MeldaProduction

scoplunk · Post by **scoplunk** » Sat Jan 16, 2016 7:38 pm

I'm a big fan of melda plugins and I find that I'm using them more and more on projects - usually in Logic. Unfortunately, now that I'm transferring some of the sounds from these projects into Mainstage for live use, I've run into a nasty problem. The two I've been trying to use - MVintageRotary and MMHarmonizer - won't unload once they've been instantiated. So, here's an example scenario:

1) start Mainstage. The opening patch is a kontakt sampled piano. CPU use 5%
2) select a patch that includes MVintageRotary. CPU use 14%
3) go back to first piano patch. CPU use 14%
4) select patch that includes both MVintageRotary and MMHarmonizer. CPU use 38%
5) go back to first piano patch. CPU use 35%!

This will never correct itself and, my guess is, it will ultimately cripple the CPU if I keep using Melda plugs in more patches. When I put up the CPU history graph, I can see why. Other plug ins remove themselves when the patch that includes them is no longer being used. Melda plug ins persist until I either turn them off or quit Mainstage and start over again. Once they're turned off, the piano is back to its normal 5% load. Melda plugs are the only ones in my system that have this problem. Obviously, this is no good for a live show, so I'm going to have to stop using them until this is sorted out.

Any ideas about why this might be happening? I love these things and don't want to have to find replacements.

BTW, this is on MS 3.2.2 with melda 9.15

waltercruz · Post by **waltercruz** » Sat Jan 16, 2016 7:46 pm

I use MVintageRotary with Mainstage in an old core 2 duo..... Have you seen Mainstage Patch Aliases?

MVintageRotary is ALWAYS processing, even without a input signal, and it's a somewhat demanding plugin. I use just one Hammond with MVintageRotary, and do some aliases of this patch.

steve2KVR · Post by **steve2KVR** » Sat Jan 16, 2016 8:59 pm

...curious...how does one "aliases of this patch?"...seems very useful.../s~

waltercruz · Post by **waltercruz** » Sat Jan 16, 2016 9:05 pm

https://brianli.com/using-aliases-reduc ... mainstage/

https://support.apple.com/kb/PH13561?vi ... cale=en_QA

I play with mainstage with a Pink Floyd cover band, so some patches have a lot of aliases (Hammond, Wurlitzer and some synths). During the concert, I only press a button on keyboard to change patches, a button that change the patch to the next on the list. I arrange my patches in the order of the songs in the show.

But, yes, MVintageRotary is somewhat CPU intensive, but I love this Leslie.

scoplunk · Post by **scoplunk** » Sat Jan 16, 2016 9:29 pm

Yes, I use aliases. But, your post got me to thinking about this some more and now that I've checked this out a little more closely, the problem may not be a melda problem. I'm beginning to think the problem is somehow related to the Logic B3. It doesn't unload and therefore keeps other plugs from the channel strip also instantiated? Hmm.

When I put MMHarmonizer on an electric piano channel strip, it unloaded correctly with a patch change. The same thing happened with MFreqShift. When I replaced EVB3 with VB3, the melda plug ins, including MVintageRotary unloaded correctly with a patch change.

It appears I jumped to the wrong conclusion. EVB3 is the problem, not the melda plugs. Sorry for the false alarm. But, I'm glad because that means I can use the melda plugs again. Great! Now, about that EVB3...

waltercruz · Post by **waltercruz** » Sat Jan 16, 2016 9:48 pm

Nice! I will test it later using Kontakt Vintage Organs to see if it solves my problem. I really miss my VB3 in a 64 bit environment.

In april, last year, I sent a e-mail to Yonac Software to ask if they had interest in porting their Galileo Organ to Mac Os X, and they said yes, but I haven't seem any progress on this one

scoplunk · Post by **scoplunk** » Sat Jan 16, 2016 10:48 pm

I feel the same way. VB3 is fantastic, but I have to run it with 32 Lives. It works pretty well, but I see a lot more spiking on the CPU, where everything is steady with EVB3. I can't help but think that a 64 bit version of VB3 would solve all of my problems. I'm not sure that's ever going to happen, though. The developer is working a lot with hardware these days, which I understand because it's difficult to steal hardware and you can charge a lot more money for it! But damn, I would sure love to have an updated version of VB3. In the meantime, I'm going to keep checking things out with EVB3. If I can't get it to behave better, I'll send a report to Apple.

waltercruz · Post by **waltercruz** » Sat Jan 16, 2016 11:01 pm

Hi scoplunk! I did a quick test here, and yes, the trouble is really VB3! Really shocked

indirectally, you solved a HUGE problem for me. I never realy tried Ni Vintage Organs inside Mainstage, cause I really like the VB3 sound, but using Vintage Organ solved the problem to me. My computer is somewhat old (MacBook core 2 duo, mid 2010), and MVintageRotary uses 15~17% of my CPU. Having that percentage of my CPU wasted, without even using the Organ Patch was driving me nuts, for something like a year or more. I even wrote a e-mail to the Melda developer, and he said that Mainstage was a BAD host. Well, looks like he was right... or at least partially right: maybe Mainstage is not a bad host, but VB3 is a bad boy.

I will migrate my organ patches to NI Vintage Organs.

Shall we both open a ticket on Apple Side? It's just two drops on a freaking huge ocean, but maybe it can make a positive difference

Someone, please: write a Hammond simulation (not samples!) that can run on 64 bit audio unit hosts.

scoplunk · Post by **scoplunk** » Sun Jan 17, 2016 12:14 am

If you're seeing the same thing over there, then we should definitely tell Apple about it. I can't imagine that this is the planned behavior. It makes no sense to be wasting CPU cycles on a virtual instrument and everything else on the channel strip even when it isn't active.

Aargh! I hate wasting so much time chasing bugs when I want to be working. Now, I've got some weird audio crackling that starts up after about 15 minutes and gets worse and worse as time goes by when I use Mainstage 3.2.2. The very same concert in 3.1.1 works flawlessly. Oh well, I'll get back to work now and chase that problem down later.

Thanks for checking this out waltercruz!

MeldaProduction · Post by **MeldaProduction** » Sun Jan 17, 2016 1:25 pm

Thank you folks for the positive comments!

As for releasing the effects - basically AU is really a messed up thing. Instead of the host asking each effect separately, the idea is that the host connects them, so they do the work (complete madness). Most 3rd party hosts don't do that fortunately, but Logic & Mainstage do from I remember. That opens a whole set of potential problems, because if any of the items in the chain does something weird, it will all fall apart.
Now... if the plugin is still processing, that means, that something is sending data to it, it has no other way than obey the command. So indeed it is the plugin before it or Mainstage.

As for Mainstage, I'd strongly recommend using something else, just anything that would make the workflow good enough. It is the least efficient host in the whole universe, it doesn't even use multicore from what I remember (maybe the newest version does, hard to say), so the huge numbers you were presenting are indeed like that because of Mainstage... I don't understand how they can sell this as a host for Live performances...

waltercruz · Post by **waltercruz** » Sun Jan 17, 2016 1:35 pm

Reasons for using mainstage: it has a very streamlined workflow for playing live. Really. And a huge collection of real good instruments. For other projects where I play loops, I find the workflow o Ableton better, but for the simpler workflow, Mainstage wins. And it uses multicore

scoplunk · Post by **scoplunk** » Sun Jan 17, 2016 7:29 pm

Mainstage has its quirks, just like every host. It's definitely multicore. In fact, that's one of the things that's different than Logic. When Logic is in live mode, which is what they call it when you're playing a virtual instrument through a channel strip, only one core is being used. This drives everyone with a modern machine crazy, because if you make a complicated signal path, it will choke a single core while all the others are sitting idle. At least this isn't a problem with Mainstage.

Thanks for the explanation of how AU deals with plug in chains. That explains a few things that I've seen happen. Sometimes, a bunch of plugs will still be taking up CPU even after a patch change, but they release later - probably because some signal is still present on the previous patch. Most of the time, this isn't a problem, but I've had third party plug ins that don't exit gracefully and a small glitch or dropout can occur. Oh well, this seems to be the price we pay for dealing with software from all sorts of different developers. OTOH, the amazing stuff we get in return is been worth it, so far.

MeldaProduction · Post by **MeldaProduction** » Mon Jan 18, 2016 5:56 pm

Got it folks

. Didn't know about multicore and definitely didn't know about Logic choking up that much!

It's not exactly the best host there, but I thought it at least is pretty stable. Well, at least I never saw this "zombie effect" problem there

.

waltercruz · Post by **waltercruz** » Wed Jan 20, 2016 6:35 pm

Hey scoplunk, Apple released a update to Mainstage that appears to have solved the trouble here (I need to test it more and load my whole Pink Floyd concert, but I will do it later). Maybe it can help you to solve your issues there.

scoplunk · Post by **scoplunk** » Wed Jan 20, 2016 8:33 pm

Awesome! I haven't had a chance to update my Mainstage machine yet, but I updated Logic on my other computer and it appears that they've finally gotten rid of the "one core for live mode" situation I described earlier in this thread. I'm throwing up different softsynths and watching the CPU meter and I can see multiple cores working where it used to be only one. I don't know how much better everything will work in everyday use, but this is a huge improvement.

OK, I've got a lot of other things to do today, but I'll check in later when I get a chance to look at the MS update.

MeldaProduction

KVR Audio

melda plug in problems in mainstage