Jim Roseberry

That sure sounds like a MIDI driver issue to me.

Are you running the latest driver for the Babyface Pro?

Always make sure to have an up-to-date backup image file of the OS (prior to applying any major updates).

If you're under tight deadline or working on a major project, wait for a period of downtime to apply updates/etc.

Most times, there won't be an issue... but there's no reason to take chances.

9 minutes ago, abacab said:

Dude, with all respect, you are just 100% wrong about this.  I have been using Windows 10 Pro for 2.5 years, and Windows never updates until I ask it to check for updates.

There is a policy setting in "Group Policy Editor" that lets you completely 100% disable automatic updating, as well as the notifications.   I have never in 2.5 years had a Windows update that ran without me first clicking on "check for updates". Of course, I always take a full Macrium Reflect image before I push that button.

Group Policy Editor > Local Computer Policy > Computer Configuration > Administrative Templates > Windows Components > All Settings > Configure Automatic Updates

The text highlighted in blue in the image above explains how the policy works if it is set to "disabled".   I am quoting it here for clarity because it seems that embedded screen captures on this forum have a max width of 400 pixels. Too small to read.

 

 

 

If you're running Win10 Pro, you can absolutely disable all automatic updates including notifications.

You can do so via the Group Policy Editor... or by adding two Registry Entries.

Once implemented, you have to manually check, download/install all Win10 updates.

We've done this scores of times. 

3 hours ago, Leadfoot said:

You definitely want to compensate on the low frequencies so that it compliments the kick. I also like to add just a touch of overdrive to give the bass some grit so it stands out a little more in the mix. I record through a POD HD Pro into an Audient interface to get my sound, but any saturation/distortion plug in can get you the extra bite. Just be very light with it so it's not too distorted.

+1

Another thing you can do is apply the distortion/dirt only to the mids/highs of the bass signal.

• Mult or duplicated the bass track
• Run a high-pass filter on the mult/copy
• Apply the distortion after the high-pass filter

This gives you a tone in the realm of Chris Squire and Geddy Lee.

You get that nice sounding distortion... without the bottom-end losing definition

3 hours ago, abacab said:

You can simply install or upgrade to Windows 10 Pro.  Then you can either turn off or have complete control over updates via "Group Policy Editor".

Yep... you can completely disable all automatic updates (including notifications).

We do this for virtually all of our clients.  Those in commercial studios don't want to be hit-up with update prompts while clients are mid-session.

Babyface is an excellent performer.

When I was running a Fireface UFX (same core components), I rarely bumped it above the 48-sample ASIO buffer size.

6 minutes ago, John Nelson said:

Jim,

Would that be an option for a Asus Prime Z370-P mb?  i7 8700K.

Hi John,

Unfortunately, no... the Prime Z370-P doesn't have a Thunderbolt-3 header (nor BIOS support) for the Thunderbolt EX3.

The Prime Z370-A  does have the Thunderbolt-3 header (and BIOS support) for the Thunderbolt EX3.

If you want Thunderbolt-3 as an option, you have to be careful with motherboard choice.  Only a handful have the necessary header and BIOS support for the AICs

3 hours ago, ien said:

Do you have any experience of a Quantum interface being used on a motherboard through an add on controller, as opposed to the board natively supporting it? I ask as my board doesn't support it natively but has the add on option and i think i read presonus say that the board needs to support it natively (although i didn't find the remark when just looking for it. I don't think i imagined it though).

The Quantum works just fine with Thunderbolt-3 AICs (add-in-controllers).

I'm using the Asus Thunderbolt EX3.

2 hours ago, Michael Fogarty said:

well, that was pretty straight coming from you Jim. But Steev, you must realize. There are professionals  on this forum whose main thrust is to keep their machines maintained and running smoothly because they are busy producers. I am one of them, and Jim makes hi - end (I don’t use the term lightly) DAW’s. You are pushing Scarletts and other gear down our throat and we are living in the MOTU/Studio 192/Apogee/ Quantum/RME world. There is a reason they cost more - the same reason that we all shell out $1000’s for the Synthogy Pianos, Omnisphere and Keyscape Plugs, UAD and Waves Plugs and know which company makes our converters. There is no pride in this. It’s just who we are and what our clients expect from us. I think Jim has shown an amazing amount of patience and restraint up to this point. 

Now Jim, the Asian rep of Presonus gave me a Studio 192/DP88 combo and Studio One to make some music on and post. He said the 192 may have problems with Sonar/Cakelab/Cubase and other DAW other than Studio One latency wise.

I have found that to be the case.

But Quantum is different?

Hi Michael,

Quantum works well with Studio One 4.1.2 (especially with their Hybrid Buffering scheme), but it also works well with all other native DAWs.

With Cubase 10, I can run Helix Native at 96k with a 32-sample ASIO buffer size.  I can do the same in Reaper, Samplitude Pro X4, and CbB.

BTW, If anyone has noticed, when monitoring at ultra low latency, that's a scenario that's not particularly well suited to using multiple cores.

This is why high CPU clock-speed is absolutely paramount.  I'm currently running a 9900k (8 cores, 16 processing threads all locked at 5GHz).

As I mentioned before, prior to Quantum, I was not a huge Presonus audio interface fan.  

When a company "gets it right", they deserve kudos for a great product.  Quantum is a great ultra low-latency performer.

Is Quantum the best fidelity money can buy?  No.  You can buy better, but it's going to be significantly more expensive.

When it comes to ultra low-latency performance, you (currently) can't do better than Quantum.

As someone who's job it is to push the limits of ultra low-latency performance, I'm always on the lookout for great performers.

To me, it doesn't matter if it's RME, MOTU, Lynx, Presonus, UA, etc, I want a rock-solid audio interface that yields the lowest possible round-trip latency.

 

 

FWIW, It's a good thing the conversation is over, when it comes to PCs, you really don't know what you're talking about.

Semantics... 

Who cares if the Thunderbolt driver has to be loaded in Win10?!?!  Takes all of 1 minute.

As I said, Windows 10 does support "PCIe via Thunderbolt"... which is what allows PCIe level performance.

 

"And as far as you being able to plug into a Thunderbolt 3 controller card is the same as saying you plugged a USB 2 device into a USB 3 port. Focusrite sternly warns Scarlett users NOT to do that in their top 10 FAQ reports in Support to answer the common newbie question of "Why doesn't my Scarlett work?""

This makes me chuckle...  😎  (it's not the same by any stretch - let me explain the details)

There is not a single Thunderbolt-2 controller on the planet that'll work under Windows for PCIe level performance.

Microsoft doesn't support "PCIe via Thunderbolt" for Thunderbolt-2 controllers.  Period.  There's no debate/question about it.

Under Windows, to achieve PCIe level performance from Thunderbolt, you HAVE to be running Win10 and a Thunderbolt-3 controller.  There's no other choice.

Since most audio interfaces are Thunderbolt-2, you have to use a Thunderbolt-3 to Thunderbolt-2 adapter.

Regarding Focusrite's "warning" about connecting to Thunderbolt-3 ports:  That was posted while they were beta-testing their "PCIe via Thunderbolt" drivers for Windows.  If you're going to run a Clarette under Windows (and want to achieve PCIe level performance), you absolutely HAVE to connect via Thunderbolt-3.

There's a really cool Focusrite rep here in town (really nice guy).  He asked two years ago at Sweetwater's Gear Fest if I was running a Clarett.  I told him, no... and explained why.  They were still sorting issues with their Windows support for Thunderbolt-3.  Meanwhile, MOTU/UA/RME/Presonus had release drivers fully supporting "PCIe via Thunderbolt."

Prior to Win10, Microsoft did not support "PCIe via Thunderbolt".  Thus, any audio interface under Win7 that's connected via a Thunderbolt-2 is running Firewire protocol via Thunderbolt.  Any audio interface under Win10 that's connected to a Thunderbolt-2 controller is running Firewire protocol via Thunderbolt.  Firewire protocol via Thunderbolt does not yield PCIe level performance. 

 

If you know anything about current generation motherboards and Thunderbolt, you know that very few motherboards actually come with an integrated Thunderbolt-3 controller.  There are a handful of motherboards that offer Thunderbolt-3 as an option.  These motherboard have a Thunderbolt-3 header... and support in the BIOS for the add-in-card that IS the Thunderbolt-3 controller.  This has zero to do with the OS natively supporting Thunderbolt.  For these motherboards, there is no Thunderbolt-3 without the add-in-controller.  

 

Regarding Quantum and round-trip latency, you can run RTL Utility to confirm (with any audio interface).  I've already done so.

http://www.oblique-audio.com/free/rtlutility

If your audio interface reports its latency accurately, CbB and Reaper will list the total round-trip latency.

 

As I said, you can buy an audio interface with better fidelity than the Quantum...

But you won't be doing it for $1k.  😉

My DAW is running 24/7.  The Quantum never disconnects... ever.

 

Regarding Studio One, Presonus has implemented a "Hybrid Buffering" scheme.

• A small buffer is used for tracks that are being monitored in realtime.
• Tracks that are merely playing back are processed using a large buffer (much more CPU efficient - especially across multiple cores)

Hybrid Buffering has been used in Logic and Samplitude for years.   Avid implemented it in ProTools 11.

Though CbB doesn't use a Hybrid Buffering approach, Quantum still works very well at ultra low latency.

 

Regarding the Apollo:  Why would I miss it?  It's still here.  😉

What you refer to as "zero latency" monitoring thru UAD plugins is not correct.

Using UA's "Unison" technology, you can monitor in realtime thru UAD plugins with 2ms round-trip latency.

That's certainly very low... but it's not zero-latency. 

True zero-latency monitoring can't happen with DSP processing.  No matter how small, the audio has to be processed using a buffer.

That buffer results in latency.  Doesn't matter if you're talking a digital console, ProTools HDX, Apollo's UAD plugins, or using native processing of VST plugins.

This is exactly why the Clarett series (and MOTU AVB, RME, etc) has higher round-trip latency than the Quantum (the onboard DSP).

 

"By the time all prosumer computers are powerful enough to support 16 and 32 buffer settings we will have already switched over to "Dente" Ethernet interfaces as an affordable medium, the wave of the future, and much faster then anything USB  and or Thunderbolt 2, 3 , 5, 6, or 7, or what ever Mac has too offer in the future."

If you mean Dante, the lowest round-trip latency you can achieve with Dante (via your motherboard's Gb Ethernet) is about 10ms.  

The Focusrite RedNet Dante cards can achieve PCIe level performance (round-trip latency on par with Quantum)... because... they're PCIe cards.  😉

PCIe is the bus that used inside a computer (Mac or PC) that connects hardware and distributes data (liken it to the Circulatory and Nervous systems).

Thunderbolt is for all intents/purposes "external PCIe". 

You can't have a peripheral that supersedes (speed wise) the bus used to connect it.   

Dante' via Ethernet will NEVER be able to outperform a PCIe card or "external PCIe" aka Thunderbolt.

Mac for high-performance???  You can't even put a second drive in any current generation Mac.

Apple has all but abandoned their power-users.  We hear from them every week.  Moving to PC because they can't get the speed/configuration they need from an iMac Pro or an 8 year-old (cylinder) Mac Pro.

 

"Thunderbolt on a Windows PC? Not an established platform of reliability to say the least. Business as useable is more like it, or a crap shoot to be more precise.  Great for bragging rights when it works, but a bone chillingly expensive soul crushing nightmare when it doesn't."

Bone-chilling soul-crushing nightmare of a crap-shoot... ONLY if you don't know what you're doing.  

If you know all the details... and configure properly, Thunderbolt is absolutely rock-solid on a PC.

We've got professional clients who are running Thunderbolt audio interfaces in commercial studios.   If they were unreliable, I'd be the first to hear it.

 

Speed only kills if you're ignorant of the danger and the details.  Speed is not synonymous with "reckless".

A rock-solid DAW isn't the result of guess-work, happen-stance, or dumb luck.  It's the result of knowledge/experience.

If you leave important details to chance, yes... that can absolutely bite you in the rear.

• A good doctor doesn't leave important details to chance
• A good lawyer doesn't leave important details to chance
• A good reporter doesn't leave important details to chance
• A good athlete doesn't leave important details to chance
• A world-class performer doesn't leave important details to chance

 

FWIW, We've seen the Fast Startup option (when dual booting) cause loss of data files.

If you're dual-booting (even two copies of Win10), make sure it's disabled.

Cakewalk on a rental basis???

Do you have any MIDI interfaces or USB connected MIDI controllers?

This can happen if a driver doesn't let go. 

A good while back, this would happen if you had a Korg keyboard or MIDI controller.

I believe the workaround was to turn off the Keyboard/MIDI controller, then close Sonar (CbB).

I hear what you're saying about Win10 and what I refer to as the "annoying components"... but those things can be reined-in with the Pro version.

Once Cortana, OneDrive, all automatic updates including notifications, privacy settings, etc have been disabled... Win10 is a fine DAW platform.

FWIW, I feel like it's more of a "dumbing down" of the OS (trying to make it easy for the less tech-savvy) than a matter of control.  And in that, it's two-steps-forward... one-step-back.  On a computer, I rarely like anything "automatic".

3 hours ago, Steev said:

First off, the newly released Presonus Quantum 2 is a Thunderbolt "2" audio interface, and plugging it into a Thunderbolt 3 port isn't going to make it run any faster, it's just backwards compatible. But KILLER audio interface BTW although killer pricy. Also Windows computers nor does Microsoft don't natively or officially support Thunderbolt, although it can be ported through a USB C on computer motherboards that support it.

It also can run under 1 msec EFFECTIVE RECORDING LATENCY not total roundtrip, not even analog recording studios can do that, LoL, the human brain even doesn't react that fast comprehending the information it receives from your ears..

 Did you know that it takes more than 6 msec to hear the sound a guitar amp traveling through the air located on the other side of a medium size stage or any room from you that located approx. 10 feet away from you from speaker cab to ear??? Assuming an air temperature of 59 degrees Fahrenheit (15 degrees Celsius), sound travels at 761.2 mph (1,225 km/h), at 1 msec per foot, actually precisely 9/10th's of a msec per foot, it will take approx. 10 msec for noise the guitar amp makes to reach your ears..

 Are you trying to say a sound 10 feet away has too much latency to jam with???

 I nor anyone else have ever heard any noticeable artifacts such as time delay, flanging, or phasing issues through any decent audio interface until around 30 msec of latency occurs.

 But all these overkill specs and sales hype aside, can a Presonus Quantum 2 pass a double blind shootout test against a Focusrite for superior sound quality? I have absolutely no disrespect for Presonus, but...…….. That would be a very doubtful and extremely pleasant sounding first for Presonus, but really just comparing apples to oranges I suppose for the average users. Once we get used to and familiar enough to like the sound of our equipment, all is good

 I've been a power user of MIDI since the mid eighties, and playing and triggering Cakewalk Session Drummer 2 & 3 and XLN Addictive Drums 1 and 2 for years now from various MIDI synths, Yamahas, Casios, Rolands, Ensoneq, Roland Octopads, V Drums, and even very inexpensive Alesis Sample pads thru either the Focusrite (1x1), Behringer X Touch (2x2), a MIDI Sport (8x8) MIDI interfaces and I don't have any MIDI performance latency issues  with either hardware or software Synths or any  dollar value, as long as they can transmit MIDI, devices like the X Touch or any other MIDI controllers.

  MIDI latency is an entirely DIFFERENT ball of wax from AUDIO latency. Poor MIDI performance latency "typically" has absolutely nothing to do with audio latency.

 But old outdated MIDI controller drivers can have a bad effect on total system and overall DAW performance.

 A good case in point is, when I upgraded from Windows 7 to v10 my Edirol PCR 500 stopped working. Heartbroken, I first tried hacked drivers that got it working, but SONAR got sluggish, I did not notice any change or artifacts in audio latency, MIDI and GUI latency was increasingly more abysmal with every track added to the project.

 But then a light bulb can on in my head and I realized the PCR 500 is nothing more then a MIDI controller, it can still be manually mapped, which is a tad tedious, and by plugging the USB into a cell phone charger for power it works fine, but once done, it's done and it doesn't need to run MIDI messages thru USB or it's own drivers, it has a standard 5 Pin MIDI port, and by interfacing that way, BAM! Splat was snappier then it ever was before. I have to credit Win 10 for that.

 

And secondly, I'm talking about 12 msec round trip audio latency running a signal directly from a mic or line input into the Scarlett's preamps, then AD/DA converters_ out to and thru Cakewalk using Cakewalk's built in "Input Monitoring" to hear any inserted "AUDIO" FX plugins in real-time, then back into the Scarlett back out through it DA/AD converters into my headphones and or studio monitors with no humanly detectable latency issues with either recording one audio track or 16 with my older Focusrite Octopre Mk II running into into it thru ADAT with everything, including MIDI devices and Cakewalk running/ slaved to the Scarlett gen 2's clock source thru one single USB 2 cable. And the coolest thing about that is.... ALL computers support USB 2.

And being I don't have any problems, nor has anyone ever complained about working with 12 msec because they can't  tell the difference if we're working with 6 msec or 12 msec, I don't even bother cranking up the frequency from 48 k/Hz to 96 K/Hz for 6 msec latency. Being humans cannot hear the difference between 24/48 and 24/96, recording at 24/96 being MUCH LARGER .wav files is nothing more than a waste of hard drive space.

Who said the Presonus Quantum was a Thunderbolt-3 audio interface???

I said it was connected via Thunderbolt-3.  "PCIe via Thunderbolt" (necessary for PCIe level performance) is only supported via Thunderbolt-3 controllers (see below).

To clarify:

• Microsoft absolutely DOES support "PCIe via Thunderbolt", but only on Windows 10 and only with Thunderbolt-3 controllers.
• USB-C is the type of Port that's used to carry Thunderbolt
• USB-C can also be used to carry USB-3.1 Gen-2
• Just because a motherboard has a USB-C port/s does not mean it has Thunderbolt or supports it.  The motherboard has to have a Thunderbolt-3 controller onboard... or it has to have a Thunderbolt-3 header and BIOS support for a Thunderbolt-3 AIC (add-in-controller card).

I'm well aware of stage latency (when not using IEMs).  Yes, you can compensate/anticipate... but walk about 50-100 feet off stage into the crowd.   I play the better part of 50 shows a year locally.  And no, I don't monitor our guitar players sound from his amp that's 20' away.  It's mic'd and the sound sent to my IEMs or wedge (much lower latency).

I gave specific examples of how anyone can test and see/hear for themselves that lower latency is indeed noticeable and feels more immediate/tight. 

Take that 12+ms round-trip latency into a session with any seasoned player... and I guarantee they're going to hear it... and not like it.

Been there and done that in Nashville way back in the day (recording the Memphis Horns, Terry McMillan, Tabitha Fair, etc).  Even the vocalists noticed the latency... and complained about it.  The solution back then was to split the signal and monitor off the board... so the performers could hear exactly what they wanted/needed (tone/EFX wise)... and not have the latency impede their performance.

If you can't feel the lag of latency until it gets to 30ms, more power to you.  I guarantee that's not the case with many folks.  I talk to them every day.   A composer using virtual-instruments to score for TV/Film doesn't want to play thru 30ms latency.  When one of our clients has Slash over to overdub guitars, he doesn't want to cope with 30ms of latency while monitoring and trying to play tight.

 

Fidelity comparison of Quantum vs. an audio interface that's half the cost?

Which do you think will have lower noise-floor, more stable digital-clock, quieter mic preamps, more dynamic range on the A/D D/A?

BTW, You can buy an audio interface that has better fidelity than the Quantum... but you'll spend a lot more to get it.

The Quantum is (right now) one of the best low latency performers.  That's why I moved from an Apollo 8 Quad and Quad Expander (lowest round-trip latency is about 3.7ms).

 

As a "power-user", if you can't feel the lag of 12+ms latency (especially on high-transient material), I don't know what to tell you. 

It's as easy to loading your favorite piano sample library... and playing at each ASIO buffer size.  The difference is obvious and immediate.

I was not talking about MIDI jitter/slop.  That's another conversation.  As you know, MIDI is a serial protocol (not parallel) and doesn't lend itself well to ultra-tight timing.  Newer MPE controllers that spit out a lot of CC data can cause MIDI data stream congestion (causing timing issues).   One reason that folks like virtual-instruments; once the MIDI part is captured, playback timing is sample-accurate (consistent)... unlike dealing with hardware instruments.  I had a Korg O1/W FD years ago that had an onboard sequencer that was all but useless.  If you tried to record joystick movements, it would clog the MIDI stream causing major timing issues.  Not much of a workstation if you can't effectively record.  

 

There's a time/place for working at higher sample-rates.  It's not practical all the time. 

It's overly simplifying the situation to say that humans can't hear the difference when recording at 44.1k vs. 88.2k and higher.  It's not just about hearing higher frequencies.  For things like AmpSim plugins (dealing with non-linearity), you can have aliasing happening in the audible range.  Perform the distortion processing at high sample-rates, and the aliasing is above human hearing range.  Even when converted back down to 44.1k, there can be a clear audible improvement.  Some plugins up-sample... as well as some DAW applications (CbB has an up-sampling option).

 

BTW, I wasn't a huge Presonus audio interface fan... until they released Quantum.  Had always used RME or MOTU, then moved the UA Apollo to take advantage of their "Unison" technology (monitoring UAD plugins with 2ms round-trip latency), then moved to Quantum due to its incredible ultra low-latency performance.  Presonus has a rock-solid performer in Quantum.

I should mention that Quantum can actually go down to a 16-sample ASIO buffer size.  I didn't mention it previously because no current machine can effectively make use of it.  Still, I'd rather have the option and not (currently) be able to use it.  Eventually, machines will be fast enough... and 16-sample ASIO buffer size will be usable. 

Note:

• 64-sample ASIO buffer size at 44.1k = 1.5ms
• 32-sample ASIO buffer size at 44.1k = 0.75ms
• 16-sample ASIO buffer size at 44.1k = 0.375ms

Round-trip latency is the sum of the following:

• ASIO input buffer
• ASIO output buffer 
• The driver's (often hidden) safety-bufferer
• A/D D/A

When it comes to round-trip latency, the audio interface's safety-buffer is the X-factor.  The best audio interfaces can use a small safety-buffer... whereas other audio interface's have to use a large safety-buffer.  The smaller the buffer, the lower the latency.

 

If I'm overdubbing drums (and trying for tightest timing), I'm certainly not going to listen to a monitor that's placed 12 feet away (~12ms latency).  I'm going to use IEMs or headphones.

 

There's absolutely nothing wrong with your Focusrite audio interface.    There's also nothing wrong with acknowledging that there are other audio interfaces out there that are better performers for those who are concerned with ultra low latency.

 

You can migrate the OS install to the new machine... but it'll be a mess.

You're much better off starting with a clean install... and taking the time (when finished) to create an up-to-date backup image file.

This way, you're starting on a rock-solid clean/lean foundation... and you've got a backup of the clean install.

"Except by default, a 32-bit user mode process can only access 2GB."

It's a whopping 3GB with the Large Address Aware switch enabled.   😉

That takes us back a few years...  

 

 

Exactly.  

I use Drop-Box on an "as needed" basis.  

But I don't want it (or any other application) automatically trying to backup my files (especially not audio/video files).

A 32Bit application can only address 4GB RAM.

If you're working with virtual-instruments, that's a tough limitation.

ie: Loading a full kit in Superior Drummer 3 takes about 2GB RAM. 

 

Development resources are finite.

In today's economic environment, it doesn't make sense for a company to spend resources on antiquated products.

BTW, CbB is not alone in this situation. 

• Steinberg doesn't make a 32Bit version of Cubase 10
• Line-6 only makes a 64Bit version of Helix Native

I'm old, but I loath Cloud Storage for anything but small files.

I use the Adobe Creative Cloud (good use from it), but they are the worst at having applets running in the background.

Even with services set to "manual start", you have to manually stop creative cloud processes.  They won't stop upon closing Photoshop, Premier, After Effects, Audition, etc.

I have the same experience with Bandlab Assistant. 😉   Granted, it's not used often

 

19 minutes ago, Starship Krupa said:

And there you have it right in a nutshell. They finally caved to the "average user's" inability not to click on every piece of crap that someone forwards to their inbox.

Both Microsoft and Apple have caved to the "average" user.

Apple has all but abandoned their power-users.  Look at their machines... virtually no user-serviceable (internal) expansion 

• A new boot drive means a trip to the Apple store
• There's no expanding RAM for any current MacBook Pro (what you buy is what you get - WYBIWYG)

Lucky for us, with the Pro version of Windows 10, we can still take full control.

2 hours ago, Steev said:

 Want to see Cakewalk by Bandlab perform like an ULTRA LOW LATENCY fire breathing dragon and actually save several hundreds of dollars on a KILLER software bundle? Latency so low you will be able to run FX in real -time recording thru Cakewalk "Input Monitoring"? Behold the World's BEST selling and most trusted USB audio interface The Focusrite Scarlett 2i2 2sd gen USB 2 achieves MUCH lower latency than any old legacy Firewire 400 audio interface can even achieve, and has noticeably better digitally controlled mic preamps and signal flow/path curcuit protection.

 The Scarlett 2i2 is a little brother to my Scarlett 18i20, it performs the same, it just has less I/O. Running on TODAYS current Focusrite ASIO drivers, tested on my 6 year old Win 10 Pro, AMD FX 8370 (8-core) with 16 gigs of DDR 3 /PC 1866 RAM installed tested with CbB with 64 bit Double Precision Engine Running.

Recording at 24/44.1 or 24/48 achieves an effective recording latency of 2.7 msec... 12.6 msec total roundtrip. (Humanly impossible to detect.)

Recording at 24/96 achieves an effective recording latency of 0.7 mesc with a total roundtrip of 6 mesc. (Humanly impossible to tell the difference.)

I'm not sure if the above was serious... or in jest

By today's standards, running 6ms total round-trip latency isn't really that low.

ie: RME Fireface-800 (now about 15 years old) connects via Firewire... and can achieve 6ms total round-trip latency at 44.1k using a 64-sample ASIO buffer size. 

 

Using a Presonus Quantum (connected via Thunderbolt-3), you can do things like run Helix Native (software only version of the Helix guitar processor) at 96k using a 32-sample ASIO buffer size.  That's 1ms total round-trip latency.  To be able to to this effectively requires a fast well-configured machine.

If the comments about latency perception were serious, you absolutely can tell the difference between 6ms and 12ms latency (especially with high transient instruments). 

Dial up your favorite piano sample library... and set your ASIO buffer size such that playback (one way) latency is 6ms.  Now double the current ASIO buffer size so that playback (one way) latency is 12ms.  Which one feels tighter and more immediate while playing?  If your audio interface permits,  cut the original ASIO buffer size in half (3ms playback latency)... and compare  playing response to the previous two settings.  It's pretty easy to feel the difference.

 

If you're going to monitor thru software: 

• Anything higher than about 6ms total round-trip latency starts to feel pretty sluggish
• Upwards of 10ms total round-trip latency starts to feel unbearably sluggish
• Below 5ms starts to feel more comfortable
• Below 3ms feels tight 

 

If you have an audio interface that allows super low round-trip latency, you can test this for yourself.

• Set the audio interface to operate a 1ms total round-trip latency
• Open a test project with a single audio track
• Insert a delay plugin that offers fine control (1ms increments) and set the delay to be 100% wet
• Use the delay plugin's time parameter to simulate various amounts of (additional) round-trip latency

 

If you're going to do something like use a V-Drum kit to trigger drum samples from Superior Drummer 3 (in realtime),  you're going to want the lowest possible round-trip latency.  Round-trip latency above about 3ms will start to affect the feel.  Round-trip latency of 6+ms and the drummer will feel he's playing thru molasses.   With round-trip latency below 3ms, the drummer will feel much more comfortable.

 

 

Assuming the OP is running a desktop/tower,  there's numerous solutions.

ie:  You can get a single 5.25" bay that has four removable 2.5" trays for SSDs (for SSDs in a single 5.25" bay).

If the OP is running a laptop or small form-factor machine, there may be zero internal expansion.

If the OP is running a laptop/SFF machine that has Thunderbolt-3, using an external Thunderbolt drive enclosure is also an option (albeit somewhat expensive).