Which DAWs have dual buffers?

Cubase’s “Asio direct monitoring” for Pc isn’t even a dual buffer, or low buffer really. It’s just highly integrated communication with your audio interface’s drivers. Essentially it takes control of the Interface’s mixer and controls it with the faders inside cubase. Completely bypassing any buffers in the DAW. So you’re hearing the sound straight from the converters, but you don’t need to use a secondary mixer (your interface’s software mixer) which makes it seem totally integrated.

I don't know why it's confusing. Help me help you. 

This "understanding" is  flawed. You are conflating the disabling of processing with dual buffers. Dual buffering...or triple buffering as Cubase uses now...is a function of the audio engine. It's nothing you engage ad hoc.

Cubase's "Constrain Delay Compensation"=Logic's "Low Latency mode"=protools "low latency monitoring"...Steinberg was the only one who called it what it was--as confusing as the name IS...they all do roughly the same thing, which is to more or less fuck up what's going on right now in service of a temporarily lower latency input. IGNORE THIS FEATURE EXISTS FOREVER.

#geritolShot #history The multiple buffering, starts with Nuendo 20+ years ago. Even though I was a Logic user longer ago--I honestly never did audio production with it....it was simply a MIDI sequencer that had some software synths/sampler in it functionally to me. I used embedded hardware for audio. Certainly, they DID at some point implement double buffering. It's handled the OPPOSITE way in Logic--you select 256, that is ONLY your input buffer. The process buffer is some multiple of THAT. Which is why people think Logic is "more CPU efficient" than Cubase. If you selected 256 in Cubase, you got a 64 input buffer  and it ran the WHOLE project on 256. The "equivalent" was to set Cubase to 1024(256 input buffer)...and compare what it could do compared to Logic THERE--equally input and process buffer latencies. Anyway--these two led the way, and were the ONLY apps with it until Studio One, which was created by ex Nuendo team members. 

Keep in mind--ProTools TDM, standard since the late 90s...is an embedded hardware system. Nuendo's developer challenge was to allow for music production IN a latent (native) environment. With HDX, they shifted to a native playback engine, hardware cue mixer/DSP...effectively making a super fancy version of Nuendo circa 1997. It's what UA does with the Apollo now--with channels automatically switching from running on the DSP mixer or the CPU mixer based on input monitor state. Technically, Steinberg/Yamaha and Presonus have their own, much lesser versions of that and have for years--NO ONE buys first party solutions...for some solid reasons. But, the side effect is that there's only so much a company can do to work around monitoring latency--since first part is the only way they have control over the hardware mixer in every interface. 

What's the debate about Cubase not having multiple buffers on Mac? It behaves the same. When it had two it had two on Mac. It's been a while since I've had a new Mac...but, I feel like Greg still does all his demonstration videos on his Mac...since every now and again he makes reference to a keystroke being "ALT on Windows" instead of Command...

Yep AAX DSP is totally dead and the best are still the old Sonnox Oxford plugins which are very expensive because they charge double than for the native. The Pro Audio DSP DSM too is great. Like I could mix something down with the Oxford and use the limiter on two bus and the reverb but if you need some ass saving magic like Drum Leveler or RX, distortion like Decapitator or Tupe, mega powerful dynamics like kotelnikov or the Weiss ds1, good amp sims or vstis, you have to leave the aax dsp and deal with more latency than going all native 😬

I don't know anything about "Constrain Delay Compensation" (CDC) in Cubase or other versions thereof in other DAWs, but, just to be clear, this is something different than dual or multiple buffers, correct? Just making sure we're on the same page. Also, why are you saying to avoid CDC or other versions thereof in other DAWs?  What is it about CDC that you don't like?

Also, I think Veggieryan DOES understand the difference between disabling plugins and dual buffers. He's just simply saying that S1 will disable certain latent plugins IN ADDITION TO using the lower/faster record path buffer, as an additional measure of reducing latency.

In any case, as I mentioned in my previous reply to you, dual buffers is still what I started this thread about, so I'm trying not to get too far off in the weeds with other subjects like CDC.  I'm really just trying to determine which DAWs have dual or multiple buffers and which DAW's implementation of dual/multiple buffers is done the best.  Thus far, S1 seems to have dual buffers which do what I want, but I'm not sure about any other DAWs. Are you saying that Cubase also has dual or multiple buffers on Mac? If so, how does it compare to the dual buffering in S1. What about Logic?

Good to know. Yeah, I was aware of the ASIO Guard thing, and was similarly confused by it, as there is no ASIO on Mac. Also, is ASIO Guard different than ASIO Direct Monitoring? That could further muddy the waters if you're not clear what each does.

In any case, it sounds like you've verified that dual buffers exists for Cubase on Mac. Good news. That's what this thread was about, and it's good to get some confirmation. I'll be curious to hear how it compares to dual buffers on S1.

Thats not what quint is asking about. "asio direct monitoring" and "constrain delay compensation" are separate to the dual buffer/hybrid buffer (ASIO-Guard).

Cubase dual buffer/hybrid buffer is ASIO-Guard which is available for Mac and Windows. (yes ASIO is a windows thing so its confusing naming.)

Cubase is pretty robust, set and forget on any modern computer.

Please don't take this the wrong way, but your posts are a little hard to decipher sometimes.  They're a little "stream of consciousness", if you know what I mean. 😜  So there are some things that are hard to follow, because the context isn't clear, so I need to seek some clarification sometimes, to double check that I'm understanding you correctly.  But I do appreciate the help, nonetheless.

In any case, just to make sure that we're on the same page here, are you including dual or multi buffers when you say "low latency modes"?  I just want to make sure I understand what you are or aren't saying to avoid.  In other words, to put it simply, are dual or multi buffers "bad" across the board, or just in SOME DAWs?  If bad in only some DAWs, which ones?

I'm with you on wanting really low or zero latency. This is why I've been on Apollos for a long while. DSP monitoring may not be zero latency, but at 96k, I was getting 1.1 ms RTL on the Apollo, unless I used certain plugins, which would then kick the latency up to a total RTL of about 2 ms. So it wasn't zero, but it was as good as I could get without going all analog monitoring (like you), and I don't know that I'm prepared to make a move to switch to all analog monitoring, at least not right now. On a related note, this is one of the reasons why I always work at 96k. The latency is lower.

So I'm exploring A) if it's possible to achieve similar levels of latency in an all native system and B) what the best ways are to achieve that same level of low latency in an all native system.

Dual buffers seems to be the best alternative I've seen. If I could reliably/stably do 2 ms RTL or less, while monitoring thru plugins, I'd consider dual buffers to be an equal and viable alternative to DSP monitoring. Based on what Gravesnumber has said in this thread, I'm optimistic that this is doable, at least with S1.

I guess the most confusing thing for me here is that how can a "dual buffer" somehow override the processing overhead needed for plugins? If a plug needs 1ms to process a block of data, it needs that on DSP or in CPU alike. The DSP advantage was always that DSP enabled plugs always processed in a deterministic amount of time, and you could parallel them up and have no further increase in processing time. But that was back when CPUs were in the MHz range and now that they're in the GHz range, they can process data 10-100x faster than most DSP chips can based on clock cycles alone and thus why DSP use in audio has been largely discarded.

The only drawback is that CPU processing is nonlinear and non-deterministic and there is an amount of overhead needed for the OS and the DAW.

But back to the plugin processing.. Buffer settings adjust the amount of samples being processed ahead, at the expense of CPU demand.

But how would a "dual buffer" change that amount of processing required?

Lowering a buffer would always increase the demand on the CPU, and vice-versa. Pre-rendering effects would be an effective strategy but it ALSO takes the exact same processing time, so doing that on-the-fly would not gain a single millisecond.

The ONLY way I can see this working is if the buffer were automatically reduced and plugins over a certain reported latency were disabled.

In Reaper, I simply change my samplerate as I need and sometimes disable certain plugs that report (in Reaper's real-time Performance Report) that have high latency or CPU usage. I haven't seen anyone actually explain to me what these other "dual buffer" DAWs are doing that is different than that.

Ok, so it explains it like this:

Threading.  It creates multiple processing threads in parallel.  Other than "VIs in one thread and audio inputs in another" it does not explain this in any further detail but I guess it does explain the "dual" aspect of the dual buffer.  But what if I don't use VIs?

It disables any plug that is more than 3ms in latency.

It disables external round-trips, analyzers and splits(I assume it means sends/returns).

The picture they use to show "Low latency" monitoring shows 6.6ms of monitoring latency at 512 samples.  I get similar numbers in Reaper with my current setup, but I think the MOTU driver has a lot to do with that.

Ok.  I went by this statement: "Under this system, the tasks of audio playback and monitoring of audio inputs and virtual instruments are handled as separate processes."  which makes it seem like it's only those two processes that are separate, but I'm sure it's more of a layman's explanation than I'm reading into it.

And yes, I turn off high latency plugs, just manually.  Last night we did vocal overdubs and I left most plugs on since I'm already working on the meat of the mixes before the band wanted more harmonies added..  I even moved the buffers from 128 to 256 and noticed almost no difference with my highest DPC at about 2000.  I'm not sure if something else is the gating factor here but the main singer and the background singers didn't mention anything about the headphone mix having latency.  

I need to do a test and see if my headphone sends set to "Pre-FX" have less latency than normal (post-FX).  I do like using the DAW to direct tracks to the headphone sends since I can enable and disable at will and do adjustments on send levels, etc.  It makes it much more flexible than using pure hardware for monitoring.

I have some theories on what I can do in Cubase to assign open/close an individual insert to an individual button on my SD.

I'm curious, have you already figured out how to do this on your Stream Deck, or does the Sideshow app you're using for SD already have that capability directly built into the app?

Also, which version of Cubase should I be looking at for purchase?