Avid Carbon

[frazzz]

Hi,

Just been looking at the support documents for the Avid Pro Tools Carbon with the expanded systems.

I've glanced over these but it seemed odd to read that the Mac Mini with the Coffee Lake processor 6C|12T - Intel CPU is supported but the Intel iMac 2020 with 8 & 10 Core Intel i9 is not supported for Carbon expanded nor is the iMac Pro with the Intel CPU's.

Years ago someone mentioned there is a big difference sometimes between not supported and doesn't work! - Meaning something may not be supported but does work!

Anyhow, maybe someone can chime in to explain the Mac Mini 2018 with Coffee Lake CPU Intel - Is supported by Avid for Carbon expanded but the other mentioned above are not - I'll double check this - Thanks

[frazzz]

One reason could be the Apple Ethernet to Gigabit adapter and 1 giga-bit LAN is twice USB 2 for the sake of argument, so a 2.5 Giga-bit LAN should work maybe?

USB 2 is good for 24 I/O simultaneous (accepted norm)
1 Gig LAN 48__for_arguments_sake
2.5 g LAN 120_I/O_simultaneous

What do you think? - This is my exptrapolation

[dominicperry]

Extrapolation won't help here - it's not the speed/bandwidth that's the problem, it's AVB compatibility for the Ethernet chipset.

Having said that, I still can't help you with whether it will work or not. But you can always use a TB port and a double-dongle (Apple only TB3 on USB-C to TB2 -> Apple only TB2 to Ethernet) or the Sonnet adapter.

I'm not clear from your post which machine you actually have, or what ports are on it.

Dominic

[nst7]

I've used the 2020 Imac I7 with the Carbon and it worked just fine. The ethernet connection on the I9 will be exactly the same. The only upgrade choice for ethernet was 10gb vs normal, but the chipset is the same.

EDIT: just saw you're talking about expanded systems, but I don't know why it would be different.

[frazzz]

Originally Posted by nst7

I've used the 2020 Imac I7 with the Carbon and it worked just fine. The ethernet connection on the I9 will be exactly the same. The only upgrade choice for ethernet was 10gb vs normal, but the chipset is the same.

EDIT: just saw you're talking about expanded systems, but I don't know why it would be different.

Well, a 1GbE = 1 Gigabit (USB 2 = 480 Mega bits per second =24 CH I/O
capable.

An expanded system would be 3 X the number of I/O and would exceed 48 channels so in my calculation could be touch and go as to whether there is enough bandwidth in the 1 GbE which is 2 X USB2 + just a little extra.

The next one up is a 2.5 GbE which is 2.5 times more bandwidth but half of USB 3 bandwidth - But still more than sufficient to use ALL I/O on an expanded Carbon system.

Apple 10GbE uses an AQUANTIA LAN chipset off the top of my head - Bandwidth wise is way more than what is needed but welcome IMO!

So from my point of view 1GbE is low on bandwidth for ALL I/O of a carbon system being used simultaneously - I've built computers and know some of the important numbers which help me ascertain certain things - Just to cover bases! and all to minimise or eliminate problems if possible.

There may be other considerations too, such as which LAN chipsets will work and won't work - Like with the Firewire cards of old - Texas Instruments (TI) was always advised.

I have a Mac Mini with 10 GbE LAN so that is good.

Some other interfaces use thunderbolt 2 (20 Gbs) loads of bandwidth - and TH3 = 40 GbS, say no more.

So, a fully expanded Carbon system would be doing 72 I/O which is more than what 1 Gigabit LAN will be able to deal with (possibly) - From my point of view from the numbers - I don't like how it looks but that's my call which is why I mentioned a 2.5 GbE - Would cover ALL I/O of an expanded Carbon System from a bandwidth point of view in my humble opinion.

USB2 interfaces are good for 24 I/O simultaneously - Other people and users have mentioned this on other forums over the years.

I don't wish to get into spats with people who don't agree! - Peace !

[BScout]

Originally Posted by frazzz

Well, a 1GbE = 1 Gigabit (USB 2 = 480 Mega bits per second =24 CH I/O
capable.

Your assumptions are wrong.

Like Dante or Ravenna, 1Gbe can handle 64 channels i/o easily. Also, AVB (including Avid's use of it) does not utilize anything over 1Gbe (so no 2.5Gbe and no 10Gbe.) The negotiated speed (even if using NIC with higher multi gigabit speeds possiblities) is always 1Gbe. It's just not in the standards (and why Avid does not demand 10Gbe or anything over 1Gbe in compatibility docs.)

(Also, as proved multiple times by RME, USB2 can handled 64 channels both in and out just fine. They have multiple interfaces that do just that. So that's wrong too.)

[Darryl Ramm]

Wait, what? This silliness needs to stop.

Carbon was designed by the Avid live audio folks, experts in AVB technology and systems, do you really think they would get stuff wrong in the design of basic network capacity?

The Carbon has a single 1 Gbit AVB based connection. It supports only direct connection to a host computer. In doing so it does not matter if the host has 1 Gbit, 2.5 Gbit, 10 Gbit, or whatever, the Ethernet port you connect a Carbon to will run at 1 Gbit. It can't do anything else, that's how Ethernet works. For all the expertise you are pointing out you have here I'd expect you to understand this.

And lets try some basic math for a maxed out Carbon system:

Carbon:
8 mic inputs
8 line inputs
8 line outputs
4 x ADAT outputs @ 196 kHz
4 x ADAT inputs @ 196 kHz

(ADAT is 24 bit at most but it likely pads to 32-bit float in the Carbon for transport over AVB, so assume 32-bit for worse case)

Carbon Pre 1:
8 mic inputs
8 line inputs
8 line outputs

Carbon Pre 2:
8 mic inputs
8 line inputs
8 line outputs

Ethernet is bidirectional so you have a theoretical bandwidth of 1Gbit/s in each direction.
So worse case at 192kHz:

AVB output from Carbon:
3 * 16 + 4 = 52 channels @ 196 kHz 32-bit float
= 52 * 196 kHz * 32 bit / 1024
= 326 Mbit/s

AVB input from Carbon:
3 * 8 + 4 = 28 channels @ 196 kHz 32-bit float
= 24 * 196 kHz * 32 bit / 1024
= 176 Mbit/s

This will likely be all the channels each way combined in one or a few AVB streams with relatively low overhead compared to seperate stream per channel. Pick an overhead 10%, 20% it just does not matter, lets assume 20% that gives

Carbon AVB output bandwidth: 326 * 1.2 / 1000 = 40% utilization

Carbon AVB input bandwidth: 176 * 1.2 / 1000 = 15% utilization.

So no, there should be no concern here about saturating the Ethernet.

I'll stay away from more precise estimating the AVB overhead because I don't know the exact stream structures and other info necessary, or know what magic changes Avid has made to AVB here, but it won't really change anything here. And again, the folks who designed this are experts in AVB, they are not going to get anything so basic wrong.

And yes Carbon is speced to use point to point connections with no other traffic on the link.

---

You really are flooding DUC with questions all over the place. And now silliness like these wrong over confident assertions. It's unclear if you are just spending your time here trolling asking question after question in thread after thread.

[frazzz]

Hi - I wouldn't say trolling - But forums are for asking questions to get answers to things - Thanks for pointing items out where you think there are incorrect assertions - Thanks

[frazzz]

Please don't have a go at me - I think some of my ideas are not as bad as you think.

I asserted the use of all I/O - 72 channels I/O including ADAT I/O but they are not ALL useable simultaneously as found in this forum link so bandwidth "MUST" play a roll in the fact that in a Carbon expanded system not ALL I/O 72 are available for use.

My assertion here in not correct - as Avid don't allow this

Please see this forum post from someone else

http://duc.avid.com/showthread.php?t=423415

I have no problem being incorrect about some things - I'm only human!

[dominicperry]

Originally Posted by frazzz

Please don't have a go at me - I think some of my ideas are not as bad as you think.

I asserted the use of all I/O - 72 channels I/O including ADAT I/O but they are not ALL useable simultaneously as found in this forum link so bandwidth "MUST" play a roll in the fact that in a Carbon expanded system not ALL I/O 72 are available for use.

My assertion here in not correct - as Avid don't allow this

Please see this forum post from someone else

http://duc.avid.com/showthread.php?t=423415

I have no problem being incorrect about some things - I'm only human!

Someone on that thread stated that Mark Corbin had told them "Max sample rate would be 96k due to some limitations."
It doesn't say they are bandwidth limitations.

The maths was done in that thread, and again in this one. The Carbon has a 1Gb/s port on it. So there's nothing that a 1Gb/s port on a Mac can't do that a 2.5Gb/s or 10Gb/s port would. You are simply mistaken about that element. If there's a limitation for your machine, it's not the speed of the Ethernet port. You still haven't said which machine you have.

Your starting point of USB - 24 i/o is just nonsense.

You are correct that Avid seem to have designated certain machines as not compatible with Carbon Expanded, but that they will run with Carbon alone. No one here knows why. You could try emailing Avid or PM Mark Corbin who is the product manager. They don't come on these forums very often.

Dominic