column of contents situated where the timeline's vertical scroll bar is supposed to be. This will go back to normal upon resizing the timeline
18. While editing text, there will be significant lag in both character input and arrow navigation, and the cursor will not blink, making it impossible to visually tell whether or not there's a freeze going on without inputting something
19. Had a single crash during playback of the whole piece, didn't happen again
20. A glissando in the first part of the piece (I think the one on measure 3) played incorrectly in playback and mp3 export a few times, but I didn't figure out what I was doing that was any different that usual, and I can't recall exactly how it sounded (I think it may have immediately stopped sliding down towards the target note and instead went straight to playing a note lower than the target note?)
21. Press play > normally very short delay (completely acceptable) but occasionally several seconds of lag. Occasionally freezes while playing and can't be paused.

Attachment Size
La Gran Banda-2.mscz 751.83 KB

In reply to by Nicola Rulli

I've loaded your score. The steps you list seem kind of random and disconnected and not related to the original problem you describe. Can you reduce it to one simple set of steps that results in the 20 second delay? You said it happens on save, so I tried loading it, then immediately doing a "save as". I can confirm that since it is a very large score it does take a while - not 20 seconds on my system, but perhaps your hard drive is unusually slow? Anyhow, large files take a long time to save, that is true.

I did check a few of the other random unrelated performance issues you mention but couldn't reproduce any. For instance, navigating via cursor key is instant as I expect. Same with entering notes or editing text. Could be your processor is also unusually slow? Mine is many years old and a "mobile" one at that so definitely not top of the line.

Some of the other random things you describe seem unrelated to performance at all but to details about the rendering or playback of specific elements. Best to start new threads to discuss those - one thread per issue, please. Then we can understand and assist better. And if someone can confirm any of the issues using the score and the precise steps you supply, then you can open an issue on GitHub to report it to the developers.

In reply to by Marc Sabatella

It must be my processor or some other part, because my disk is a brand new SSD (my old HDD had begun having minute-long periods of unresponsiveness and 100% use, whereas now disk usage during saving of a musescore project peaks at 2,5% and averages around 1%, and I never have those computer-wide freezes anymore,). However, as stated in another comment, nominal resource consumption stays below 50% for both ram and cpu, so I'm not sure what other bottleneck there could be
I listed all of the issues I have that don't seem to happen frequently to other users (I also have several issues that are common, IE mixer size, occasional inability to click invisible elements, ties going crazy especially in continuous view, audio stutter at standard sample rate and buffer size, etc. and I didn't include those), since them being rare and happening to a machine that has so many performance issues might indicate that it's the program messing up because it doesn't get its "answers" quick enough from the machine.

In reply to by frfancha

Yeah, but I had no such problems (and still don't) with MS3. Or rather, I had a standard lag upon any basic action that was 25% of the lag I get for the same action in MS4, and it didn't increase more-than-proportionally but rather proportionally (or even less-than-proportionally) by increasing the number of affected elements / giving many inputs in a row, whereas it does in MS4.

I THINK I FOUND THE PROBLEM. It might be related to the "hairpins not showing up in parts" problem.
Aaaaaand now the site isn't accepting my log files as an attachment. Why? It gives no error. I press okay and nothing happens... Anyway, apparently any time I save, or an autosave happens, there's a bunch of these:
2024-01-26T20:23:59.900 | WARN | main_thread | SpannerWriter::SpannerWriter | SpannerWriter: spanner (HairPin) doesn't have an endpoint!
2024-01-26T20:23:59.901 | WARN | main_thread | SpannerWriter::SpannerWriter | SpannerWriter: spanner (HairPin) doesn't have an endpoint!
2024-01-26T20:24:00.078 | WARN | main_thread | SpannerWriter::SpannerWriter | SpannerWriter: spanner (HairPin) doesn't have an endpoint!
2024-01-26T20:24:00.079 | WARN | main_thread | SpannerWriter::SpannerWriter | SpannerWriter: spanner (HairPin) doesn't have an endpoint!
2024-01-26T20:24:09.418 | WARN | main_thread | SpannerWriter::SpannerWriter | SpannerWriter: spanner (TextLine) doesn't have an endpoint!
2024-01-26T20:24:09.423 | WARN | main_thread | SpannerWriter::SpannerWriter | SpannerWriter: spanner (TextLine) doesn't have an endpoint!
2024-01-26T20:24:09.428 | WARN | main_thread | SpannerWriter::SpannerWriter | SpannerWriter: spanner (HairPin) doesn't have an endpoint!
2024-01-26T20:24:09.429 | WARN | main_thread | SpannerWriter::SpannerWriter | SpannerWriter: spanner (HairPin) doesn't have an endpoint!
2024-01-26T20:24:17.749 | INFO | main_thread | NotationProject::doSave | success save file: "C:/Users/Hp/Desktop/Tesi/La Gran Banda-2.mscz.autosave"

Note that the actual save happens 18 seconds after the first error.
It would be interesting to see if users that have the same lag problems all have these in their log, and users who don't, don't.

Do you still have an unanswered question? Please log in first to post your question.