As an electronics designer, I want to point out that the way you calculate the required bandwidth is fine, as long as the connection is capable of transferring the data you need within the minimum latency you required without adding any significant latency of it's own to the mix. I'm not saying that USB 2.0 isn't fast enough for audio, I only wish to add to the conversation :) I'm going a bit deep, but I'm sharing my experiences when designing and specifying data connections for applications. Hopefully it has something to add. Let's say you have 32 sample data buffer and 100 channels of 44.1khz@24bits of data. Thus each sample accounts for ~23 microseconds of real time and thus 32 sample buffer accounts for ~0,74 milliseconds. That amount of data in this example comes to 24bits/sample * (32samples/buffer)= 768bits/buffer * 100 channels, so required bandwidth is 76,8kbits / 0,74 milliseconds (0,00074s), or ~103megabits / second. [ 76,8kbits * (1s/0,00074s) = ~103megabits/s ] If the bandwidth of the connection is such, that to transfer of the data happens before the next buffer is to be sent, it should be fine, right? Because no dropouts occur. But that minimum required transfer speed means that it takes as long as the size of the buffer in REAL TIME to transfer all the data over the connection that barely meets the requirements. And that effectively DOUBLES the latency from 0,74ms to 1,48ms.. just from the speed of the connection. In this instance, USB2.0 has ~4 times the required bandwidth and thus _shouldn't_ introduce any meaningful latency to the connection, right? Oh, but USB 2.0 is HALF DUPLEX, thus unless your connection is completely one way only and doesn't transfer audio back to the interface outputs as well, we've halved our total bandwidth from 480mbits/s to 240mbits/s. But we've also affected the LATENCY because to transfer that data over the connection first requires the input data to be sent over the connection, then we have to wait for the output data to be sent or they're out of sync. Thus if you can transfer 100 channels in e.g 1ms one way, it takes 2ms to transfer both inputs AND outputs. So half duplex not only halves your total bandwidth, it also doubles your latency for bidirectional communication UNLESS it can be completely out of sync where you can just push the data to the IO when it arrives without waiting all other channels. But that's undesirable. Also, each transfer has overhead in regards to protocol AND because the chip-sets themselves have RX/TX buffers that in most instances you cannot change or affect. all that adds latency, depending. So what if I double the buffer? Well now you've increased the latency obviously, but we've not changed our bandwidth requirements at all. So what did we learn? Our available bandwidth ALWAYS adds to the overall latency of the system and at MINIMUM bandwidth required, it doubles the latency (obviously more than that, because overheads and everything else). And even if you decrease the latency requirements, only the percentage of the total latency that the data connection is responsible of, decreases. But it cannot add less than what's minimum required to transfer the data without dropouts. And USB isn't very efficient anyway. There's other chatter on the bus, OS's don't do very good job at minimizing overheads that exist beyond drivers abilities, you have the chip-sets RX/TX buffers to contend with, the IRQ stuff... so it can be either very solid or VERY POOR. and you might not reach very low buffers in many cases and even if the driver says "32 samples" the total latency might be something completely different. Thunderbolt is VERY efficient and VERY fast. Thus it adds almost nothing to the overall latency. It's less chatty, is full duplex and has less between it and the processors DMA controller than USB, thus you can achieve way more stable setups in very low latencies. And it eats less CPU resources since it HAS the DMA and direct access to the processor without the OS interfering. Thus other stuff going on in the system doesn't so easily cause buffer under runs on software side norwithin the processor interrupt system because it uses less interrupts (DMA). USB3.0 is also ways better, but it's still USB with some issues that USB2.0 had as well. Are they really worth it? Well... _maybe_? :D It all depends on what you do and what you need exactly. I hope this gives any insights or understanding :)
I have an old firewire interface that I still use for the very reasons you noted. The limitations of processing audio signals to and from your DAW without introducing unacceptable latency. My interface my not be near as fast as USB 3.1, but it has broader bandwidth and can simultaneously snd and receive data. Would love to move up to Thunderbolt, but there are too few options and most are cost prohibitive.
I'm very glad you made these points, because over the past 25 years I've used many different Audio Interfaces with connections ranging from USB 2.0 through to Thunderbolt 1, 2, 3 and up to PCIe cards setup for DMA access (RME AES to be precise). The worst audio performance I had was using USB 2.0, which had lots of dropouts, huge latency, and if MIDI was also connected, there always seemed to be stability issues. Thunderbolt 1 to 3 got progressively better, with an AVID OMNI Native HD thunderbolt system giving me very stable performance and large track counts with tons of plugins including a MIDI USB interface. But the best performance by far was using an RME AES card on a PCIe bus on a Windows PC. MIDI was flawless. Latency was unbelievably good, where I could run large track counts at 64 samples while recording with no dropouts or glitches. Audio was also flawless, and whatever I threw at this system it did without complaining. The downside was the messing around with different components to find a Graphics card which gave me the most stable audio. Since I now no longer own a Windows computer with PCIe slots, a Mac with TB-3 is the next best thing, and I can get excellent performance and stability using a Lynx Aurora (n) with TB-3 and MIDI over USB 3.0. If you paid me, I would never go back to using USB 2.0 for audio + MIDI. But I'm still hankering for a PCIe based setup on a PC even though I think the Mac Core Audio is a lot easier on the nerves, than getting a Windows PC to work efficiently with audio.
@@theocorfiatis8456 I never understood how PCIe interfaces actually work in practice. I know you plug the card directly into the computer motherboard, but the analog instrument still needs a device to convert the signal to analog. Or do you plug the analog instrument directly into an input on the card itself? Either I'm really stupid or that whole process has never been adequately explained for newbies.
I have a MOTU 8 Pre-es which supports both USB 2.0 and Thunderbolt 1. I'm also running an AMD Ryzen 9 and I was super pumped until I learned after the fact about the lack of support for Thunderbolt on AMD. The reality of it is most of my projects are mixing projects for clients and not recording. I don't use outboard gear, so lots of ins and outs aren't needed. USB 2.0 works awesome for me.
Amd straight up said they will not support Thunderbolt because it's an enemy's tech(Intel). I've always had insane lens with AMDs stuff. Everything from GPU problems to motherboard problems, cpu's failing and needing to be replaced. All sorts of shit. So it's unfortunate that Intel is sucking for the past 8yrs. Still, I have always had less issues with them.
@@rileyioacura Intel recently announced a new Thunderbolt version with speeds of up to 80gbps. They haven't said if they will call it Tb5 or Tb4v2, but it's definitely on the way.
Wonderful video! Nice to see another RME fan making good work. When I was first starting out I remember getting a second gen Scarlett 2i2 and not using it to its full potential. I would have at least 700ms of monitoring latency through Studio One and decide to just use the Scarlett's direct monitoring function. This went on for about a whole 6 months before I started doing more research into audio engineering and understood what buffering and sampling rate were and how they worked. I properly configured it and had basically sub 100ms latency and was very happy that my 2i2 has that capability. During this whole time I was wondering if I needed to upgrade to Thunderbolt but was always held back by the costs. Nowadays I feel like Intel and Apple just need to push it as "better" simply because they have stakes in the technology and not necessarily because "Oh, it's just better dude." Went through a couple other interfaces, all USB, and all perfectly capable (with great emphasis on my old Audient id22, fantastic preamps) and eventually bought a Babyface Pro on sale and have been loving this thing ever since. I hardly feel like I'll ever need to upgrade (emphasis on need, I do want to upgrade to a Fireface like yours in the video) since my work consists mostly of my voice over work, sound design, light mixing and post-production, I don't work with music production so I'll never need to run 200 tracks simultaneously. Very refreshing to see a well made video take a pragmatic approach to this topic! Congrats again!
4:38 - It's not a misconception. I have two Focusrite Clarets 8pre, except one is thunderbolt 2 and the other one is USB 2. There's something in the protocol or in the way thunderbolt is wired directly into the CPU lanes, I don't know what it is - but the round trip is considerably lower via thunderbolt with these interfaces with near zero latency at 96 KHz, without problems. Sadly, I now need to let the thunderbolt interface go, as it no longer works with my new PC with thunderbolt 4, no matter what adapter or hub I try. I'm thinking of switching to RME, but having 24 channels with pres included is going to cost around 4k or 5k with RME, compared to 2k my Clarett 8preX with two Clarett Octopres costed me. So... I'm really trying to avoid the switch, looking for solutions.
@@bassyey Not true. There are just USB ports on hubs with no devices connected to the other ports. No USB has or can have direct memory access like thunderbolt.
Good choice! I myself have been a RME user for the past 17 years. I started off with the Fireface 800 and now use a Fireface 802. Rock solid with great latency performance.
Is there difference in sound 800&802? For me ff800 sounded crystal clear and my babyface pro little flatter and not so clear as 800. Now I want to change it to rack rme, cause I think their analog components are better, so thinking about 802 or ufx+. Thanks
USB drivers are less efficient in general due to the way USB indirectly interfaces with the CPU, and USB itself involves a fixed bus latency. WIth thunderbolt and PCIE based audio interfaces, you have a lower starting point for your round trip latency which gives you another unit of buffer size, sample rate, or plugin delay to play with before approaching perceivable monitoring delay. This can be invaluable in some situations. Also, since you can do dynamics and EQ processing in totalmix fx with that audio interface, the interface does in fact have a DSP chip. I'm assuming you meant you just can't chuck on proprietary processors at any stage of the signal path like UAD and PTHD interfaces can. I am very glad you mentioned the bandwidth misconception and driver efficiency though. USB 2.0 is more than enough and RME drivers absolutely slap.
I agree, the fixed bus latency is a key reason for using thunderbolt over USB. For many applications it doesn't matter that much, but for some it matters a lot.
@@crackedmagnet 100 percent, i'll reiterate what you said by saying that USB is definitely more than adequate for majority of recording situations. I believe a lot of people get conned into buying thunderbolt interfaces due to bad experience with suboptimal drivers from brands that don't care.
@@sorbpen live tracking into a project that was given to you with a bus processor that has delay on it but is necessary for the vibe so you don't have to bother with creating sub projects, live monitoring chains, disabling half the project, or freezing. Only to undo all of that again for the final mix. Saves you a lot of time. And as I said in the comment, it gives you one extra thing to play with. You can reclaim some of that latency with the minimum buffer setting that you'd have by working on 48khz instead of 96khz, which is more CPU efficient. Higher number of plugins overall regardless of if they introduce latency or not. Stability if you use your daw in live music to run lights and Fx chains, monitoring and foh. Etc
@@jaydensydes3478 I fail to see where in any of those use cases the difference between RME 2.0 USB drivers, and thunderbolt latency would make any difference at all, except possibly running lights, but any DMX system worth it's money will let you offset so you can sync with whatever audio you are feeding to your audience and or live performers. Your guitarist is used to at least 12-25ms delay just from standing a few feet from his amp, etc etc. So i still think the word "invaluable" is a bit strong here. USB can poll up to 1000hz, i don't know what poling rate the RME drivers or other interfaces are running at, someone might be able to chime in, if it was say 125hz like early USB devices sure then we can talk. I was considering getting a PCI-e card from RME at one point, but decided on picking up a Fireface 400 because i came to the same conclusion as this video the latency difference would be neigh unnoticeable if i went looking for it and for my use-cases completely moot, that interface is what 15 years old now? I got it because it had firewire. My reasoning was that since I'm a bedroom producer, and programmer i kinda need to use my computer for other things as well. And in my experience the quality of USB drivers from different manufacturers of peripherals are of varying quality to put it kindly. So keeping it on a separate bus sounded like a good idea. Now please correct me if I'm wrong but the use case for thunderbolt that i see would be massive interfaces with hundreds of I/O, and to my knowledge there are very few of those on the market
Excellent video. As an RME owner in the past and a potential UFX+ customer I am completely sold on RME products. What is missing from the video are any numbers on the round trip latency in your setup between USB 2-3 or Thunderbolt. I find that omission troubling,. However latency as a feature of products lacks an accepted standard so I get why it may be missing. The only thing I find missing in the current RME approach is DANTE. Again I understand DANTE licensing adds $300ish dollars to your product and their approval process my require revealing more than RME wants to reveal. Love it or not DANTE has become an internationally accepted process with lots of "Big Guy" adherents so it belongs on a piece like a UFX+ with it's otherwise swiss army knife approach to sound recording. Anyway, it's a great video. Thank you for posting it.
Great overview and explanation regarding the bandwidth and utility USB 2 still offers. Been an RME fan for quite a few years now, and a quite satisfied user of the Fireface UC and Babyface.
1:34 USB 4 provides 40Gb/s, not 20Gb/s. The version of USB that provides 20Gb/s is USB 3.2 Gen2x2. (I know, the naming scheme is terrible. It hurts me too.)
One reason to consider an Apollo's interface for latency is for the loopback. If you monitor your mic through loopback (obs streaming), there is noticeable latency.
@@JoBu133 oh I know the why, it still sucks though. It can accept dc power, why not just disable the dynamics when bus powered, enable them when dc powered instead of removed altogether. Just figured that would have been better. I mean, I still went and bought a babyface last week anyway so I'm gonna enjoy it regardless :)
This is a great video however, there are a few things that are unfortunately incorrect, so let me kindly point them out : 1) The bandwidth shown here are the *theorical* max bandwidths and not the real-world transfer speeds bu the bulk packet *size*. In real world use, bandwidth is occupied by communications with the host device (your computer) as well as being affected if the cable provides both power and data connection (as in all bus-powered interfaces) because then a significant portion of the conductive materials in the cable are utilized to transfer the 5V necessary to power the interface. In real-world, a Thunderbold 3 device of top performance will give you around 2.3 to 2.5 gigs per second for data transfer rates. SInce we are all nerds here is the real explanation "USB 2 uses 1 millisecond frames, and in High Speed (480 Mb/s) mode they are divided into 8 micro-frames. The maximum size of bulk packets (used by USB mass storage devices) is 512 bytes. According to this very informative document the theoretical maximum is 13 packets per microframe. So the theoretical maximum speed of a USB 2 drive is: 1000 * 8 * 512 * 13 = 53248000 ~= 53 MB/s" Therefore, 46MB/s is dangerously close to the theorical max speed os USB 2.0, so why not use a 3.0 bus which is super-cheap anwway? That's quite weird to me. Kind of like they want to prove a point and their drivers are amazing enough to keep up but still, if you don't have a really perfect connector on your computer's side, you will run into issues if you full that interface up. 2) USB 3.0 gives you 4.8 Gb/s; USB 3.1 gen 1 gives 5Gb/s bulk packet max size (the difference is marketing really), USB 3.1 gen 2 gives you 10 Gb/s max size; Thunderbolt 1 gives you the same 10Gb/s size (*not* speed). Thunderbolt 2 (usually uses MiniDisplay port, although USB-C ports can also be used in certain drive enclosures) has a bulk packet max size of 20Gb/s. Thunderbolt 3 gives you 40gb/s max size, and USB4 *also* give you 40gb/s max size. This is because USB and Thunderbolt are competing protocols, but USB is universal, while Thunderbolt is proprietary and belongs to Apple and Intel. Therefore it takes less time for them to produce it because it does not have to go through 2 bazillions tests like a universal protocol has to. Then they can sell TB ports to other manufacturers. In reality USB ports are a much better value for money. They sound more generic; USB 4 is the same as TB 4. It just came out a little bit after. Thunderbolt is 100% marketing, and in fact, is inferior since it's not tested as rigorously nor is it available on most devices. In practice, USB 4 and TB 3 are the exact same, TB 2 and USB 3.1 gen 2 are also the exact same, and so on. I know, the nomenclature is really confusing.
I am using AMD. I wonder if that contributes to my problems with Thunderbolt on my PC. Thanks. The RME effects work well for tracking. I agree that UA plugins are premium quality, but they are also much more expensive and processor-intensive. So it’s not really a one-to-one comparison.
One thing that I think you haven't touched on, especially when it comes to different versions of USB, is power delivery. Some ADC/audio interfaces come with built-in processing (Revelator io series, Apollo, Antelope come to mind) which require more power. The different preamps and phantom power supplies also demand different types of power. USB 2.0 can only do around 2.5 w (@ 5V 500mA), while USB 3.0/3.1 can do 4.5W (@ 5V 900mA). Thunderbolt on the other hand, can do around 9.9W (@ 18V 550mA). Something to keep in mind
I really appreciate the info in this video. Thank you! Also the comments from @Mtaalas below add some additional info that is really important and needed to round out the conversation. I run a 7.1.2 Atmos system via thunderbolt to my Mac book pro but I have used my interface for 21 simultaneous mics being recorded in large concert recording. My interface is also Dante enabled which is how we passed 21 channels from the stage to front of house position simply over ethernet cable. It really does just come down to what your needs are and getting a device that meets those and possibly provides a little room for occasional sessions needing larger simultaneous input
My next interface will be the focusrite dante enabled red 16line. It’s thunderbolt. I don’t know if there are any usb interfaces that offer dante compatibility at that high of I/O
My Apollo X4 is thunderbolt 3/4 only. I love the plugins. When i switched from an Apollo Twin USB to the X4, the difference in my audio was clearly noticable. There may be other things going on but i love the sound difference.
If anyone tells you to DM them on telegram in the comment section in TH-cam, they are scammers. Pay attention and let them know you know they are scammers
Top audio interface! I’ve been using MKI for a few years, and it’s great. Been eyeing this MKII with on-board recording and improved on-device controls among other things.
Excellent tutorial on some of the nuts and bolts behind the scenes. I’ve never been happier since I realized these trivialities have nothing to do with making music, and are only convenient distractions from productivity.
yes sir! finally someone nailed it ... except RME with its own video! this thunderbolt advertising has been driving people nuts for years now. running a RME Fireface UFX II via USB 2 by the way :-) roundtrip latency much under 3 ms if I want to :-) nice job dude! love from germany!
Seems like there was a key point missed on thunderbolt. Thunderbolt is a dedicated port, where USB is shared with multiple ports on your system, it can be congested if the rear of your machine is "streaming gamer ready" with 30 different devices just to make a single light swirl colors.
I used to use a Baby Face FS and the drivers are fantastic. The latency is imperceptible as well and tons of headroom along with zero noise. My thunderbolt Apollo provides all of these things, minus the superior drivers but, allows for live tracking with plugins. RME is absolutely top notch and I will definitely be picking up a converter of some sort in the future if my setup and work flow expands. Thanks for the video!
@@MaximusWhyman Aww dang I'm in the U.S. we don’t have many places that do good trades. They'll try to give me $700 for it at best in a store which ain't happening. 😆
Thanks for the Infos ..! How do you connect the the interfacce USB2 to the Mac M1 USB-C Port...? USB hub..or adapter ? Any side effects on the interfacce performances or power issues?
Great tutorial .my set up in addition to the scarlet audio interface 6i6 gen 1 uses an outboard mixer for monitoring so I don't have the latency issues as I avoid monitoring from the DAW/interface
I hear you, but theres times I'd sure LIKE to be able to monitor thru the box - I'm sure the RME's a perfectly nice box, but that cute new socket's just TOO tempting,...
Latency on effected audio is the reason I still use an old Line 6 UX2. It is one of the cheapest interfaces with onboard effects, so I can plug in my guitar directly and use the Pod Farm software to load up a nice tone without needing to open my DAW at all. This way, I can record a clean DI and still hear whatever tone I want to hear and play accordingly. Best of both worlds. I firmly believe onboard FX is the way to go. They don't need to be great, just good enough to get you through the recording process. That said, I still use Pod Farm on many projects, so even so called outdated software still has its place.
I've been thinking about the bandwith thing with USB2 for years. 🤔 Thanks for the to-the-point illustrations! 🙌 And you also got me interested in RME products 😁 - Eero
I've got the motu Ultralite avb, and since I record classical music I don't need very low latency. should I need it I can use direct monitoring through the motu's mixer with zero latency
I'm not sure if its my Focusrite interfaces or USB but I've had 2 interfaces with USB and they occasionally both lose sync and sound like crap, and I have to reconnect them to get them to sync. Never had that issue with Firewire.
@@AudioUniversity I hope you never do, I know of others that have not hit this issue, but I've hit it with 2 different PC's and 2 different DAW HW - both Focusrite Scarletts... that might be the weak link but I just don't know.
I have had issues with my Focusrite Scarlett 4i4 for a while. Those issues stopped happening after I followed Focusrite advice and stopped using the interface connected to an USB hub (regardless or they are powered). Instead, it's now directly connected to the laptop's USB root hub.
Can you please show what issues might arise when using Thunderbolt? I though any Thunderbolt 3 device would automatically fallback to USB 3, when the host does not support Thunderbolt? What is the actual latency with minimal processing?
@@lorenhoward8178 Here I have an i5 13600k, Scarlett Solo (3rd Gen) 32 buffer and 48k (my configuration) 5.17ms overall latency in ableton. (I don't use 96k). But at 32 and 96k it is 3.67ms.
@@lorenhoward8178 Here I have an i5 13600k, Scarlett Solo (3rd Gen) 48k with 32 buffer (my configuration) is 5.17ms overall latency in ableton. (I don't use 96k). But at 96k with 32 it is 3.67ms.
Thanks, that is a comprehensive summary. However my experiences with USB for audio are quite different. For me USB is problems and usually audio errors. I have to use it since they stopped mounting FW sockets on laptops. FW was far more reliable - rock solid recording. I`ve been using MOTU UltraLite hybrid mk3. I have to buy a thunderbolt laptop and try to use FW-to-Thunder adapters to see if that works, as the current USB situation on my laptop makes the interface unreliable.
I can confirm it works perfectly fine on a modern mac with thunderbolt 3; but the easiest and cheapest way is to use two dongles (thunderbolt 3 to thunderbolt 2 then thunderbolt 2 to firewire); those are the official apple dongle and will set you less than 100 eurodollars. I have the FW only version which is officially not supported anymore but the two interfaces share the same driver. Only issue I have is I can't put the computer to sleep when the interface is on or I get a kernel panic.
@@AudioUniversity thanks for the great video! Have you used DigiCheck much? I just set up 5.1 surround this week and I'm finding the surround meter and other tools extremely useful for calibrating and setting up routing.
Thanks for the video. I was pleased to come across it. Hearing for a very long time how great RME Audio interfaces are. And that I've been hearing for a long time how Great Thunderbolt is compared to most other popular connectivity. Your video was good to see, as I'm considering a USB interface upgrade from my trusted old : Muto 2408 mk3. I may even get a second hand RME.
I have a Apollo Twin X thats connecting and working onnmy computer but it seems as if what i here in the headphones is not what getted printed in my daw. Do you think this is my computer are is it because im using a cables matter thinder 4 cable. I also used a belkin thunderbolt 3 cable and i still get a degraded sound of audio.
Hi. Thanks your highly educative and informative videos. I have a bit of a challenge with my Babyface (original) interface. It has served me well since 2010, but recently I got a used 2015 iMac and the interface just won't work with it. I can't even resell the interface because I won't get the value for it here in Lagos, Nigeria. If I sell it on eBay the cost shipping to its new owner will wipe out the money I get for it. I have searched for solution online and nothing has worked. Any advice on a workaround will be appreciated.
I agree usb may be sufficient but why is it that rme hdspe aio pro is faster than their and basically any existing usb interfaces? Pcie is still faster than usb for now at least.
Maybe for the home studio or project studio market but definitely not the Professional Commercial market. Thunderbolt is an external PCIE connection as it doesn't get any better than that given that Protools HDX has the lowest latency in the industry with PCIE. Thunderbolt gotten pretty close. The HDX cards perform around 0.7 ms. Thunderbolt is a pro level studio connection esp for large scale deployments. USB can only handle a limited number of i/O.
You get lower latency and less load on the CPU when you use thunderbolt. Thunderbolt is really an external PCIE connection. Most high end professional studios use anything but USB. Thunderbolt, Digilink is pretty much the industry standard in commercial facilities, along with Dante, AVB and MADI that uses a lot i/O. USB has a limitation on i/O and performance which is why you rarely see USB ports on high end converters.
@@eman0828 That doesn't really answer the guys question ;) That said, I tend to agree- my understanding is there's a bit of unavoidable latency with the way USB works, and I'd REALLY like near-instant monitoring in the headphones, which is why I'm hanging out for Thunderbolt for my next Motherboard (come ON AMD pick it up will ya!) (ok I'm sure RME's drivers do just great & I sure wouldn't turn my nose up at that little box.. But having had to deal with echoey lonng monitor latencies, I'd like everything as snappy as possible, thanks! Glenn Frickers gone DANTE, but for me thats a tad overkill ;) )
You don't see USB ports on high end converters cause no wants to spend the money to write drivers for it and get tons of emails about things are not working, that is why they all have AES, ADAT. The RME PCIE AES card is standard but there is no reason not to have the RME UCX II, you get a lot of connectivity, its portable and its nice addition to any studio and if you looking to control Eurorack it has DC coupled outputs, also if you are using Midi, you will get incredible timing cause they writing the driver, they have it down, now if Burl could team up with RME that would be a game changer.
@@PylonRecordsdotcom Lol RME AES is NOT a standard. That would be Thunderbolt and AVID HDX. RME market share in the US is rather small since the UA Apollo out sells them around the world. Lynx Auora, Apogee Symphony i/O and AVID dominates the high end commercial studio market. MOTU and Focusrite network audio interfaces are also found in high facilities.
The reason why USB3, Thunderbolt and Firewire are inherently better than USB2 for Audio interfaces is not bandwidth. Its because USB2 uses polling to get data. Thats means that the device can only sent data when requested, which is pure garbage for real-time applications like audio. Its also the reason why people kept using PS/2 keyboards for ages. USB3 also doesnt do that. Can you make it work for audio? Yeah, if you work on it for 20 years and spend billions you can make everything work. RME has these impressive figures because they 100% do the USB stack themselves in their interfaces. Most other manufacturers use readily available interface chips.
I got steinberg's mr816 csx on firewire and texas instruments pci-e card years ago and it worked great on win7. after I switched for win10 and upgraded drivers I got annoying pop outs and all of sound nightmares. was digging net to find solution, tried everything. nothing worked. some time ago I browsed all files of that driver and found stand alone file in utility folder. it brings up the window with choice of ieee 1394 buffer size. was set on small. I set it on large and all problems were gone. no info about that anywhere so this is last time I go for steinberg interface. It still seems strange for me why it all worked well on small buffer with win 7. Im afraid of going win11 because of close to none driver support from steinberg.
Hi, I'm interested in your opinion. I use Steinberg UR28 audio interface. Will I improve the sound with the RME Fireface UCX II. I use plugins, vst instruments and record keyboards and guitars as needed. Tnx
I'll probably stick with my PCIe card as long as I can, though I notice that less places are selling them. I've been using an RME HDSPe AIO Pro for about a decade now and it's been very reliable and capable of low-latency when necessary, probably lower than the lowest possible latency of USB.
When measured, nearly all PCIe card suck in latency compared to asio usb. You're lucky to have one of the few expection with the model you have. RME drivers are on a different level.
I still have a firewire interface. I upgraded my SiiG Pcie card lately for a newer one with better TI chip, and now I can go down to 2ms latency 96khz/64samples buffer on win11 with reaper.
If this is true why do thunderbolt audio interfaces achieve lower latency figures than usb? RME's thunderbolt capable interfaces and presonus 2626 all come in a few ms lower in terms of round trip latency. There must be a reason for it. The 2626 can achieve around 2ms RTL at 32 buffer and the UFX+ around 2.5ms. RMEs usb interfaces seem to achieve more like 5-6ms at the same bugger size.
when working on a big project... 120 tracks or so. Does it matter if the interface is using usb 2.0 vs TB? I get the point about the physical IO, but what about playing larger projects? Thanks!
Great question. The number of channels going between your interface and computer is the main consideration. That could mean recording, playback, outboard inserts, or even using the interfaces built-in DSP as an insert to the DAW. The number of tracks in the DAW session will be dependent on the DAW, RAM, CPU, etc. Not the connection to the interface.
When you say ‘switching back to USB’, what thunderbolt interface did you own? I have a UCX mk1, it’s great. I bought a Thunderbolt Antelope Orion interface, and although the driver performance was not as good as the RME, the Antelope has more depth and clarity, it just sounds better.
I use the DAW as an effects processor for live performance. The Baby Face Pro FS with USB2 has a round trip latency of 22.3 ms. My Quantum 26x26 thunderbolt has a roundtrip latency of only 1.52 ms. So, I don't see the reason in not using the thunderbolt interface. The only problem many people may run into is the thunderbolt cable. You MUST use an ACTIVE thunderbolt cable. They are usually pricey, from $60 and above. I got my 2 meter 'active' thunderbolt 4 cable for $23 from Amazon. There are unlimited VST plugins out there. So this solution is much better and cheaper than buying an Eventide H9000 or a Lexicon PCM92. In response to the title. I will never go back. All my USB2 devices will be sold or given away. My Baby Face Pro FS was pricey but good I thought. Until I found the Quantum 26x26, which is better and cheaper with or without a $100 active cable.
Agreed, I bought a Presonus Quantum 26x26 for $400 on eBay. Using thunderbolt 3. Why would I "switch back" to 2008 technology ? My latency problems are gone.
@@AudioUniversity oh man I was looking at that one too for mobile and laptop solutions. So you do like that one? I swear I did not know what I was NOT hearing until I went to RME. It's like someone took a blanket off my music.
Hello! I have a question. I'm looking to buy studio monitors and would appreciate your advice. My room is about 12ft by 10ft wide and from the floor to ceiling is about 8ft tall. The floor is wood so it would be reflective. I'm not sure which size of monitors would be good or which brand (I compose and interested in orchestration and ensemble, so I guess songwriting and film scoring? I also enjoy worship, neo soul, lofi, gospel... so music with a band setting too... All with the goal of giving glory to Yahweh and leading people to the Messiah, the Way, the Truth and the Life). I've read a book on recording and they recommended monitors not going under 6" but I've taken a look at KRK 5" ones. Also looked at the Yamaha HS5. Any recommendations? Thank you and Yahweh bless! :)
I found once you have a serious CPU at some point drivers and connection may be the bottleneck. I stick with thunderbolt since it's really just a pcie lane so i can monitor what's in my daw directly including processing plugins. That said, i have also have a USB 2 interface for when i only need quick audio not my whole recording rig. Cool video, you are slowly talking me into rme products too!
Thanks for your opinions. I've used many different Audio interfaces and computers over the past 25 years and from experience found USB 2.0 to be by far the most laggy and with the most dropouts. I disagree with your analysis based on many years of experience. There are many factors which determine how efficient and stable Audio interfaces are on a particular computer, and they are not just related to bandwidth over a USB port. The most efficient and stable I owned was an RME AES card plugged into a PCIe slot in a Windows 7 PC. It left USB 2.0 for dead in every respect, including latency, dropouts, overall system stability and the sheer number of tracks and plugins I could run on a moderately spec'd PC was perfect for doing large mixes. Those interfaces all use DMA which bypasses a lot of the interrupts caused by other system resources such as USB. Those other resources can drastically degrade audio performance, like having an SSD plugged into another USB port close to the one the Audio interface uses, which can drag down the bandwidth of the Audio interface quite drastically resulting in dropouts. Using MIDI over USB could also cause dropouts, if there is a lot of MIDI data over one USB 2.0 port, which happens to be on the same USB controller as the Audio interface. Thunderbolt 3, which I use now, bypasses the USB SSD drives, and I get excellent stability and no dropouts. I can understand why you oversimplified the explanation of why USB 2.0 is good for Audio. At face value you are correct, but in reality, there are always other factors which contribute to real-time system stability. When I started using computers to record, plugins used to have a static UI, with a few sliders on them, and were very efficient in terms of graphics. Nowadays many of the top brands of plugins have complex interactive graphical interfaces (which I love), displaying all sorts of real time information. This fact makes it mandatory to have a moderately well spec'd graphics card so that you don't get lots of audio dropouts, because the graphics card takes away real time processing power from the CPU. I discovered this the hard way when I got a Mac Mini (late 2018) which has a very decent i7 CPU but extremely low spec'd on-board graphics chip. Despite the good CPU performance, the computer was unusable for serious mixing, because the CPU spent most of its valuable clock cycles processing the UI data from the plugins I was using, and of course USB ports have lower priority than Graphics, because Graphics uses DMA, but so does Thunderbolt. The Audio interface I am using on it is an RME Babyface pro FS, which as you know uses USB. So I added a moderately powerful external GPU though TB-3, and all of a sudden, I had stable and very usable Audio without dropouts. I've since relegated that computer to the recording room and use a much higher spec'd machine for mixing (with a TB-3 interface) . All things being equal, yes RME make amazingly good interfaces, with excellent drivers, which I would recommend to anyone without a second thought. USB 2.0 should have tons of bandwidth for multiple audio channels, and it does if you have say 50 mono tracks in a DAW without plugins, and there are no other devices attached to other USB ports. But in reality, you can't really make music unless you have some good plugins and SSD drives for your samples and backups and Audio and all your extraneous system apps have been pruned right down while you are recording. Ok, maybe that was hasty, because if you have a room full of analogue hardware you can render each track to your daw, without using any plugins at all. But how many of us have a hundred thousand dollars to get all those magical analogue boxes? I've never met a real working musician who can afford this stuff, so I guess it's plugins for most of us, and so that's why I'm making this point that just looking at USB bandwidth alone is just as impractical as looking at CPU speed. Neither of these metrics give you an idea of whether a computer can produce stable audio.
Very interesting. I've had many interfaces over the years, firewire (x 3) USB ( x 2) PCI (x 2) and my little Thunderbolt 2 Zoom interface has very similar low latency to my very old and expensive PCI MIX hardware from Pro Tools 5 (from the early 2000s G4 days). I've always had issues with latency on USB, and to a lesser extent, firewire. Thunderbolt and the very old PCI stuff wins for low latency, in my experience.
you can't beat PCI for latency certainly; and thunderbolt gives you pretty much direct PCI access so it's virtually identical. Firewire was still a bit better than USB though.
That's what I'm saying. PCIe-424 MOTU to FOUR 2408mk3 rackmounts for 32 balanced I/O analog TRS channels. 96k 24bit all day. Still killing it. Hackintosh build on a 768GB RAM twin i7 CPU server.
I had a horrible latency issue using a USB interface plugged into the USB-A ports of a 2018 MAC Mini. The only workaround I could find was to buy a Thunderbolt dock with a USB-C port and plug my Interface into that.
I thought I remember learning from _somewhere_ that Thunderbolt was, indeed a "highway with a faster speed limit," to use your analogy. This source claimed that, due to an inefficiency with how USB performs data transfer (error checking algorithm, maybe?), there is a bottleneck when it comes to latency in USB, even with low channel counts, and such a bottleneck was not present (or at least greatly reduced) in Thunderbolt. Anyone who knows more than me, feel free to let me know if I'm totally off-base. I'll admit that this presumption may be a desire to satiate my confirmation bias, as I dropped $2500 on a used Focusrite Red 8Pre (a Thunderbolt 2.0 interface)
As an electronics designer, I want to point out that the way you calculate the required bandwidth is fine, as long as the connection is capable of transferring the data you need within the minimum latency you required without adding any significant latency of it's own to the mix.
I'm not saying that USB 2.0 isn't fast enough for audio, I only wish to add to the conversation :)
I'm going a bit deep, but I'm sharing my experiences when designing and specifying data connections for applications. Hopefully it has something to add.
Let's say you have 32 sample data buffer and 100 channels of 44.1khz@24bits of data.
Thus each sample accounts for ~23 microseconds of real time and thus 32 sample buffer accounts for ~0,74 milliseconds.
That amount of data in this example comes to 24bits/sample * (32samples/buffer)= 768bits/buffer * 100 channels, so required bandwidth is 76,8kbits / 0,74 milliseconds (0,00074s), or ~103megabits / second. [ 76,8kbits * (1s/0,00074s) = ~103megabits/s ]
If the bandwidth of the connection is such, that to transfer of the data happens before the next buffer is to be sent, it should be fine, right?
Because no dropouts occur.
But that minimum required transfer speed means that it takes as long as the size of the buffer in REAL TIME to transfer all the data over the connection that barely meets the requirements.
And that effectively DOUBLES the latency from 0,74ms to 1,48ms.. just from the speed of the connection.
In this instance, USB2.0 has ~4 times the required bandwidth and thus _shouldn't_ introduce any meaningful latency to the connection, right?
Oh, but USB 2.0 is HALF DUPLEX, thus unless your connection is completely one way only and doesn't transfer audio back to the interface outputs as well, we've halved our total bandwidth from 480mbits/s to 240mbits/s. But we've also affected the LATENCY because to transfer that data over the connection first requires the input data to be sent over the connection, then we have to wait for the output data to be sent or they're out of sync. Thus if you can transfer 100 channels in e.g 1ms one way, it takes 2ms to transfer both inputs AND outputs.
So half duplex not only halves your total bandwidth, it also doubles your latency for bidirectional communication UNLESS it can be completely out of sync where you can just push the data to the IO when it arrives without waiting all other channels. But that's undesirable.
Also, each transfer has overhead in regards to protocol AND because the chip-sets themselves have RX/TX buffers that in most instances you cannot change or affect. all that adds latency, depending.
So what if I double the buffer? Well now you've increased the latency obviously, but we've not changed our bandwidth requirements at all.
So what did we learn? Our available bandwidth ALWAYS adds to the overall latency of the system and at MINIMUM bandwidth required, it doubles the latency (obviously more than that, because overheads and everything else). And even if you decrease the latency requirements, only the percentage of the total latency that the data connection is responsible of, decreases. But it cannot add less than what's minimum required to transfer the data without dropouts. And USB isn't very efficient anyway. There's other chatter on the bus, OS's don't do very good job at minimizing overheads that exist beyond drivers abilities, you have the chip-sets RX/TX buffers to contend with, the IRQ stuff... so it can be either very solid or VERY POOR. and you might not reach very low buffers in many cases and even if the driver says "32 samples" the total latency might be something completely different.
Thunderbolt is VERY efficient and VERY fast. Thus it adds almost nothing to the overall latency. It's less chatty, is full duplex and has less between it and the processors DMA controller than USB, thus you can achieve way more stable setups in very low latencies. And it eats less CPU resources since it HAS the DMA and direct access to the processor without the OS interfering. Thus other stuff going on in the system doesn't so easily cause buffer under runs on software side norwithin the processor interrupt system because it uses less interrupts (DMA).
USB3.0 is also ways better, but it's still USB with some issues that USB2.0 had as well.
Are they really worth it? Well... _maybe_? :D
It all depends on what you do and what you need exactly.
I hope this gives any insights or understanding :)
Yeha that right , people learn here .
Wow! This is absolutely fantastic added insight and info. Thank you!!
I have an old firewire interface that I still use for the very reasons you noted. The limitations of processing audio signals to and from your DAW without introducing unacceptable latency.
My interface my not be near as fast as USB 3.1, but it has broader bandwidth and can simultaneously snd and receive data.
Would love to move up to Thunderbolt, but there are too few options and most are cost prohibitive.
I'm very glad you made these points, because over the past 25 years I've used many different Audio Interfaces with connections ranging from USB 2.0 through to Thunderbolt 1, 2, 3 and up to PCIe cards setup for DMA access (RME AES to be precise). The worst audio performance I had was using USB 2.0, which had lots of dropouts, huge latency, and if MIDI was also connected, there always seemed to be stability issues. Thunderbolt 1 to 3 got progressively better, with an AVID OMNI Native HD thunderbolt system giving me very stable performance and large track counts with tons of plugins including a MIDI USB interface. But the best performance by far was using an RME AES card on a PCIe bus on a Windows PC. MIDI was flawless. Latency was unbelievably good, where I could run large track counts at 64 samples while recording with no dropouts or glitches. Audio was also flawless, and whatever I threw at this system it did without complaining. The downside was the messing around with different components to find a Graphics card which gave me the most stable audio. Since I now no longer own a Windows computer with PCIe slots, a Mac with TB-3 is the next best thing, and I can get excellent performance and stability using a Lynx Aurora (n) with TB-3 and MIDI over USB 3.0. If you paid me, I would never go back to using USB 2.0 for audio + MIDI. But I'm still hankering for a PCIe based setup on a PC even though I think the Mac Core Audio is a lot easier on the nerves, than getting a Windows PC to work efficiently with audio.
@@theocorfiatis8456 I never understood how PCIe interfaces actually work in practice.
I know you plug the card directly into the computer motherboard, but the analog instrument still needs a device to convert the signal to analog. Or do you plug the analog instrument directly into an input on the card itself?
Either I'm really stupid or that whole process has never been adequately explained for newbies.
this youtube channel is radicaly solving all my last unsolved problems, cheers Audio University I love you so much !!!
❤
You can almost forget about Thunderbolt if you have an AMD processor. Very few AMD boards support Thunderbolt and the ones that do are expensive.
I have a MOTU 8 Pre-es which supports both USB 2.0 and Thunderbolt 1. I'm also running an AMD Ryzen 9 and I was super pumped until I learned after the fact about the lack of support for Thunderbolt on AMD. The reality of it is most of my projects are mixing projects for clients and not recording. I don't use outboard gear, so lots of ins and outs aren't needed. USB 2.0 works awesome for me.
Thanks for sharing, LittleStudiosOnline!
The new Ryzen 6000 Chips can use USB4 even with Thunderbolt 3 support..
USB4 2.0 will be able to use 80Gbps twice as fast as Thunderbolt 4
Amd straight up said they will not support Thunderbolt because it's an enemy's tech(Intel). I've always had insane lens with AMDs stuff. Everything from GPU problems to motherboard problems, cpu's failing and needing to be replaced. All sorts of shit. So it's unfortunate that Intel is sucking for the past 8yrs. Still, I have always had less issues with them.
@@rileyioacura Intel recently announced a new Thunderbolt version with speeds of up to 80gbps. They haven't said if they will call it Tb5 or Tb4v2, but it's definitely on the way.
Wonderful video! Nice to see another RME fan making good work.
When I was first starting out I remember getting a second gen Scarlett 2i2 and not using it to its full potential. I would have at least 700ms of monitoring latency through Studio One and decide to just use the Scarlett's direct monitoring function.
This went on for about a whole 6 months before I started doing more research into audio engineering and understood what buffering and sampling rate were and how they worked. I properly configured it and had basically sub 100ms latency and was very happy that my 2i2 has that capability.
During this whole time I was wondering if I needed to upgrade to Thunderbolt but was always held back by the costs. Nowadays I feel like Intel and Apple just need to push it as "better" simply because they have stakes in the technology and not necessarily because "Oh, it's just better dude."
Went through a couple other interfaces, all USB, and all perfectly capable (with great emphasis on my old Audient id22, fantastic preamps) and eventually bought a Babyface Pro on sale and have been loving this thing ever since. I hardly feel like I'll ever need to upgrade (emphasis on need, I do want to upgrade to a Fireface like yours in the video) since my work consists mostly of my voice over work, sound design, light mixing and post-production, I don't work with music production so I'll never need to run 200 tracks simultaneously.
Very refreshing to see a well made video take a pragmatic approach to this topic! Congrats again!
Thanks for sharing your thoughts here, Tanekoshima! Very helpful. Glad you've found the Babyface Pro. Great interface!
This is a very important video on a topic I’ve been thinking about for months…thanks!
Glad it was helpful! Thanks for watching, Britt!
The Motu 828ES supports Thunderbolt and USB2 which is one big reason I bought it. Drivers have been rock solid so far (on Thunderbolt).
4:38 - It's not a misconception. I have two Focusrite Clarets 8pre, except one is thunderbolt 2 and the other one is USB 2. There's something in the protocol or in the way thunderbolt is wired directly into the CPU lanes, I don't know what it is - but the round trip is considerably lower via thunderbolt with these interfaces with near zero latency at 96 KHz, without problems.
Sadly, I now need to let the thunderbolt interface go, as it no longer works with my new PC with thunderbolt 4, no matter what adapter or hub I try. I'm thinking of switching to RME, but having 24 channels with pres included is going to cost around 4k or 5k with RME, compared to 2k my Clarett 8preX with two Clarett Octopres costed me. So... I'm really trying to avoid the switch, looking for solutions.
There are USB ports directly wired to the CPU too. The motherboard tells which ones are.
@@bassyey Not true. There are just USB ports on hubs with no devices connected to the other ports. No USB has or can have direct memory access like thunderbolt.
Thank you, I was surprised when I looked at the Babyface Pro performances and saw a USB 2.0 port ! Now I get it
Good choice! I myself have been a RME user for the past 17 years. I started off with the Fireface 800 and now use a Fireface 802. Rock solid with great latency performance.
Nice! Thanks for watching and sharing your experience, Roy!
Is there difference in sound 800&802? For me ff800 sounded crystal clear and my babyface pro little flatter and not so clear as 800. Now I want to change it to rack rme, cause I think their analog components are better, so thinking about 802 or ufx+. Thanks
Just M2C: I still run a RME Hammerfall 9652 in my studio in Munich.
It is "on duty" since 2002. That's what you can call quality and durability.
Your videos are awesome. I've shared your channel to many of my friends who need a professional teacher like you...
You're the best..
Thank you! I’m glad you enjoy the videos.
I have the ucx and the Babyface and it's extremely stable and fast. I record at near zero latency about 35 with regular USB 2.
Nice! Thanks for sharing, Jermaine!
USB drivers are less efficient in general due to the way USB indirectly interfaces with the CPU, and USB itself involves a fixed bus latency. WIth thunderbolt and PCIE based audio interfaces, you have a lower starting point for your round trip latency which gives you another unit of buffer size, sample rate, or plugin delay to play with before approaching perceivable monitoring delay. This can be invaluable in some situations.
Also, since you can do dynamics and EQ processing in totalmix fx with that audio interface, the interface does in fact have a DSP chip. I'm assuming you meant you just can't chuck on proprietary processors at any stage of the signal path like UAD and PTHD interfaces can.
I am very glad you mentioned the bandwidth misconception and driver efficiency though. USB 2.0 is more than enough and RME drivers absolutely slap.
I agree, the fixed bus latency is a key reason for using thunderbolt over USB. For many applications it doesn't matter that much, but for some it matters a lot.
@@crackedmagnet 100 percent, i'll reiterate what you said by saying that USB is definitely more than adequate for majority of recording situations. I believe a lot of people get conned into buying thunderbolt interfaces due to bad experience with suboptimal drivers from brands that don't care.
In what situations is that "invaluable" Jayden.... name one.
@@sorbpen live tracking into a project that was given to you with a bus processor that has delay on it but is necessary for the vibe so you don't have to bother with creating sub projects, live monitoring chains, disabling half the project, or freezing. Only to undo all of that again for the final mix. Saves you a lot of time. And as I said in the comment, it gives you one extra thing to play with. You can reclaim some of that latency with the minimum buffer setting that you'd have by working on 48khz instead of 96khz, which is more CPU efficient. Higher number of plugins overall regardless of if they introduce latency or not. Stability if you use your daw in live music to run lights and Fx chains, monitoring and foh. Etc
@@jaydensydes3478 I fail to see where in any of those use cases the difference between RME 2.0 USB drivers, and thunderbolt latency would make any difference at all, except possibly running lights, but any DMX system worth it's money will let you offset so you can sync with whatever audio you are feeding to your audience and or live performers.
Your guitarist is used to at least 12-25ms delay just from standing a few feet from his amp, etc etc.
So i still think the word "invaluable" is a bit strong here.
USB can poll up to 1000hz, i don't know what poling rate the RME drivers or other interfaces are running at, someone might be able to chime in, if it was say 125hz like early USB devices sure then we can talk.
I was considering getting a PCI-e card from RME at one point, but decided on picking up a Fireface 400 because i came to the same conclusion as this video the latency difference would be neigh unnoticeable if i went looking for it and for my use-cases completely moot, that interface is what 15 years old now?
I got it because it had firewire. My reasoning was that since I'm a bedroom producer, and programmer i kinda need to use my computer for other things as well. And in my experience the quality of USB drivers from different manufacturers of peripherals are of varying quality to put it kindly. So keeping it on a separate bus sounded like a good idea.
Now please correct me if I'm wrong but the use case for thunderbolt that i see would be massive interfaces with hundreds of I/O, and to my knowledge there are very few of those on the market
Excellent video. As an RME owner in the past and a potential UFX+ customer I am completely sold on RME products. What is missing from the video are any numbers on the round trip latency in your setup between USB 2-3 or Thunderbolt. I find that omission troubling,. However latency as a feature of products lacks an accepted standard so I get why it may be missing. The only thing I find missing in the current RME approach is DANTE. Again I understand DANTE licensing adds $300ish dollars to your product and their approval process my require revealing more than RME wants to reveal. Love it or not DANTE has become an internationally accepted process with lots of "Big Guy" adherents so it belongs on a piece like a UFX+ with it's otherwise swiss army knife approach to sound recording. Anyway, it's a great video. Thank you for posting it.
They just released a new dante product itz like 2 grand
How does it handle latency from iZotope plugins active during record?
That’s dependent on your computer.
Great overview and explanation regarding the bandwidth and utility USB 2 still offers. Been an RME fan for quite a few years now, and a quite satisfied user of the Fireface UC and Babyface.
I knew most of that, but loved hearing it all again! Thank you!
1:34 USB 4 provides 40Gb/s, not 20Gb/s. The version of USB that provides 20Gb/s is USB 3.2 Gen2x2. (I know, the naming scheme is terrible. It hurts me too.)
Thanks for the correction!
One reason to consider an Apollo's interface for latency is for the loopback. If you monitor your mic through loopback (obs streaming), there is noticeable latency.
RME has loopback and Totalmix is AMAZING. Apollo is sometimes terrible on PC.
@@Mr.Plutonium I'm not sure if it's better than Apollo's considering it's thunder bolt.
@@lilsafmusic Thunderbolt has nothing to do with quality, just quantity. With good drivers, USB 2.0 will deliver no difference.
So… to connect RME UFX II with usb 2.0 to my m1 mac studio which cable I need use? Usb 2.0 to Usb-c? Thank You
You’ll use a USB-B to USC-C cable. It is a USB 2.0 connection.
Clear, informative, and not condescending. I've subscribed, I think I can learn a lot from you.
The RME interfaces with EQ, dynamics and FX, do have a DSP chip onboard. It has even a DSP meter in TotalMix FX. The UCX2 as well.
It's a shame the dsp in babyface is stripped back to eq and reverb, no Dynamics at all :(
@@OfficialLemnisc8 That’s because of the power consumption from a bigger dsp. It would not be able to be bus powered anymore.
@@JoBu133 oh I know the why, it still sucks though. It can accept dc power, why not just disable the dynamics when bus powered, enable them when dc powered instead of removed altogether. Just figured that would have been better. I mean, I still went and bought a babyface last week anyway so I'm gonna enjoy it regardless :)
This is a great video however, there are a few things that are unfortunately incorrect, so let me kindly point them out :
1) The bandwidth shown here are the *theorical* max bandwidths and not the real-world transfer speeds bu the bulk packet *size*.
In real world use, bandwidth is occupied by communications with the host device (your computer) as well as being affected if the cable provides both power and data connection (as in all bus-powered interfaces) because then a significant portion of the conductive materials in the cable are utilized to transfer the 5V necessary to power the interface. In real-world, a Thunderbold 3 device of top performance will give you around 2.3 to 2.5 gigs per second for data transfer rates.
SInce we are all nerds here is the real explanation
"USB 2 uses 1 millisecond frames, and in High Speed (480 Mb/s) mode they are divided into 8 micro-frames. The maximum size of bulk packets (used by USB mass storage devices) is 512 bytes. According to this very informative document the theoretical maximum is 13 packets per microframe. So the theoretical maximum speed of a USB 2 drive is:
1000 * 8 * 512 * 13 = 53248000 ~= 53 MB/s"
Therefore, 46MB/s is dangerously close to the theorical max speed os USB 2.0, so why not use a 3.0 bus which is super-cheap anwway? That's quite weird to me. Kind of like they want to prove a point and their drivers are amazing enough to keep up but still, if you don't have a really perfect connector on your computer's side, you will run into issues if you full that interface up.
2) USB 3.0 gives you 4.8 Gb/s; USB 3.1 gen 1 gives 5Gb/s bulk packet max size (the difference is marketing really), USB 3.1 gen 2 gives you 10 Gb/s max size; Thunderbolt 1 gives you the same 10Gb/s size (*not* speed). Thunderbolt 2 (usually uses MiniDisplay port, although USB-C ports can also be used in certain drive enclosures) has a bulk packet max size of 20Gb/s. Thunderbolt 3 gives you 40gb/s max size, and USB4 *also* give you 40gb/s max size.
This is because USB and Thunderbolt are competing protocols, but USB is universal, while Thunderbolt is proprietary and belongs to Apple and Intel. Therefore it takes less time for them to produce it because it does not have to go through 2 bazillions tests like a universal protocol has to. Then they can sell TB ports to other manufacturers.
In reality USB ports are a much better value for money. They sound more generic; USB 4 is the same as TB 4. It just came out a little bit after. Thunderbolt is 100% marketing, and in fact, is inferior since it's not tested as rigorously nor is it available on most devices.
In practice, USB 4 and TB 3 are the exact same, TB 2 and USB 3.1 gen 2 are also the exact same, and so on.
I know, the nomenclature is really confusing.
Thunderbolt direct to CPU because use PCIe, USB need passing the chipset first. Thunderbolt better latency because short chain to CPU rather than USB
An RME driver for Linux then ???
Great video! If you are on AMD you can experience troubles with thunderbolt.
RME is great interfaces but effects are crappy in comparison with UAD.
I am using AMD. I wonder if that contributes to my problems with Thunderbolt on my PC. Thanks.
The RME effects work well for tracking. I agree that UA plugins are premium quality, but they are also much more expensive and processor-intensive. So it’s not really a one-to-one comparison.
One thing that I think you haven't touched on, especially when it comes to different versions of USB, is power delivery. Some ADC/audio interfaces come with built-in processing (Revelator io series, Apollo, Antelope come to mind) which require more power. The different preamps and phantom power supplies also demand different types of power. USB 2.0 can only do around 2.5 w (@ 5V 500mA), while USB 3.0/3.1 can do 4.5W (@ 5V 900mA). Thunderbolt on the other hand, can do around 9.9W (@ 18V 550mA). Something to keep in mind
I really appreciate the info in this video. Thank you!
Also the comments from @Mtaalas below add some additional info that is really important and needed to round out the conversation.
I run a 7.1.2 Atmos system via thunderbolt to my Mac book pro but I have used my interface for 21 simultaneous mics being recorded in large concert recording. My interface is also Dante enabled which is how we passed 21 channels from the stage to front of house position simply over ethernet cable.
It really does just come down to what your needs are and getting a device that meets those and possibly provides a little room for occasional sessions needing larger simultaneous input
My next interface will be the focusrite dante enabled red 16line. It’s thunderbolt. I don’t know if there are any usb interfaces that offer dante compatibility at that high of I/O
Check the RME Digiface Dante
Dante is an amazing technology! Nice!
My Apollo X4 is thunderbolt 3/4 only. I love the plugins. When i switched from an Apollo Twin USB to the X4, the difference in my audio was clearly noticable. There may be other things going on but i love the sound difference.
If anyone tells you to DM them on telegram in the comment section in TH-cam, they are scammers. Pay attention and let them know you know they are scammers
"Telegram Spam, Telegram Spam.."
'Telegram SAM!'
"No, its ok I sent him an email"
What he says @Dane Reid
The RME has DC coupled inputs? Know the outputs are DC coupled. Thx for the explanation
Top audio interface! I’ve been using MKI for a few years, and it’s great. Been eyeing this MKII with on-board recording and improved on-device controls among other things.
Lol, it's basically material from the RME channel about USB...
Excellent tutorial on some of the nuts and bolts behind the scenes. I’ve never been happier since I realized these trivialities have nothing to do with making music, and are only convenient distractions from productivity.
yes sir! finally someone nailed it ... except RME with its own video! this thunderbolt advertising has been driving people nuts for years now. running a RME Fireface UFX II via USB 2 by the way :-) roundtrip latency much under 3 ms if I want to :-) nice job dude! love from germany!
Can you connect the Rme Thunderbolt 2.0 or usb to the new Mac studios that are Thunderbolt 4.0. Thanks
Yes. I have had great results using my Babyface Pro FS with my MacBook Pro.
Seems like there was a key point missed on thunderbolt.
Thunderbolt is a dedicated port, where USB is shared with multiple ports on your system, it can be congested if the rear of your machine is "streaming gamer ready" with 30 different devices just to make a single light swirl colors.
I used to use a Baby Face FS and the drivers are fantastic. The latency is imperceptible as well and tons of headroom along with zero noise. My thunderbolt Apollo provides all of these things, minus the superior drivers but, allows for live tracking with plugins. RME is absolutely top notch and I will definitely be picking up a converter of some sort in the future if my setup and work flow expands. Thanks for the video!
Thanks for sharing your thoughts, Akash! Glad you liked the video!
I traded in my Apollo X4 for RME Babyface. Not having to use a power source makes the Babyface a better choice for me. To much bloat with The Apollo.
@@MaximusWhyman where did you make this trade? I'm looking to offload my X4 as well?
@@v3zMedia Long & Mquade in Canada.
@@MaximusWhyman Aww dang I'm in the U.S. we don’t have many places that do good trades. They'll try to give me $700 for it at best in a store which ain't happening. 😆
I see that TH-cam plaque! Congrats, man. Another great video. Keep it up.
Thanks, Michael!
So how do you connect the RME to the MacBook, considering that it only has Thunderbolt with a physical usb-c input?
The Thunderbolt ports on new MacBook Pros work with USB-C Devices.
@@AudioUniversity The audio interface you are showing seems to have only USB-A ports. Is there something missing from your video?
Thanks for the Infos ..! How do you connect the the interfacce USB2 to the Mac M1 USB-C Port...? USB hub..or adapter ? Any side effects on the interfacce performances or power issues?
I would really like to know how performance changes from driver to usb class compliant.
Thunderbolt has always be rock solid for me on MOTU. Always going to use a Mac for DAW so it just makes sense to stick with Thunderbolt.
Good call! Because Thunderbolt is far superior to USB, because it doesn’t tax the CPU.
I still ask the same question
How to connect 2 to 3/4 thunderbolt
Apollo twin duo 😢
Great tutorial .my set up in addition to the scarlet audio interface 6i6 gen 1 uses an outboard mixer for monitoring so I don't have the latency issues as I avoid monitoring from the DAW/interface
I hear you, but theres times I'd sure LIKE to be able to monitor thru the box - I'm sure the RME's a perfectly nice box, but that cute new socket's just TOO tempting,...
Latency on effected audio is the reason I still use an old Line 6 UX2. It is one of the cheapest interfaces with onboard effects, so I can plug in my guitar directly and use the Pod Farm software to load up a nice tone without needing to open my DAW at all. This way, I can record a clean DI and still hear whatever tone I want to hear and play accordingly. Best of both worlds. I firmly believe onboard FX is the way to go. They don't need to be great, just good enough to get you through the recording process. That said, I still use Pod Farm on many projects, so even so called outdated software still has its place.
That’s a solid ass card
I've been thinking about the bandwith thing with USB2 for years. 🤔 Thanks for the to-the-point illustrations! 🙌 And you also got me interested in RME products 😁
- Eero
M1 ALL THE WAY... RME thanks mate. 😊 x
I've got the motu Ultralite avb, and since I record classical music I don't need very low latency. should I need it I can use direct monitoring through the motu's mixer with zero latency
Nice. Thanks for sharing, Matthias!
I have a very similar approach for Classical and small Jazz, except I have a MOTU 8pre, which does everything I ask and just begs for more!
@@leumas75 great stuff! I've got a RME 4pre when I need additional mics
I'm not sure if its my Focusrite interfaces or USB but I've had 2 interfaces with USB and they occasionally both lose sync and sound like crap, and I have to reconnect them to get them to sync. Never had that issue with Firewire.
Hmm... I haven't had that issue. Thanks for sharing though, lastdaysguitar.
@@AudioUniversity I hope you never do, I know of others that have not hit this issue, but I've hit it with 2 different PC's and 2 different DAW HW - both Focusrite Scarletts... that might be the weak link but I just don't know.
I have had issues with my Focusrite Scarlett 4i4 for a while. Those issues stopped happening after I followed Focusrite advice and stopped using the interface connected to an USB hub (regardless or they are powered). Instead, it's now directly connected to the laptop's USB root hub.
@@NexuJin I'm directly connected to the PC, no hub but yes that is good advice.
Can you please show what issues might arise when using Thunderbolt? I though any Thunderbolt 3 device would automatically fallback to USB 3, when the host does not support Thunderbolt? What is the actual latency with minimal processing?
Great video. Went from Firewire to USB for the same reasons.
Nice. Thanks for sharing, Paul!
Interesting info. I purchased the Antelope ZenQ as I thought Thunderbolt reduces latency in comparison to USB interfaces!
You will have less latency using the Zen Q vs using just about any USB interface, including RME.
@@lorenhoward8178 How much less latency?
@@lugac For me, about 6ms at 96k with 64 buffer is acceptable for a good USB interface. You'll probably see about 3ms with the Zen Q.
@@lorenhoward8178 Here I have an i5 13600k, Scarlett Solo (3rd Gen) 32 buffer and 48k (my configuration) 5.17ms overall latency in ableton. (I don't use 96k).
But at 32 and 96k it is 3.67ms.
@@lorenhoward8178 Here I have an i5 13600k, Scarlett Solo (3rd Gen) 48k with 32 buffer (my configuration) is 5.17ms overall latency in ableton. (I don't use 96k).
But at 96k with 32 it is 3.67ms.
Just amazing. That is your explanation and reason for staying with Usb
What would you recommend for triggering superior drummer 3 live ?
Thanks, that is a comprehensive summary. However my experiences with USB for audio are quite different. For me USB is problems and usually audio errors. I have to use it since they stopped mounting FW sockets on laptops. FW was far more reliable - rock solid recording. I`ve been using MOTU UltraLite hybrid mk3. I have to buy a thunderbolt laptop and try to use FW-to-Thunder adapters to see if that works, as the current USB situation on my laptop makes the interface unreliable.
I can confirm it works perfectly fine on a modern mac with thunderbolt 3; but the easiest and cheapest way is to use two dongles (thunderbolt 3 to thunderbolt 2 then thunderbolt 2 to firewire); those are the official apple dongle and will set you less than 100 eurodollars. I have the FW only version which is officially not supported anymore but the two interfaces share the same driver. Only issue I have is I can't put the computer to sleep when the interface is on or I get a kernel panic.
its really up your workflow.
up to your budget and sessions.
That's right, camille!
As a happy UFX II user I wholeheartedly agree. It's by far the most stable, versatile, and reliable interface I've ever worked with.
Nice! Thanks, Steven. Couldn't agree more. I've been very impressed so far.
@@AudioUniversity thanks for the great video! Have you used DigiCheck much? I just set up 5.1 surround this week and I'm finding the surround meter and other tools extremely useful for calibrating and setting up routing.
@@AudioUniversity are you sure there is no dsp in FF UCXII? In the manual it says it does. If not, how does it work then?
Thanks for the video. I was pleased to come across it. Hearing for a very long time how great RME Audio interfaces are. And that I've been hearing for a long time how Great Thunderbolt is compared to most other popular connectivity. Your video was good to see, as I'm considering a USB interface upgrade from my trusted old : Muto 2408 mk3. I may even get a second hand RME.
Is there a big difference in latency and sound between connecting with lightning or connecting with USB
I have a Apollo Twin X thats connecting and working onnmy computer but it seems as if what i here in the headphones is not what getted printed in my daw. Do you think this is my computer are is it because im using a cables matter thinder 4 cable. I also used a belkin thunderbolt 3 cable and i still get a degraded sound of audio.
Hi! RME audio interfaces support Core Audio?
Hi. Thanks your highly educative and informative videos. I have a bit of a challenge with my Babyface (original) interface. It has served me well since 2010, but recently I got a used 2015 iMac and the interface just won't work with it. I can't even resell the interface because I won't get the value for it here in Lagos, Nigeria. If I sell it on eBay the cost shipping to its new owner will wipe out the money I get for it. I have searched for solution online and nothing has worked. Any advice on a workaround will be appreciated.
Great video! Thank you for sharing such informative stuff!
thank you for making this easy to understand for us noobs that are still learning and wonder where to go when upgrading...keep up the great content 👍
Glad to help! Thanks for watching, Mark!
I agree usb may be sufficient but why is it that rme hdspe aio pro is faster than their and basically any existing usb interfaces? Pcie is still faster than usb for now at least.
Yes, definitely. USB is not the fastest - the point is that it’s fast enough for many interfaces.
USB 4 is already being adopted, making thunderbolt mostly obsolete now
It will be interesting to see how it plays out!
Not just obsolete, but also confusing for consumers that aren't well versed in various standards. TB is kinda becoming the next Firewire.
Maybe for the home studio or project studio market but definitely not the Professional Commercial market. Thunderbolt is an external PCIE connection as it doesn't get any better than that given that Protools HDX has the lowest latency in the industry with PCIE. Thunderbolt gotten pretty close. The HDX cards perform around 0.7 ms. Thunderbolt is a pro level studio connection esp for large scale deployments. USB can only handle a limited number of i/O.
Yep, just like USB micro B is obsolete and we haven't seen any devices ship with one since 2014. Oh wait...
I just built a $4K PC and none of the high end motherboards featured USB4. The spec was announced in 2019. Adoption has been pathetically slow.
Is there no preamp like midas preamp in rme card?
You get lower latency and less load on the CPU when you use thunderbolt. Thunderbolt is really an external PCIE connection. Most high end professional studios use anything but USB. Thunderbolt, Digilink is pretty much the industry standard in commercial facilities, along with Dante, AVB and MADI that uses a lot i/O. USB has a limitation on i/O and performance which is why you rarely see USB ports on high end converters.
What's the limitation?
@@ivolol when was the last time you worked on an SSL nor Neve Console? Perhaps using a ton of outboard or doing live sound?
@@eman0828 That doesn't really answer the guys question ;) That said, I tend to agree- my understanding is there's a bit of unavoidable latency with the way USB works, and I'd REALLY like near-instant monitoring in the headphones, which is why I'm hanging out for Thunderbolt for my next Motherboard (come ON AMD pick it up will ya!)
(ok I'm sure RME's drivers do just great & I sure wouldn't turn my nose up at that little box.. But having had to deal with echoey lonng monitor latencies, I'd like everything as snappy as possible, thanks!
Glenn Frickers gone DANTE, but for me thats a tad overkill ;) )
You don't see USB ports on high end converters cause no wants to spend the money to write drivers for it and get tons of emails about things are not working, that is why they all have AES, ADAT. The RME PCIE AES card is standard but there is no reason not to have the RME UCX II, you get a lot of connectivity, its portable and its nice addition to any studio and if you looking to control Eurorack it has DC coupled outputs, also if you are using Midi, you will get incredible timing cause they writing the driver, they have it down, now if Burl could team up with RME that would be a game changer.
@@PylonRecordsdotcom Lol RME AES is NOT a standard. That would be Thunderbolt and AVID HDX. RME market share in the US is rather small since the UA Apollo out sells them around the world. Lynx Auora, Apogee Symphony i/O and AVID dominates the high end commercial studio market. MOTU and Focusrite network audio interfaces are also found in high facilities.
as i have booked ucx 2 please make a walk through video of it.. 🙏
It's very straightforward to setup! RME has some good resources on their website if you need guidance on setup!
Is there any advantages for RME Fireface UFX III (USB 3) compare it to RME Fireface UFX II (USB 2)?
this is why I love this channel, informative, scientific, easy to understand, detailed, and helpful. Thanks, Kyle.
Glad you enjoy it, Dewa! Thanks again!
Can I connect Apollo tb3 with windows usb c ?
The reason why USB3, Thunderbolt and Firewire are inherently better than USB2 for Audio interfaces is not bandwidth. Its because USB2 uses polling to get data. Thats means that the device can only sent data when requested, which is pure garbage for real-time applications like audio. Its also the reason why people kept using PS/2 keyboards for ages. USB3 also doesnt do that. Can you make it work for audio? Yeah, if you work on it for 20 years and spend billions you can make everything work. RME has these impressive figures because they 100% do the USB stack themselves in their interfaces. Most other manufacturers use readily available interface chips.
I got steinberg's mr816 csx on firewire and texas instruments pci-e card years ago and it worked great on win7.
after I switched for win10 and upgraded drivers I got annoying pop outs and all of sound nightmares.
was digging net to find solution, tried everything. nothing worked.
some time ago I browsed all files of that driver and found stand alone file in utility folder. it brings up the window with choice of ieee 1394 buffer size. was set on small. I set it on large and all problems were gone. no info about that anywhere so this is last time I go for steinberg interface.
It still seems strange for me why it all worked well on small buffer with win 7. Im afraid of going win11 because of close to none driver support from steinberg.
Hi, I'm interested in your opinion. I use Steinberg UR28 audio interface. Will I improve the sound with the RME Fireface UCX II. I use plugins, vst instruments and record keyboards and guitars as needed. Tnx
What about using an adapter? Thunderbolt to usb?
Very well explained. Everyone should watch this so this thunderbolt nonsense (for audio) stops
I'll probably stick with my PCIe card as long as I can, though I notice that less places are selling them. I've been using an RME HDSPe AIO Pro for about a decade now and it's been very reliable and capable of low-latency when necessary, probably lower than the lowest possible latency of USB.
When measured, nearly all PCIe card suck in latency compared to asio usb.
You're lucky to have one of the few expection with the model you have.
RME drivers are on a different level.
@@hypnoz7871please elaborate
I still have a firewire interface. I upgraded my SiiG Pcie card lately for a newer one with better TI chip, and now I can go down to 2ms latency 96khz/64samples buffer on win11 with reaper.
If this is true why do thunderbolt audio interfaces achieve lower latency figures than usb? RME's thunderbolt capable interfaces and presonus 2626 all come in a few ms lower in terms of round trip latency. There must be a reason for it. The 2626 can achieve around 2ms RTL at 32 buffer and the UFX+ around 2.5ms. RMEs usb interfaces seem to achieve more like 5-6ms at the same bugger size.
The wher's the trick with thunderbolt ? 🤔
Awesome video. I never knew that the transfer speeds were the same
Thunderbolt has faster transfer speeds with more bandwidth. It just matters less for small channel counts of audio.
Perfectly explained. Thank You!
Glad you enjoyed it!
@@AudioUniversity I did! 👍
when working on a big project... 120 tracks or so. Does it matter if the interface is using usb 2.0 vs TB? I get the point about the physical IO, but what about playing larger projects? Thanks!
Great question.
The number of channels going between your interface and computer is the main consideration. That could mean recording, playback, outboard inserts, or even using the interfaces built-in DSP as an insert to the DAW.
The number of tracks in the DAW session will be dependent on the DAW, RAM, CPU, etc. Not the connection to the interface.
When you say ‘switching back to USB’, what thunderbolt interface did you own? I have a UCX mk1, it’s great. I bought a Thunderbolt Antelope Orion interface, and although the driver performance was not as good as the RME, the Antelope has more depth and clarity, it just sounds better.
UA Apollo Twin DUO MkII
I use the DAW as an effects processor for live performance. The Baby Face Pro FS with USB2 has a round trip latency of 22.3 ms. My Quantum 26x26 thunderbolt has a roundtrip latency of only 1.52 ms. So, I don't see the reason in not using the thunderbolt interface. The only problem many people may run into is the thunderbolt cable. You MUST use an ACTIVE thunderbolt cable. They are usually pricey, from $60 and above. I got my 2 meter 'active' thunderbolt 4 cable for $23 from Amazon. There are unlimited VST plugins out there. So this solution is much better and cheaper than buying an Eventide H9000 or a Lexicon PCM92. In response to the title. I will never go back. All my USB2 devices will be sold or given away. My Baby Face Pro FS was pricey but good I thought. Until I found the Quantum 26x26, which is better and cheaper with or without a $100 active cable.
Agreed, I bought a Presonus Quantum 26x26 for $400 on eBay. Using thunderbolt 3. Why would I "switch back" to 2008 technology ? My latency problems are gone.
Kinda wish they would go to USBC just for future proofing sake. There's also none of that "oops wrong way" bs either
I have this exact interface and it is indeed a beast.
Nice! I'm loving it! The Babyface Pro FS is great, too!
@@AudioUniversity oh man I was looking at that one too for mobile and laptop solutions. So you do like that one?
I swear I did not know what I was NOT hearing until I went to RME. It's like someone took a blanket off my music.
I think the Fireface UCX II and Babyface Pro FS are excellent!
how can i fix the red> Connect UA Device on my Imac please someone should me tnaks
Always fantastic with precise explaining in your videos, you’d make a great teacher, very pragmatic!
Thanks, Paul! I appreciate that and I'm glad you like the videos!
@@AudioUniversity you’re welcome, they’re very well done and I hope sweetwater keeps hiring you to do their videos too!
I'm thinking in buying an AMD laptop whit USB 4. If I go to a studio whit an UAD thunderbolt interface. Will I have compatibly problems?
It depends on the OS and drivers that you're dealing with.
@@AudioUniversity Windows I don't like the apple ecosystem
Hello! I have a question. I'm looking to buy studio monitors and would appreciate your advice. My room is about 12ft by 10ft wide and from the floor to ceiling is about 8ft tall. The floor is wood so it would be reflective.
I'm not sure which size of monitors would be good or which brand (I compose and interested in orchestration and ensemble, so I guess songwriting and film scoring? I also enjoy worship, neo soul, lofi, gospel... so music with a band setting too... All with the goal of giving glory to Yahweh and leading people to the Messiah, the Way, the Truth and the Life). I've read a book on recording and they recommended monitors not going under 6" but I've taken a look at KRK 5" ones. Also looked at the Yamaha HS5. Any recommendations? Thank you and Yahweh bless! :)
I’ve been using the ADAM T7V and have really enjoyed them! It is just as important to consider acoustic treatment for your room!
@@AudioUniversity Thanks for the reply! What's a good rule of thumb when treating such a small room?
I found once you have a serious CPU at some point drivers and connection may be the bottleneck. I stick with thunderbolt since it's really just a pcie lane so i can monitor what's in my daw directly including processing plugins. That said, i have also have a USB 2 interface for when i only need quick audio not my whole recording rig. Cool video, you are slowly talking me into rme products too!
Thanks for your opinions. I've used many different Audio interfaces and computers over the past 25 years and from experience found USB 2.0 to be by far the most laggy and with the most dropouts. I disagree with your analysis based on many years of experience. There are many factors which determine how efficient and stable Audio interfaces are on a particular computer, and they are not just related to bandwidth over a USB port. The most efficient and stable I owned was an RME AES card plugged into a PCIe slot in a Windows 7 PC. It left USB 2.0 for dead in every respect, including latency, dropouts, overall system stability and the sheer number of tracks and plugins I could run on a moderately spec'd PC was perfect for doing large mixes. Those interfaces all use DMA which bypasses a lot of the interrupts caused by other system resources such as USB. Those other resources can drastically degrade audio performance, like having an SSD plugged into another USB port close to the one the Audio interface uses, which can drag down the bandwidth of the Audio interface quite drastically resulting in dropouts. Using MIDI over USB could also cause dropouts, if there is a lot of MIDI data over one USB 2.0 port, which happens to be on the same USB controller as the Audio interface.
Thunderbolt 3, which I use now, bypasses the USB SSD drives, and I get excellent stability and no dropouts. I can understand why you oversimplified the explanation of why USB 2.0 is good for Audio. At face value you are correct, but in reality, there are always other factors which contribute to real-time system stability. When I started using computers to record, plugins used to have a static UI, with a few sliders on them, and were very efficient in terms of graphics. Nowadays many of the top brands of plugins have complex interactive graphical interfaces (which I love), displaying all sorts of real time information. This fact makes it mandatory to have a moderately well spec'd graphics card so that you don't get lots of audio dropouts, because the graphics card takes away real time processing power from the CPU.
I discovered this the hard way when I got a Mac Mini (late 2018) which has a very decent i7 CPU but extremely low spec'd on-board graphics chip. Despite the good CPU performance, the computer was unusable for serious mixing, because the CPU spent most of its valuable clock cycles processing the UI data from the plugins I was using, and of course USB ports have lower priority than Graphics, because Graphics uses DMA, but so does Thunderbolt. The Audio interface I am using on it is an RME Babyface pro FS, which as you know uses USB. So I added a moderately powerful external GPU though TB-3, and all of a sudden, I had stable and very usable Audio without dropouts. I've since relegated that computer to the recording room and use a much higher spec'd machine for mixing (with a TB-3 interface) .
All things being equal, yes RME make amazingly good interfaces, with excellent drivers, which I would recommend to anyone without a second thought. USB 2.0 should have tons of bandwidth for multiple audio channels, and it does if you have say 50 mono tracks in a DAW without plugins, and there are no other devices attached to other USB ports. But in reality, you can't really make music unless you have some good plugins and SSD drives for your samples and backups and Audio and all your extraneous system apps have been pruned right down while you are recording. Ok, maybe that was hasty, because if you have a room full of analogue hardware you can render each track to your daw, without using any plugins at all. But how many of us have a hundred thousand dollars to get all those magical analogue boxes? I've never met a real working musician who can afford this stuff, so I guess it's plugins for most of us, and so that's why I'm making this point that just looking at USB bandwidth alone is just as impractical as looking at CPU speed. Neither of these metrics give you an idea of whether a computer can produce stable audio.
Very interesting. I've had many interfaces over the years, firewire (x 3) USB ( x 2) PCI (x 2) and my little Thunderbolt 2 Zoom interface has very similar low latency to my very old and expensive PCI MIX hardware from Pro Tools 5 (from the early 2000s G4 days). I've always had issues with latency on USB, and to a lesser extent, firewire. Thunderbolt and the very old PCI stuff wins for low latency, in my experience.
you can't beat PCI for latency certainly; and thunderbolt gives you pretty much direct PCI access so it's virtually identical. Firewire was still a bit better than USB though.
That's what I'm saying. PCIe-424 MOTU to FOUR 2408mk3 rackmounts for 32 balanced I/O analog TRS channels. 96k 24bit all day. Still killing it. Hackintosh build on a 768GB RAM twin i7 CPU server.
A nice little ad for RME they will be happy.
I had a horrible latency issue using a USB interface plugged into the USB-A ports of a 2018 MAC Mini.
The only workaround I could find was to buy a Thunderbolt dock with a USB-C port and plug my Interface into that.
I thought I remember learning from _somewhere_ that Thunderbolt was, indeed a "highway with a faster speed limit," to use your analogy. This source claimed that, due to an inefficiency with how USB performs data transfer (error checking algorithm, maybe?), there is a bottleneck when it comes to latency in USB, even with low channel counts, and such a bottleneck was not present (or at least greatly reduced) in Thunderbolt.
Anyone who knows more than me, feel free to let me know if I'm totally off-base. I'll admit that this presumption may be a desire to satiate my confirmation bias, as I dropped $2500 on a used Focusrite Red 8Pre (a Thunderbolt 2.0 interface)
what you think about a old Usb2 fireface UFX the first one, i ask becouse i try to change to fireface + but is too much expensive to me right now
If you have the first version you don't need the II or the +