KVR Audio

layzer · Post by **layzer** » Wed Mar 30, 2016 11:47 pm

ZonderP wrote:
layzer wrote:just tried the latest x32 win build. it crashed EnergyXT2 (which rarely ever crashes on my laptop) i
was just playing a few chords. no details given. my be some other plugin but we'll see if anyone
else reports daw crashes.
Same for me...
Windows 7 64 bit, EnergyXT 2.7, PG-8X (2016-03-28b)
Finally managed to record some bars before the crash. Always crashes EnergyXT2 after playing those bars in a loop after some time. This happens reproducible with Juno-60 Bass preset in 'DrewDalePG8XBank2.PG8Xbank' after some time.
No fancy stuff like automation or MIDI Learn involved...

yes, i did not mention that it does not crash immidiatly, but after several notes have been played. strange!
no controls moved, nothing altered.

bstageboss · Post by **bstageboss** » Thu Mar 31, 2016 5:31 am

Had a crash in REAPER too, after getting a warning message that REAPER was going over it's allotted memory maximum. This happened shortly after loading and playing some patches.
After some investigating I found that after a while, the RAM use skyrocketed (or at least increased steadily until it hit the limit I set in the preferences). I deleted the plugin from the track but the RAM use persisted, although it did not increase further. The plugin also showed up in the list of plugins in the performance meter, even though it was no longer on the track. It remained in the list even after I deleted the track itself (no other tracks were added); same with the RAM use.

I'll see if I can post a screenshot of the performance meter here.

STV

layzer · Post by **layzer** » Thu Mar 31, 2016 6:47 am

bstageboss wrote:Had a crash in REAPER too, after getting a warning message that REAPER was going over it's allotted memory maximum. This happened shortly after loading and playing some patches.
After some investigating I found that after a while, the RAM use skyrocketed (or at least increased steadily until it hit the limit I set in the preferences). I deleted the plugin from the track but the RAM use persisted, although it did not increase further. The plugin also showed up in the list of plugins in the performance meter, even though it was no longer on the track. It remained in the list even after I deleted the track itself (no other tracks were added); same with the RAM use.

I'll see if I can post a screenshot of the performance meter here.

STV

yep, def something wrong. i loaded up task manager and then PG8X in EXT2 and watched "commit change" in the bottom right climb steadily from 644M/6875M as i played some notes till it crashed EXT2.

martin_l · Post by **martin_l** » Thu Mar 31, 2016 8:19 am

Ouch!

I just discovered that I forgot to comment out an memory allocation in the main process routine.

I was experimenting with some code to try to fix the aliasing problem and introduced some temporary arrays. I did comment out the delete[] lines, but not the "new" ones, hence the quite severe memory leak.

Will try to fix it tonight.

Thanks for the detailed bug reports.

Martin

Teksonik · Post by **Teksonik** » Thu Mar 31, 2016 3:37 pm

Just to let you know the latest version (3-28) experienced Stuck Notes in FL Studio 12.3 (beta). The "panic button" in FL did not work to shut off the stuck notes, only deleting the plugin would stop the notes. Not sure if this related to the memory leak issue but thought I would report it anyway......

martin_l · Post by **martin_l** » Thu Mar 31, 2016 7:03 pm

New versions (2016-03-31) uploaded, which should fix the memory leak.
That should (hopefully) also solve the stuck notes in FL. If not, give me a shout.

Cheers,
Martin

sprnva · Post by **sprnva** » Thu Mar 31, 2016 7:31 pm

Thanks for the updates Martin! This new one is working well here. Although there's still a small jump in CPU usage when the plugin is first opened (from 7% to 12%), once it's producing sound the usage doesn't vary whether the plugin UI is open or not. Maybe it's just something with how it initialises in Live.

Initially I noticed higher usage in a small project I used it on, but then comparing the ini files I noticed that the old one had the default voice count at 6 while the new one defaults to 12.

Finally, regarding the ini file, it's created when the plugin is in my user library vst folder, but not when it's in the system library vst folder. This is usually where I keep my plugins. Maybe it's a sandboxing thing with 10.10? I can work around this by putting the plugin into my user library first and letting it create the ini file. Then moving both files to the system library.

martin_l · Post by **martin_l** » Thu Mar 31, 2016 7:59 pm

sprnva wrote:Thanks for the updates Martin! This new one is working well here. Although there's still a small jump in CPU usage when the plugin is first opened (from 7% to 12%), once it's producing sound the usage doesn't vary whether the plugin UI is open or not. Maybe it's just something with how it initialises in Live.

Initially I noticed higher usage in a small project I used it on, but then comparing the ini files I noticed that the old one had the default voice count at 6 while the new one defaults to 12.

Finally, regarding the ini file, it's created when the plugin is in my user library vst folder, but not when it's in the system library vst folder. This is usually where I keep my plugins. Maybe it's a sandboxing thing with 10.10? I can work around this by putting the plugin into my user library first and letting it create the ini file. Then moving both files to the system library.

Yes. If you want it in the System VST folder, it's best to create an empty PG-8X.ini file by hand. You might have to do that with admin permission.

Cheers,
Martin

provoc · Post by **provoc** » Thu Mar 31, 2016 9:56 pm

Old preset loaded in latest version in mono key mode instead of unison but seems ok otherwise.
Reaper x32 on Windows 10 x64

layzer · Post by **layzer** » Thu Mar 31, 2016 11:59 pm

testing latest win32, looks like the "commit change" in task manager is staying put as i play notes now.
i can keep playing for a few hours, but i'm pretty sure you have fixed it, Martin.

on to more testing!

Teksonik · Post by **Teksonik** » Fri Apr 01, 2016 12:55 am

martin_l wrote:New versions (2016-03-31) uploaded, which should fix the memory leak.
That should (hopefully) also solve the stuck notes in FL. If not, give me a shout.

Cheers,
Martin

Hi Martin, sadly the stuck note issue is still present. Can someone else check with FL Studio please ? I'm using Beta 12.3 so I want to make sure that it's not an FL issue. Although none of the other plugins experience stuck notes.

martin_l · Post by **martin_l** » Fri Apr 01, 2016 6:12 am

Teksonik wrote:
martin_l wrote:New versions (2016-03-31) uploaded, which should fix the memory leak.
That should (hopefully) also solve the stuck notes in FL. If not, give me a shout.

Cheers,
Martin
Hi Martin, sadly the stuck note issue is still present. Can someone else check with FL Studio please ? I'm using Beta 12.3 so I want to make sure that it's not an FL issue. Although none of the other plugins experience stuck notes.

Could you, please, give me some more information? In particular, which key mode you are using (poly1 or poly2), is portamento on or off?

Could you send me a MIDI file, which causes stuck notes? I am particularly wondering whether there are overlaps on the same note (which is meaningless, but can easily be produced by most DAWs)

Do you also have a non-beta version of FL? Does it behave the same way?

Cheers,
Martin

Teksonik · Post by **Teksonik** » Fri Apr 01, 2016 1:37 pm

martin_l wrote: Could you, please, give me some more information? In particular, which key mode you are using (poly1 or poly2), is portamento on or off?

This is using the INIT patch however it is set.

martin_l wrote:Could you send me a MIDI file, which causes stuck notes? I am particularly wondering whether there are overlaps on the same note (which is meaningless, but can easily be produced by most DAWs)

It occurs when I simply hit one note on my midi controller keyboard. It also occurs when I enter a single note into the Piano Roll so no overlapping notes.

martin_l wrote:Do you also have a non-beta version of FL? Does it behave the same way?

Not on this system which is why I asked others to test. I will try on the studio computer the next time I boot it up. I have a non beta version of FL and several other DAWs to test on there.......

sergiofrias · Post by **sergiofrias** » Fri Apr 01, 2016 6:23 pm

Hello ,i'm using Fl studio 12.2 (beta 3) x64

No stuck notes in my system!

Note to the aliasing on my previous post:

maybe the problem resides in the ADSR envelope and not in the filter.
i'ved compared a similar frequency sweep modulated with the lfo ,and a frequency sweep with the envelope( same frequency range ,atack and decay [1 sweep only])
the aliasing spikes in the LFO method are very low compared to the ADSR envelope method.

martin_l · Post by **martin_l** » Fri Apr 01, 2016 7:01 pm

sergiofrias wrote:Hello ,i'm using Fl studio 12.2 (beta 3) x64

No stuck notes in my system!

Note to the aliasing on my previous post:

maybe the problem resides in the ADSR envelope and not in the filter.
i'ved compared a similar frequency sweep modulated with the lfo ,and a frequency sweep with the envelope( same frequency range ,atack and decay [1 sweep only])
the aliasing spikes in the LFO method are very low compared to the ADSR envelope method.

Yes, the aliasing only appears with the envelope. And it is true that the envelope generates faster slopes than the LFO. In particular because the envelope is "stepped". If I LP-filter the envelope enough to get rid of the aliasing, I cannot get sharp attacks anymore.

So, I am currently looking into improving the filter algorithm, in particular a topology preserving method, which should behave better with time varying parameters. It will be tricky, though, to achieve that without additional CPU hit.

PG8X (inspired by the JX8P): new beta version uploaded