[Done] Speed Up Song Loading

Hi,

since the new beta 5 is out, Renoise takes longer to open a file, causes a high CPU load at that time and doesn’t respond for a moment. I’m not sure if this is wanted behavior, so I didn’t post this as a bug.

I noticed this on Linux, where the Compiz crash manager indicated that Renoise doesn’t respond.
This didn’t happen until the last beta 4.
It’s not very critical (for me), since Renoise continues working after the song is loaded but the process itself seems to have become heavier. I thought I’d mention this in case this is not really intended.

Gilzad

A high CPU load on loading is expected, this was always like this. But it seems loading this song now simply takes longer?

Have you changed sample rates, latency or other preferences for beta 5? All this can have a impact on how long loading a song takes.

Do you use VST plugins in that song?

It’s doing this on my system too. The bit where it says "“Releasing the old song…” was so fast I never noticed it but now it is taking 3 seconds.

Eeermm… For me it was already 2.1 which took about a minute for starting up and for closing it took differently - 10-30 sec maybe… But I cannot test it anymore as I totally fed up with Ubuntu -uninstalled it!- and switched back to Windows. Seven is doing good job and I have now more time for making music besides tweaking and fixing my system :)

Sorry for useless or maybe a off-topic post (I’ve been up for a couple of days in a row which makes it really hard express my thoughts in a fluid manner)

Should have mentioned this is on Windows 7 in my case.

I have this too, on Linux.

Here’s a basic benchmark. I created two completely empty songs a and b.

Beta4
Load song a directly after startup: ~6 sec
Open song b (unloading a): ~3 sec

Beta5
Load song a directly after startup: ~6 sec
Open song b (unloading a): ~6 sec

The interesting difference is in opening the second song: it’s a lot slower in beta5. I confirmed this several times alternately loading the two empty songs.

I used strace to spy on the renoise process. In beta5, every time I load a new song the process stops a second or two after writing MIDI: Using global MIDI actions from resources, which happens twice. In beta 4 those two stops are much shorter. I don’t know if that accounts for all the extra time, but it’s certainly some of it.

Please let me know if there’s something more I can do to help.

I guess this could be related with the maximum amount of midi automatable slots in the matrix which increased to 128… But i could be wrong!

Well they were 64, so 2* 64 = 128, why would 2*3secs not be equal to 6secs?

I’ll take a look at this. My guess is that this is sideeffect of this plugin related fix: Plugin Settings May Get Randomized With an Attached Automation Device
I wanted to be on the safe side here, but theres always room for optimizations.

Johan Ekenberg: Are there any or lots of plugins in “song a”?

Don’t think so.
Ekenberg:

It’s the same here. Loading an empty song in b4 takes about 5 sec, in b5 it takes about 9 sec.
First, the “Releasing the old song…” takes longer.
Also, when loading a song in b5 an Event-status-bar message will say “Prepare loading …” at some point for about 1-2 sec. I’ve never seen this “Prepare loading …” in pre b5 releases - maybe because the message is only shown in an instant in older releases?

Thats indeed the problem, but this will be easy to fix. Will do so for the next beta.

Thanks for the help guys!

Great! Perhaps move this topic to Bug Reports?

Does this make anything better?

Eh, sorry - just trying to be helpful. Won’t happen again…

Hey, that wasn’t serious. Relax ;)

Still, this is no “malfunction”, nothing which has high priority. And I anyway always do small performance tweaks here and there if theres time. Just need some hints where to look at, what should/could be faster…

:o Thanks, I thought I’d stepped over a line and my license was just about to be revoked prematurely, after you had remotely started “rm -rf /” through your hidden backdoor in renoise…

Great, I’ll keep strace loaded then if I find anything else that might merit performance attention.

Done. Should be much better in the next beta, maybe even a bit faster than in 2.1…

Great, looking forward to trying it (beta7, right?) out when it arrives. Will report any findings back here…

It seems to be working very well, thanks for looking into this!