Thunderbolt performance

[Synthia Cynthia Payne]

Hi Jacktrippers!

Please share your experience with using a Thunderbolt adapter for
JackTrip sessions. I really want to upgrade my Mac, but mid-2012 was the
last year that MacBook Pros had an ethernet port and I am not sure that
the Thunderbolt will be fast enough.

Thanks!

[Jon Raskin]

I’ve never had an issue with the Ethernet adapter

Sent from my iPhone

> On May 6, 2022, at 12:56 PM, Synthia Cynthia Payne <synthi...@gmail.com> wrote:
>
> Hi Jacktrippers!

>
> Please share your experience with using a Thunderbolt adapter for JackTrip sessions. I really want to upgrade my Mac, but mid-2012 was the last year that MacBook Pros had an ethernet port and I am not sure that the Thunderbolt will be fast enough.
>
> Thanks!
>

> --
> You received this message because you are subscribed to the Google Groups "jacktrip-users" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to jacktrip-user...@googlegroups.com.
> To view this discussion on the web visit https://groups.google.com/d/msgid/jacktrip-users/b2eb319d-51ae-0a30-85db-c28caf3afb12%40gmail.com.

[Marcin Pączkowski]

Most USB-Ethernet adapters would be just fine, doesn't even have to be Thunderbolt. 

The only thing I suggest is getting one that doesn't need extra drivers in macOS (check item description and/or reviews). 

Aside from the ones offered on Apple website, I know that Anker 3-Port USB 3.0 HUB with 10/100/1000 Ethernet works fine and does not need drivers (but you'd need USB-A to USB-C adapter for it).

Hope this helps!

Marcin

To view this discussion on the web visit https://groups.google.com/d/msgid/jacktrip-users/6AC66A29-87C6-4095-B46E-4B7868BB8319%40yahoo.com.

[Mike O'Connor]

hi Synth,

you mean "thunderbolt on the new Mac"? that'll be Thunderbolt 3 or 4 which is pretty darn fast. here's a pretty good comparison. Thunderbolt 3 runs at a theoretical max of 40 Gbps each way (40,000 Mbps)

https://www.lifewire.com/what-is-thunderbolt-high-speed-2260190

we Jacktrippians like it when we see internet connections that are provisioned at 5 to 10 Mbps each way so that our 1.5 Mbps stereo signals don't get goofed up. the theoretical 40,000 Mbps of a Thunderbolt 3 link will more than cover that.

mike O'C

[Synthia Cynthia Payne]

thanks mike, marcin and jon for your feedback so far. Thanks Mike for
the hard data - really helps to understand. I was also wondering about
the adapter itself and its ability to make a good connection to the
ethernet cable on the one end and the computer on the other. But I guess
you get what you pay for and spending a bit more might buy higher
quality materials and workmanship.

synth

[Steve C]

I was thinking that the question would be regarding latency rather than speed.... are there any estimates of the delay introduced when having to use an adapter, be it thunderbolt or usb?   Just curious since even a millisecond can be valuable when it comes to naving leeway to absorb jitter.

To view this discussion on the web visit https://groups.google.com/d/msgid/jacktrip-users/d92dc5ea-a79a-5f7f-1970-24c7faf81412%40gmail.com.

[Wm Leler]

Just to clarify. Steve C is absolutely correct. The biggest metric we care about is latency. Not bits per second. The second metric we care about is jitter, which is how much latency varies over time.

Furthermore, a unit like bps (bits per second, or Megabits per second or Gigabits per second) is NOT SPEED. It is BANDWIDTH. Speed, bandwidth, latency, and jitter are separate things. In particular, for playing music over the internet, we don't need gigabit fiber. Audio doesn't need a huge amount of bandwidth, it needs very low latency.

An analogy: Let's say you are standing by a highway counting the number of cars that pass by every second. You measure 1 car passes every second (1 cps). That's bandwidth. But that does not tell you the speed the cars are going, even on average. If the highway has many lanes, the cars could be traveling quite slowly and still passing at 1 cps. After all, if your commute to work takes too long, you might want your government to add a lane or two to the highway so you can get to work faster.

What we want to know is how long it takes a car to go from point A to point B. That is latency. Knowing the cps (cars per second) cannot tell us the latency, because there could be a delay somewhere (like a stoplight, or even a traffic jam). For playing music together, these delays are one of the most important things (along with distance, since the speed of light puts a limit on how fast music can go from point A to point B).

If we have an adapter in the system, it can introduce added latency, especially if it has to do any buffering of data. How much latency can vary widely, and unfortunately few manufacturers will tell you how much latency their adapters add to a signal. Mostly because for most internet traffic, latency is unimportant. If you are reading a webpage, you probably don't care if the page took 100mS (a tenth of a second) to get to you. But for two-way audio, 100mS would be terrible.

To view this discussion on the web visit https://groups.google.com/d/msgid/jacktrip-users/CABoo1GePMEP%2B4E0_TA6JAYN%2B7LKHJ%2B%2BA9GFZYBT6OSK3Da-7XQ%40mail.gmail.com.

[Kenneth Fields]

hi wm,

100mS would be terrible.

It sounds like you’re assuming that localized music making (<50ms) is somehow fundamental to music interaction itself. 

It’s actually just a habit of historical practice (pre-telephony).

we live in a networked music age now. tokyo can play perfectly well with london at 230ms delay.

Ken

To view this discussion on the web visit https://groups.google.com/d/msgid/jacktrip-users/CADDX_n7yaOea-JHhTXTR4NvQxncTPjzbCUTex7cT_5QADN3Q2g%40mail.gmail.com.

[Marcin Pączkowski]

Looping back the conversation to the technical aspect:

It's a good point about the latency and continuous performance as opposed to peak throughput.

I have not seen any noticeable difference in latency between the built-in ethernet interface and USB one. Ping on local network in both cases is on the order of fractions of millisecond and I've never seen USB ethernet behaving worse (except for one situation when a 3rd party drivers was causing USB issues with the USB audio interface connected to the same hub... that's why I suggest the devices that don't require 3rd party drivers).

Thunderbolt is essentially PCI-express on the wire, so in principle should provide the same performance as built-in interface (note that the true TB3 ethernet adapters are quite expensive though)

we live in a networked music age now. tokyo can play perfectly well with london at 230ms delay.

Statements regarding latency are very much context-dependent. Some music can be played "together" even over zoom (~500ms latency or so)... But it would be really hard for a jazz/funk rhythm section to groove well (also depending on the definition of "well") when latency between participants is above 15-20 ms (maybe even less). But again, this depends A LOT on the music, players etc and thus I think it's better to avoid blanket statements like "latency X is totally fine".

Marcin

[Wm Leler]

Latency problems definitely depend on the type of music, and other issues. When I play music, it is normally in a jam session, where we are playing off against each other. Network latency is a big problem in that case.

I would also disagree with the statement "It’s actually just a habit of historical practice (pre-telephony)."

Telephony is not the only thing that introduces latency problems. So does air! Sound in air travels in the neighborhood of 1mS per foot. Have you ever wondered why when a (largish) orchestra plays (or larger choir sings), they always have a conductor? While if the same musicians play in a smaller group, they don't need one? Because for a large group, they need a timekeeper to keep them together because of latency!

--wm

--

You received this message because you are subscribed to the Google Groups "jacktrip-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to jacktrip-user...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/jacktrip-users/CAEBNZeBU_3DMi-y43v%3DPAUn53kYjoji4Q7rOWK_iHr%2B0R25jvA%40mail.gmail.com.

[Synthia Cynthia Payne]

thanks all for this good thread! I really appreciate the extra mile to consider the speed vs. latency question. Perhaps a case study in action is needed. Thundertrippers unite!

synth

To view this discussion on the web visit https://groups.google.com/d/msgid/jacktrip-users/CADDX_n4KQVfYGTPqy%3Dui7UPoGivBrZ%2BJP1MuC0TTF4GCx2JvjA%40mail.gmail.com.

[Mike O'Connor]

Thundertrippers unite!  a great rallying cry.

does there exist an easy-to-use cross-platform gizmo to test throughput AND latency?  i'm thinking of something like GeekBench (CPU testing) or Black Magic (disk-speed testing).  

barring that, how about a simple cross-platform recipe that can be followed by folks of less-than-stellar tech ability (like me)?

mike o'connor  

To view this discussion on the web visit https://groups.google.com/d/msgid/jacktrip-users/c1abe82f-8381-58ed-e896-75c2250d4a62%40gmail.com.

[Wm Leler]

Jacktrip will tell you the roundtrip latency. But it varies wildly!

Also, you can do a ping on most operating systems to a person you are connected to, and that is an easy (but somewhat inaccurate) measurement. See https://shapeshed.com/unix-ping/

The speed of light is only one of the contributors to latency, so distance is not the only factor.

Buffering is often a bigger problem. Buffering (saving up data so you can send it in a packet) is used when there is a wireless connection, and (even worse) when a signal goes from one internet provider to another (like if one person is on Comcast, and another is on Google Fiber, there is a gateway between them that can cause a traffic jam). Even if everyone is using the same provider, there can be congestion, including other people in your neighborhood using the same circuits.

--wm

To view this discussion on the web visit https://groups.google.com/d/msgid/jacktrip-users/9E842D02-E6DB-42B3-AB79-EC6CAC2FA8AD%40gmail.com.

[Carlos Aguayo]

I like the cloudflare speed test, which includes bandwidth, latency, and jitter reports, but it's not end-to-end:

   https://speed.cloudflare.com/

For end-to-end measurement, I like iperf3.  Your mileage may vary:

   https://en.wikipedia.org/wiki/Iperf

In my experience, Comcast is the worst for jitter, Centurylink is great over both fiber and "bonded copper" DSL.

Carlos

On Thu, May 12, 2022 at 12:52 PM Mike O'Connor <oconn...@gmail.com> wrote:

To view this discussion on the web visit https://groups.google.com/d/msgid/jacktrip-users/9E842D02-E6DB-42B3-AB79-EC6CAC2FA8AD%40gmail.com.

[Wm Leler]

Many people use https://speedtest.net to measure the throughput of their connection to a nearby major server. But that won't include any gateways between companies, or how long it takes your signal to get to the person with whom you are communicating. It is also sad that their name includes "speed" when of course they aren't measuring speed (latency) at all.