Hello folks -

New Mac time over here. I’m also considering building a PC to act as a satellite host for VEPro.

With a pretty large template on the PC, and a new Mac Pro Ultra or Studio Ultra, I’m hoping to have a minimum of latency between the two machines. So with that said - and since it’s kind of soon to ask how those Macs would do, I’ll ask about the M1-based Studio Ultra.

What kind of multiplier is needed when using VEPro on a newer (Gen 13 Intel or Ryzen 9) PC? The PC’s libraries would be on nvme drives on an internal bus. The Mac would have a pretty large template too, and set to as low as the machine would tolerate; I have been hearing some say that it never goes above 128, but I don’t think those are 300-instrument-track projects.

Any success stories or miserable failures?

I don’t think that everyone who does this professionally wants a free solution that’s feature-poor and doesn’t have a company behind it with the resources to fix things. I like little developers but I don’t want my whole rig to be waiting on someone sitting at a computer in their basement who isn’t paid to support the app. The idea of endless free software is a dead-end one to me - hoping something works, building a rig around it, and then when the dev has to go work for a pharmaceutical company, having to change platforms entirely. But at least it was free, so I saved - what: $300US for two licenses? And lost days and days of work? That’s the definition of false economy.

If anyone is running a Mac Studio Ultra and a recent or new PC for external VEPro hosting, what kind of buffer sizes (and buffer multipliers in VEPro) do you need when working with a large template? Is it possible to keep the Mac at 128 or 256 if the PC is doing a lot of the work, and how small can the multiplier be?

I mean, by the way, if the PC is playing lots of stuff - not just Superior Drummer or something, but full strings/brass/winds. Full sections, lots of articulations.

Curious also about the handling of cores on 13k PC’s vs. AMD Ryzen 9’s - I’m really skeptical of the idea of half as many performance cores as efficiency cores.

Do go on.

I’m not interested in internet conflict over software releases. I’m actually not engaging in conjecture beyond assuming that it’s complicated to do and that they intend to stay in business. You can argue those points on your own, as much as you like. Won’t make anything happen faster, hence my point about reality. Sure will be great when it’s all updated.

It's my understanding that the reason VEPro works is low-level code, directly addressing hardware - which when it was originally written was Intel.  Now that Apple Silicon is here, and it's a completely different hardware platform and architecture, it's got to be like writing the whole thing from zero.  Maybe the M-series CPU's don't deal with these direct calls as well.  That's all beyond me.  But I think of this like a team of developers writing an entirely new app that has the goal of enterprise-level reliability and the greatest efficiency and speed possible.  How long would one say that should take?  Especially if they have other things that people are clamoring for.

Some folks are acting like the folks at Vienna don't care about users' feelings about this.  Of course they do.  But they can't do anything about those big feelings except put out the best product they can, and for our part we can maybe not act like our feelings trump reality.

Thanks for your reply!



That’s MIR’s buffer multiplier, correct? If my buffer is set low-ish to start - say, 64 but ASIOGuard bumping it up as it does - can I set MIR 3D to a specific size or is it a multiple of the DAW buffer? I ask because I wonder if MIR would shift its value when I put a track in record.
EDIT: I see the answer in your picture.


Cubase can always be on a lower size on the record track, so I shouldn’t have to change that. But by “deactivate” you mean “remove”, not just bypassing, yes? And so when I’m done playing on that track, I need to re-instantiate it on that channel and recall a preset? Hmm. Even doing that with track presets on a macro (“recall selected track’s preset with ‘MIR’ in the name” or something) seems pretty cumbersome when I’m in a hurry. I will have to think about that.



But if Dynamic Processing is on, then at least it wouldn’t be using the cycles for convolution etc., correct? Just sort of holding the door open, with the latency as well?



I will give it a go when the new computer shows up.



Very true, and very helpful. It’s my understanding that the reason why one has to be careful with bussing is that everything in a channel is going to be processed by the same core, so if there’s a lot of elaborate bussing it’s much harder to be efficient, and the system can get inordinately bogged down. But I’m a composer, not a software engineer. 😁

Your comments are greatly appreciated!

Hi folks -

As a composer I mix as as I go and I also perform a lot of things with modeled instruments and VSL libraries, so latency is always a concern. I am preparing to do a rather large upgrade to the rig, and considering adding MIR 3D when I do. (I’m currently using some other placement/ER software along with various reverbs for cinematic tail.) My impression is that MIR 3D is not specifically designed for this use case - which is not a weakness in the app, just not what it’s for. If this isn’t the case - that it functions at very low latency sufficient for real-time performance - I would love to hear about it.

Anyone using a Mac Studio Ultra on large templates not 25 tracks, but hundreds - and using MIR 3D in Cubase? And how much latency are you experiencing on new VI tracks at Orchestra-playback track counts?

Any special bussing setups that optimize DSP usage? My thought was that I would for example arrange string sections from various libraries such that each one might get a little pre-processing and then they would all feed a “First Violins” bus into MIR. Feasible? Inefficient? Would that play hell on ASIOGuard?

And beyond that - if I set my template up with MIR and then bypass it per channel, will its additional buffer latency go away? Is it still using DSP? (So I could bypass it for a performance that requires minimal latency and then re-enable it.)be, but mir is its own weird animal. wondering how cubase handles this. also wondering if mir will allocate dsp of a single instance of its use per channel in the way that other plugins do - that its dsp stays on the same core the channel is using - or if because of its buffering scheme it uses another set of cores that all instances send to.

finally, wondering if an apple silicon native version of this is imminent.

So if I get the Synchron-ized Chamber Strings for example - I know that an IR was used to place them in the hall; but what control do I have over that? Can I separate the “close” (original) perspective from the IR hall or is there just a balance control or what? Or is it a multi-out IR that simulates some other perspectives?

Seems worth doing if I can also have the same perspective as the VI version with the revised samples and the player.

Hello All -

I'm having this very issue.  I thought I'd localized it to using a local instance of VEPro in Cubase, but this has turned out not to be the case.  It does go away when I disable ASIOGuard on the plugin.  I think their dual buffer scheme is responsible, as it seems to be that switching between guarded and unguarded during play is what makes this happen.

When you are looking for a device to contain all of these drives, take care to read the specs. There are enclosures that will handle multiple drives but have limited data throughput for individual drives, which is what I believe you were trying to set up. Many multi-drive enclosures will give you ~1300MB/sec reads with a RAID but more like 200 with platter drives and 350 with SSD’s. And it’s not an issue with thunderbolt bandwidth at all.

This enclosure will do ~1500 MB/sec in JBOD mode. https://eshop.macsales.c.../TB3QMJB000/?fullDetails

I haven’t had much luck using USB drives for heavy streaming of libraries, and definitely not conventional platters. Though I did some years ago when I was touring and had to use FireWire drives, but it was basic VSL VI’s and not the de rigeur multichannel monsters we know and love today.

Honestly - if money allows, to me the coolest thing so far has been using nvme drives in an expansion chassis (as my 2013 MP has no slots) on PCIe cards. Just not a question of speed anymore at all.

Hi Paul - 

What's the deal with turning ASIO Guard off?  If Cubase compensates for that correctly should it be an issue?  Is the same directive in place for Studio One which has its own dual-buffer system?

Thanks-

Richard

Hi Steve -

I'm asking if anyone has compared using VEPro in server mode as the streaming IO with using it in standalone mode with, say, Network MIDI or MOL and using Dante as the IO.

Anybody?

Hello folks -

Doing some research into new interfaces. Lots of my favorite candidates now use Ethernet to connect to the host - Focusrite Red stuff via Dante, Avid Carbon via AVB - and I’m wondering if VEPro can coexist with Dante on a PoE switch network. (Carbon doesn’t work with a switch, so barring use of a Thunderbolt-to-Ethernet cable, it gets its own port.)

I realize that the Mac I use has two Ethernet ports, and that I could also use the Thunderbolt connection or the Avid HDX connection on the Focusrite, but I’m trying to figure out what’s possible. At the moment I’m enjoying the utility of having the second Ethernet port handling the wired internet network. I suppose the Ethernet adapter for TB could handle that as well.

I’ve heard some folks say that there is a latency improvement when using Dante with a Rednet card and network MIDI to control VEPro in standalone mode. Wondering if anyone here has experienced this. I know that you lose faster than real-time bouncing and recall, but most times these things are possible to live without.

One reason the Focusrite stuff is on the table is that it can function as a ProTools IO but also as a native DAW IO - in my case predominantly Cubase - and I wouldn’t be paying for DSP in the Carbon that I couldn’t use unless I was running ProTools. The UA is a little less, and maybe doesn’t sound as good, but lots of DSP is available - though their plugin ecosystem is fairly expensive, and many of their plugs are not suited for a real-time sequencing environment.

So I’m chewing on these points and wondering if others here have tried any of these things.

In a related area - can I run Catalina on my main Mac and Mojave on my satellite Mac? Any VEPro 7 issues there?

I’ll respect your wishes by not mentioning the name - though I think that the courts of the world would be clogged if someone got sued every time - or even one percent of the time - for complaining on the Internet - and also I’ll refer you to this thread where what you wish for was being talked about - I have long been a proponent of VEPro doing a lot more than it does now for MIDI processing. Here’s hoping we all get what we want.

The thread

Regarding price - I think if it this way: if a tool or sound library makes things faster for me, it will pay for itself rapidly. I work all the time (a very fortunate thing) and the music business has declined in rates generally (an unfortunate thing), so whatever makes it possible for me to do more with my time is desirable. I hazard a guess as to what my hourly rate would be based on how much I need to work to hit a monthly mark, project how many hours it would save me per month to use the product, and if it makes sense and if the product’s price is comparable to others (if any), that decides it.

Hello folks -

It’s coming time to build a new PC for remote VEPro duties. So I’m wondering what things I can do to address large buffer sizes in the remote machines I use. My main machine is a 2013 Mac Pro 12- Core with 128 GB of RAM and ProTools HDX with an HD|IO and the satellite is currently a 2010 Mac Pro 3.7Ghz 12-Core with 64 GB RAM. The remote machine is handling many things, most notably the SampleModeling brass which are pretty demanding CPU-wise, but also the OT Metropolis Arks and several percussion libraries (it’s also where I’ve “quarantined” the EW Play player). The main machine is doing various libraries including my VSL winds and strings, and also AudioModeling winds and SampleModeling strings, Superior Drummer 3, Damage 2 and a host of other Kontakt instruments, as well as Altiverb and a few other verbs for tail. I use a local VEPro to host the instruments, and the Reverb’s are all in Pro Tools. I sketch with most instruments turned off but a piano, and I can do this at a low buffer size, but when it’s all up and in use, I have to - have to - use a 1024 buffer to avoid the usual pops etc. (I’m working at 48/24.) In particular, the brass instances on the remote machine (an instance per section, and a Kontakt instance per chair) require 3 buffers in the instance to play back without artifacts. Everything in the satellite Mac is streaming off of nvme drives in a Sonnet 4x4 PCIe card, and throughput and latency are excellent. WiFi is off. Gigabit Ethernet connections through a Netgear switch. Cat6 cabling. Short runs. My instances of VEPro are on instrument tracks. I also use aux inputs for multi-out returns from VEPro.

So I can manage the delay after the fact very reliably - compensation has been fine; but obviously it’s no pleasure to play through. What I would like is a minimum of latency for real-time performance. Where can I look to start improving this? If I did less on the main computer and more on a very fast remote computer or computers, or if I did it all locally on a much faster computer? I realize that the newest tech in there is seven years old (though I bought the 2013 Mac in 2015 new). I’m just trying to figure out where the logjams are so a) I don’t configure my machines poorly and b) so I don’t spend a ton just to discover that I’ve gained 5ms of latency performance.

Also, I’m doing some transitioning for certain kinds of work into Cubase. I still use VEPro on the remote Mac but have found that for my workflow needs, using Cubase to directly host plugins works a little better.

So, VSL geniuses, what’s the path to awesome here? What aspects of a system govern buffer sizes the most - CPU speed, drive latency and throughput, other? How do I get a system to allow me to play a remote DSP-hungry instrument without having an eighth-note delay?

Thanks!

When the app you are referring to came out, there was nothing that did what it did.  Still isn't anything that does that.  I would certainly love to see VEPro do this, and in doing so become the object-oriented library-dominator it should be.  (See my recent post in the Ensemble software section.)

That said, I'm curious as to what you are basing your statements of inflated price on.  I mean, I know how much it is - I bought it.  It's not a feature of a mammoth app like Cubase or something - it's a specific app by a small developer that does a certain thing that some people find useful.  (I wish it were more like TransMIDIfier in terms of transforming the data as it's split, and I think they are working on that.) I don't resent them, or have a bad taste in my mouth or any of that.  It's a very iffy thing to decide what someone else deserves for what they do, especially when nothing else does it, and there's not much point in being bitter about it, if that's what you are doing.  Price doesn't require a warning - it is what it is.  Any consumer can decide if the feature set is worth it to them, and it's not the developer's fault if the expense isn't justified.  It was for me.  When I use it it's great.

Also, to put a finer point on it -

I’d like to see a divisi ensemble maker mode, where one could either:

assign targets defined by their size (which would then combine or split based on divisi rules), that could be told what the largest ensemble you want is and the largest one you have - say, if you had a solo viola and a four-viola ensemble, it could auto-create another instance of the four-violas to make 8, with randomization applied. It could be aware of what VSL libraries you have, for example, and make those automatically.

Or, one could build ensembles using solo instruments. Obviously this would work best with modeled instruments, but would be great for extending their functionality.

It could also engage layers as desired.

My goal is to be able to create a sketch on a piano track, then send some of it to a track that would output something useful as multi instruments. To write a four-voice thing, send it to “horns” and have it played back by four horns, and possibly using sustain pedal to indicate to VEPro that a line needed to be a4.

I also think this should happen because I’m not sure where they are going with this software but other than built-in early reflections placement software, I don’t know what will justify the next big version.

I agree. Because it would make one’s template (which naturally would include some setups using this tech) sequencer platform agnostic.

I can’t help but wonder if this is an issue with MIDI coding and timing that would not work outside of real-time - that if one were doing a non-real-time bounce that somehow that makes it too hard for them to do.

Yes, and I have many scripts that kind of do those things, but not flexibly enough and not outside that instance of Kontakt - so not that flexible. Rather than make a limited transformer that lives in the Synchron or VI player, why not put it where it does the most good?

Hi folks -

So what I would love to see in VEPro is the ability to take a MIDI input and apply modifications or rules to it before sending it to any instrument loaded.  To be able to load up sounds in VEPro and then be able to do things like  

CC-based articulation switching, even across libraries.

conditional layering (e.g. adding chamber staccatos to other libraries above a certain velocity) with key ranges.

constrained randomization of parameters including timing. pitch and cc values.

divisi logic that could handle voice-splitting and also switch articulations as section sizes changed - scalable from solo to chamber to standard to appasionatas, for example.  That could be set so that one had normal access to one's instrument targets, but also could have an input for chordal/polyphonic playing that would address these same targets via the divisi logic - with an option to capture the resulting splits as MIDI data to be drag/dropped into the daw.

library-agnostic combinations with modifiers to either unify the response of different libraries or cause one in a combination to respond on a curve - say, if I were using Synchron Strings Pro but wanted to make the orchestra even larger by layering it with some other library, but only as the CC dynamics approached 110 and thus scaling up the other library's dynamics at that point.

being able to easily combine sounds from different percussion libraries on the same incoming channel while still maintaining the sounds' separate mix settings.  So a master orch perf patch with elements from all over, balanced and reverb-matched, addressed with a single MIDI channel.

Yes, I know many of these things can be done within a sequencer - especially one like Cubase.  And I have Divisimate and TransMIDIfier.  But I'd like to have as simple of a master sequence as possible, with as few tracks as possible.  And I figure it could be the job of the plugin host to do useful things to whatever the user sends it.  And I feel like it could be very elegant to do it that way.  I love VEPro, and I want more from it.

Hello folks -

I’m considering AMD for the first time. Looking at building a machine that is made to run a lot of hungry VI plugins - modeled instruments primarily - as well as the traditional VSL and Kontakt instruments. If you have a 20+core AMD, can you contrast it to other setups you have or have had?

Also, though theoretical answers are what make the internet go ‘round, please answer only if you have one of these - I’m looking for first-hand experience with a cpu handling a large template for ten hours a day. I know what the theory is - I’d like to hear about it in practice. And we don’t need to bring price into it.

Thanks -

Richard