Hey there :-)

This is a very tough one and so far, I've ruled out that this must be a Windows 10/11 thing:

When I'm sending NDI from vMix (or NDI Screen Capture as well) by Multicast to a few clients everything look ok at first sight. As soon as I'm reducing the speed of network interface of one of the receivers (Birddog Play) all other receivers become laggy, like the one with the reduced NIC speed. It seems like a kind of flow control's going on but AFAIK there's no such feature for UDP (NDI is sent as UDP when it comes to multicast), so I'm pointing at Windows atm.

To sort a few things out:

Q: Why would you use reduced NIC Speed on the receiver?
A: I found this out by accident, there was a faulty ethernet cable (speed was reduced to 100MBit/) as the receivers acting weirdly
(knowing the fact, that UDP shouldn't care if a receiver's got the packets) I start investigating by simulating the faulty ethernet cable.

Q: Do you have general Multicast issues in your network?
A: No, as soon as the Multicast sender is not a windows machine, everything works as expected.

Q: Did you try a different machine?
A: Yes, they all beave the same:
M1 Windows 10 Pro, Intel Core i7 10750H, Nvidia RTX2070, 10 GBE NIC (Marvel)
M1 Windows 11 Enterprise, Intel Core i7 10750H, Nvidia RTX2070, 10 GBE NIC (Marvel) (same machine but clean install with just vMix on it)
M2 Windows 11 Pro, AMD Ryzen 9 7950X3D, Nvidia RTX4090, 2,5 GBE NIC (Intel)
M3 Windows 11 Pro, Intel Core i7 12700HX, Nvidia RTX 3050, 2,5 GBE NIC (Realtek)

Me and (maybe) some others would be so glad to find a solution to this one. Hopefully, you can help!

Thanks, Denny

Hi Mavik,

sorry, I've forgot to mention that I've turned on Multicast via NDI access manager on the test machines. As soon as I disable it, NDI is transmitted as unicast for sure and the everything runs smooth.

Meanwhile I spent weeks now in researching and also tried various things:

- Switches (unmanaged, L2 Managed, L3 Managed)
- Different Gateways (Router)
- Different NIC (Realtek 2.5G, Intel 2.5G, Marvell 10G
- Different Networks (IP Ranges)
- and as Mentioned before: Machines and OSes



You're right but: Multicast means that it's UDP!
In the fact that UDP is a stateless protocol it makes no sense to me that the decoders influencing each other, except windows's doing some extra magic.

I bet 10 Bucks that you'll able to verify the same issue:

- get 2 Hardware NDI Decoder (set to Multicast Receiver)
- Set vMix Machine to Multicast by NDI Access Manager (restart vMix after changing)
- Both NDI Decoder subscribing to the same NDI source.
- Now set one Switch Port of one of the Decoders to 100 MBit/s and see what happens.

I've IGMP Snooping configured properly.

I have 2 Zyxel XMG1915-18EP.
One of them is acting as Querier.
In NDI Access Manger I set the Broadcast Group to 239.255.171/24.... As soon as the Decoders got their signals, I see the Multicast addresses in the switch's table accordingly to the ports where the decoders / The vMix Machine (sender) are

What's weird as well: Even with unicast I'm getting some dropped frames on my Birddog Minis when set to UDP as receiving method (tcp is working fine). That gave me the idea that there's a problem in general with UDP sending and Windows (Hardware decoding a multicasting hardware encoder works like a charm).

Hi Mavik,

First of all: I really appreciate that you're trying to help! But...










...You would safe us a lot of time if you had read my post more carefully ;-)




Imagine you have a Multicast Environment with 15 receivers spread over a whole building and you're sending Multicast to it:...
The symptoms I've observed, means that the whole installation would fail at the moment when just one single decoder has a faulty LAN cable connection (whether it's a decreased Speed or CRC Errors). If this is not worth investigating what is it then?