• Registered: 2021-04-03
  • Posts: 3

Topic: HDSPe AIO vs RayDat vs Madi chipset generations

So the old AIO uses an older chipset than the RayDat, but I'm having a bunch of trouble determining what generations the various MADI offerings are from.  For my usecase the AIO is less than optimal and has higher internal latency, where as the RayDat will probably work for my usecase.  The AIO Pro is also a possibility.  But I can't determine where the generations change over for Madi.  I'm basically trying to avoid any Madi offerings chipsets that are older than the RayDat, comparible to the AIO.  I also get the impression that the AIO Pro is a newer generation than the Raydat\AES. 

I get the impression the Madi FX is based off the AIO Pro chipsets and the HDSPe Madi is based off a similar chipset to the Raydat, where as the Hdspe AIO is based off an older Hdsp chipset.

Sorry for such a clumbsy question, I don't know how to word this better since this discusses stuff that is sort of beneith the marketing of the cards.  I only ask this question because I didn't realize the Hdspe AIO has a much higher internal latency than other offerings in the line which made me question what I'm getting into here.

  • ramses
  • DAW Mastermind
  • Online
  • From: UTC+1
  • Registered: 2012-10-14
  • Posts: 13,267

Re: HDSPe AIO vs RayDat vs Madi chipset generations

HDSPe AIO and HDSPe RayDAT have nothing to do with MADI. Was it late typing this? wink

In the "digital domain", inside of the card you have extremely low latencies, nothing to worry about.
Same if you connect cards via ADAT, SPDIF, AES or MADI.

You have more latency when it comes to A/D and D/A conversion. There it might make a little difference if you are counting beans (need lowest latency for playing virtual instruments or if you have special demands for your monitoring).

Higher latencies you have, usually only through the connection to the computer. There, ASIO buffer sizes make a significant difference to the total RTL (round trip latency - A/D to computer/DAW/and back  - D/A / monitoring.

You should be clear on this, what's important for you and your use cases.

No matter which transport technology to the computer, we talk of. Whether it is Firewire, USB, TB, PCI, PCIe. The RME drivers more or less on par, deliver the same excellent performance. There is not much difference, look here from my blog space: https://www.tonstudio-forum.de/attachme … es-v2-jpg/

I would concentrate on how many and what type of I/O ports you need and would avoid stacking of recording interfaces as you can not route channels between recording interfaces without sacrificing one ADAT connection to build a crosslink between interfaces. Better to have one MADI setup where any Input and SW playback channel can be routed directly to any output.

BR Ramses - UFX III, 12Mic, XTC, ADI-2 Pro FS R BE, RayDAT, X10SRi-F, E5-1680v4, Win10Pro22H2, Cub14
  • Registered: 2021-04-03
  • Posts: 3

Re: HDSPe AIO vs RayDat vs Madi chipset generations

Sorry this is a deliberate non-answer of my question.  I am trying to obtain sub 2ms RTLs and can't do that with AIO Generation cards.  What I am trying to understand is where these card generations change over since it is not clear in the marketing.  The RAYDAT is capable of this, and I am curious what chipset generation the various MADI offerings are.  I actually get the impression that the MADI offerings are a generation ahead of the RAYDAT.

So my interests might be perfectly served with an older MADI card or better served with a newer MADI card vs RAYDAT.  My needs are somewhat maliable and I can use either card, but I'm trying to decide which is the best solution which I can only do with transpency and knowledge.  It's not ideal to be bench testing these things myself.

What I can say for certain is purchasing an AIO card would have been a mistake based on selecting strictly on I\O needs because 4ms base latency is too high for my intentended usecase.

4 Reply by ramses (edited by ramses 2023-02-14 00:03:53)

  • ramses
  • DAW Mastermind
  • Online
  • From: UTC+1
  • Registered: 2012-10-14
  • Posts: 13,267

Re: HDSPe AIO vs RayDat vs Madi chipset generations

Sorry, but I can only answer questions that make sense and that can be answered.

The concrete requirement sub 2ms RTL you delivered later. IMHO, this has nothing to do with the card itself.
This I tried to explain to you, that internal digital processing inside the card (in the FPGA) has the lowest latency, also digital transport over digital transports like ADAT or AES.

What I also did not understand that you elaborated on the RayDAT as if it would be a MADI card. I do not understand what you mean by this. This is simply not the case. It's a PCIe card without any analog converter. So, it also makes no sense to elaborate on the RTL of this card when it has no analog A/D and D/A converter. It depends on the HW you connect to it through ADAT.

Also irritating, you mention so many interface cards, with different number and types of ports, that it really makes no sense to compare these to each other. Especially as the latency inside of the FPGA (internal processing) is IMHO not that different. What has more relevance / effect is transport to PC and back (most) and converter latency (if the cards have any analog I/O).

The highest latency you have anyway with the transport to the computer and back. With RME there is not much difference between PCIe, USB, FW, TB. RME doesn't use 3rd party chips, all is done by the FPGA.

To put an example with a product that I know: the lowest RTL (incl. A/D and D/A) you get with the UFX+ over thunderbolt, but RTL is still around 2.6ms.

I have some doubts that you will find any solution which gives you sub 2ms RTL (incl. A/D and D/A).

The next questions are: for how many channels and at what sample rate.
You might know that A/D and D/A converter latency is a little (!!!) bit lower at higher sample rates.
But by that, you put a severe workload on the computer.
I have doubts that you can get a stable setup that works permanently at sub 2ms for everything.

And for what you have this requirement? I can without any issues play Guitar over a virtual amp with a RTL up to around 13ms, better is below 10ms. This you can achieve with an UFX+ when using an ASIO buffersize of 128 samples at single speed.

If you talk about MADI, when multiple devices are connected serially, the MADI I/O of each OctaMic XTC causes a
delay of 3 samples at single speed (6 at double, 12 at quad speed).

Here is a list of RTLs of different products: you can see that the different transport methods are not that different (USB, FW, TB): https://www.tonstudio-forum.de/attachme … es-v2-jpg/

Maybe you should better start to describe your requirements and use cases.
I think it's more relevant, which devices you need to connect (so what number and types of ports you need).

BR Ramses - UFX III, 12Mic, XTC, ADI-2 Pro FS R BE, RayDAT, X10SRi-F, E5-1680v4, Win10Pro22H2, Cub14