1 (edited by LiftedSeven 2023-05-28 00:49:08)

Topic: HDSPe Raydat vs RME UCX II converters, EQ, latency.

Do both units use same converters?

How closely matched are these units in term of conversion, and if it is identical.
I have RME HDSPe Raydat, and latency is amazing, but I don't like the fact that RME removed EQ out of TotalMix with Raydat

With all due respect, I don't know what you guys were thinking, when even Babyface has EQ.
I can't tell if the logic was behind it to make more money on UCX and Babyface units, and force people to buy other more expensive units from RME, but I can see EQ not being present in DSP software of RayDat is a huge downside to this incredible unit.

I mean latency cannot be beat, correct? But EQ is gone, so in my opinion this card shouldn't be close to 1000 USD, or at least have EQ implemented back into the unit. If electrically possible, maybe it is missing some transistors and chips on it to have EQ like Babyface.

I just hope it is not a Software limitation that was implemented by RME to keep a wall between Raydat and other more expensive units.

I would like to know:

1. How my RME Raydat compares in latency to UCX II
2. How low can you go with buffer sample sizes on UCX II compared to Raydat. Raydat easily stays at 64 samples and 92khz, with an occasional click and pop here and there. That is on i9-13900k and Windows 11.
3. Why is EQ missing on my RME RayDat.

I want to keep Raydat, but I wanted that EQ function in DSP/TotalMix

I might return Raydat and get UCX II, but I need low latency and that EQ is almost a sealed decision, but the price is 750 bucks more.

I use Saffire Pro 40 as a speaker driver in standalone, and I route ADAT in and out of Saffire for clocking to RME Raydat and moving 4 preamps to RME over ADAT with 96khz 24bit rate.

I can do all those things on UCX II, but how is the latency and converters in comparison to Raydat?

Sorry for a bit of rant, but I feel like RayDat is wonderful not to have EQ on it. That's a huge miss for us PCIe guys.


I am considering RME HDSPe AIO Pro to get TotalMix FX option

2

Re: HDSPe Raydat vs RME UCX II converters, EQ, latency.

I deleted the other obsolete thread (1 post) of yours.

Latency of HDSPe cards is a platform property, they all use the same driver and have the same latency.

The AIO Pro also does not have EQ and FX.

On a good working computer UCX II latency is nearly identical to HDSPe.

EQ and FX on a multichannel interface need an extra DSP chip, which is why these are more expensive.

Regards
Matthias Carstens
RME

3 (edited by LiftedSeven 2023-05-28 11:30:37)

Re: HDSPe Raydat vs RME UCX II converters, EQ, latency.

MC wrote:

I deleted the other obsolete thread (1 post) of yours.

Latency of HDSPe cards is a platform property, they all use the same driver and have the same latency.

The AIO Pro also does not have EQ and FX.

On a good working computer UCX II latency is nearly identical to HDSPe.

EQ and FX on a multichannel interface need an extra DSP chip, which is why these are more expensive.

Thank you for the answer, I was misled by Youtuber that said that AIO Pro has EQ

Can you tell me:

1. Do you consider PCIE or USB 2.0 to have more drop outs, or clicks? I tried about 5 devices from your competitors and they all had horrible drop out with USB 2.0. HDSPe Raydat has occasional click and pop, but it's once in a while.
How does UCX II compare to Raydat in that regard?
2. What is the general difference in latency between UCX II and HDSPe units?
I am not asking for exact number, as that is impossible to tell, but generally speaking how much latency difference would I see on UCX II
3. Does UCX II have better clock compared to Raydat? Could be part of question 1, but still worth asking.
 

My PC is very capable, I run i9-13900k with 32GBs of DDR4 6400mhz ram.

I need to justify this purchase besides of EQ, as well as absence of pops and clicks

I am having odd issues with HDSPe Raydat where I am having pops and clicks, but I cannot say for certain if my Focusrite Saffire Pro 40 is causing issues when signal converts from Digital to Analogue (to the monitors) from ADAT being fed to Saffire Pro 40

I speculate there is an issue with Saffire Pro 40, but I don't have another ADAT piece of gear to test RME's output with.
Happens more when I set my RME Raydat to Internal clock.

4 (edited by ramses 2023-05-28 12:25:09)

Re: HDSPe Raydat vs RME UCX II converters, EQ, latency.

> Is there a PCIE card by RME that has EQ? (Besides of HDSPe MADI FX)

No

> What is your general difference in latency between UCX II and HDSPe units?

Extremely low, check my comparison for different products here:
https://www.tonstudio-forum.de/attachme … es-v2-jpg/

As a side note … which latency do you mean?

A) Converter Latency?

The RayDAT has no AD/DA converter, this you can't compare and depends on what device you connect via AD/DA

B) Latency for transport through PCIe/USB? Nearly the same.

C) RTL (round trip latency, incl. AD, DA, transport from/to PC)?

Depends to the highest degree on the selected ASIO buffer size.

Converter latency: Is in the range of 5–6 samples for the UCX II which is 0.11/0.13ms for AD/DA (worst case at 44.1 kHz, with higher sample rates it becomes even lower). Again: RayDAT has no converter of its own.
RTL: In contrast to that, the RTL can be between 3-~100ms depending on whether you choose an ASIO buffer size of 32 or 2048 samples at single speed.

As a side note: not much important, but I want to mention it. There is a little difference between the lowest ASIO buffersize that you can select in different drivers. The RME HDSPe driver has 32 samples as the lowest ASIO buffersize at single speed. The RME USB driver 48 samples. The MADIface driver, 32 samples.
But this is kind of "artificial". Nobody would permanently work with 32 samples because the risk of getting audio drops is simply too high if you work on DAW project with different VST and maybe also VSTi.
Normally, everything up to 128 samples ASIO buffer lets you stay under 10ms, which is critical for playing VSTi.
And for pure recording, you would anyway use the highest value to have the best stability and that nothing gets lost during a recording session.

I used 32 samples only at the beginning when I tested Amp VSTi. Otherwise, I made the experience that I can play guitar very well in time if the RTL stays below 10ms. And this is the case with any ASIO buffer size of up to 128ms at single speed.
Well, but this is a special case and not applicable for everybody.

D) DPC Latencies

Regarding "My PC is very capable". Maybe in usual synthetic benchmarks, but they do not check whether the system can
handle audio applications with real-time requirements well, which is a different thing compared to pure number crunching.

DPC latencies can happen when badly written drivers block CPU cores for too long. You do not hear it as a delay (like an audio effect) but if the PC cannot process audio in time, then you will get audio loss at a certain point.
If an audio process is scheduled to such a CPU core, then audio loss can easily happen, no matter how powerful your PC or CPU actually is.

Should you use very low ASIO buffer sizes, then the risk is increased that a PC might not be able to process audio in time.

There are other things around energy saving, BIOS, 3rd party USB chips, bad mainboard design which can also be reasons, why a PC can be powerful but not well suited for processing applications with near-realtime processing requirements.

BR Ramses - UFX III, 12Mic, XTC, ADI-2 Pro FS R BE, RayDAT, X10SRi-F, E5-1680v4, Win10Pro22H2, Cub13

5 (edited by LiftedSeven 2023-05-29 02:06:37)

Re: HDSPe Raydat vs RME UCX II converters, EQ, latency.

ramses wrote:

> Is there a PCIE card by RME that has EQ? (Besides of HDSPe MADI FX)

No

> What is your general difference in latency between UCX II and HDSPe units?

Extremely low, check my comparison for different products here:


As a side note … which latency do you mean?

A) Converter Latency?

The RayDAT has no AD/DA converter, this you can't compare and depends on what device you connect via AD/DA

B) Latency for transport through PCIe/USB? Nearly the same.

C) RTL (round trip latency, incl. AD, DA, transport from/to PC)?

Depends to the highest degree on the selected ASIO buffer size.

Converter latency: Is in the range of 5–6 samples for the UCX II which is 0.11/0.13ms for AD/DA (worst case at 44.1 kHz, with higher sample rates it becomes even lower). Again: RayDAT has no converter of its own.
RTL: In contrast to that, the RTL can be between 3-~100ms depending on whether you choose an ASIO buffer size of 32 or 2048 samples at single speed.

As a side note: not much important, but I want to mention it. There is a little difference between the lowest ASIO buffersize that you can select in different drivers. The RME HDSPe driver has 32 samples as the lowest ASIO buffersize at single speed. The RME USB driver 48 samples. The MADIface driver, 32 samples.
But this is kind of "artificial". Nobody would permanently work with 32 samples because the risk of getting audio drops is simply too high if you work on DAW project with different VST and maybe also VSTi.
Normally, everything up to 128 samples ASIO buffer lets you stay under 10ms, which is critical for playing VSTi.
And for pure recording, you would anyway use the highest value to have the best stability and that nothing gets lost during a recording session.

I used 32 samples only at the beginning when I tested Amp VSTi. Otherwise, I made the experience that I can play guitar very well in time if the RTL stays below 10ms. And this is the case with any ASIO buffer size of up to 128ms at single speed.
Well, but this is a special case and not applicable for everybody.

D) DPC Latencies

Regarding "My PC is very capable". Maybe in usual synthetic benchmarks, but they do not check whether the system can
handle audio applications with real-time requirements well, which is a different thing compared to pure number crunching.

DPC latencies can happen when badly written drivers block CPU cores for too long. You do not hear it as a delay (like an audio effect) but if the PC cannot process audio in time, then you will get audio loss at a certain point.
If an audio process is scheduled to such a CPU core, then audio loss can easily happen, no matter how powerful your PC or CPU actually is.

Should you use very low ASIO buffer sizes, then the risk is increased that a PC might not be able to process audio in time.

There are other things around energy saving, BIOS, 3rd party USB chips, bad mainboard design which can also be reasons, why a PC can be powerful but not well suited for processing applications with near-realtime processing requirements.


Great measurements in that picture you linked, but I wish you did UCX II in there too.
Since you have not included UCX II in that ranking, what is approximate percentage difference between it and Raydat?

I am not looking for exact here, but generally what would you say is the percentage between them (if you know) ?

This is my latency report, and that is the same be that an hour or 5 minutes of running, stays pretty fixed, only current value jumps up to 25 once in the while
https://tinypic.host/images/2023/05/29/Desktop-Screenshot-2023.05.28---17.44.33.60.png

What do you think of it?


You see, I am battling pops and clicks that happen when I listen to music, look through Youtube, or simply play games, I am not even talking about recordings. It seem like I have pops and clicks even with higher buffers and HDSPe Raydat.

I tried to isolate the issue, but the only thing that could be part of the issue imo is the GPU and my older Saffire Pro 40.
Perhaps it's the clock that is being unstable on the unit itself, or transient spikes of RTX 4090 (well documented issues with 4090 cards)

I started having FW Active lock issues with Saffire Pro 40, so I put it in standalone and use HDSPe as Master clock, and tried to sync to Saffire by running Saffire as Master as well.

I still have an occasional pop. I ran 64 samples and 96khz (same as Saffire) on HDSPe Raydat, then after having couple of clicks, I bumped it up to 256 and 512, and I actually didn't see much difference, so I set it back to 64 samples 96khz 24 bit.

I speculate it's an older Saffire Pro 40, but I cannot be 100% sure. I have no other ADAT device to sync HDSPe Raydat to.
I was thinking of buying DAC for Raydat, or returning it and getting UCX II, because it has what I need + EQ in Total Mix.

There is something wrong with that signal chain, but I cannot trace it, it happens way less with HDSPe, but it still happens, and it's random, be that high buffer or low through Hammerfall DSP software.

I would jump to UCX II instantly if I knew I would get rid of these pops and clicks completely :-(

Right now UCX II is also sold out on Sweetwater, so I can't purchase one to try, and return window on my HDSPe is running out in next couple of weeks.

Don't know what to do for certain.
Other brands like Focusrite and Audient, SSL2+ and such had constant drop outs, way worse than my Saffire Pro 40 over Firewire.

So, I returned them all.

Have you experienced pops and clicks while doing normal things such as Youtube with USB 2.0 RME devices?
I know RME is quality, but it's also picking a right configuration, and knowing how low can I run my buffers, as higher buffers affect some things negatively for me, besides of recording audio.

I play competitive games, where a 5 ms can make a difference as I listen to footsteps and such.
I also hate recording with 10 ms latency, that is disorienting for a style I record in. It makes me slow down a lot.


Thanks for input
I have done a lot of reading on these units, but the ultimate answer eludes me

6 (edited by ramses 2023-05-29 10:01:30)

Re: HDSPe Raydat vs RME UCX II converters, EQ, latency.

> Great measurements in that picture you linked, but I wish you did UCX II in there too

I can only include products that I am using.

> Since you have not included UCX II in that ranking, what is approximate percentage difference between it and Raydat?

Seems that you still didn't fully understand the meaning of the list
- the differences between different RME solutions and drivers is very low
- it doesn't make a significant difference whether you use a recording interface with or without analog I/O
Concentrate on choosing the recording interface that has the needed I/O ports and features.

> What do you think of it?

Needs to run for a longer time to catch any potential peaks, but if the values stay at this, there is no reason to complain.

> I would jump to UCX II instantly if I knew I would get rid of these pops and clicks completely

Nobody can tell, you need to try. But if you get pops and clicks with the RayDAT when using RME driver, then it might be that the root issue is still there, regardless of which recording interface you use.

Reminds me of my issues with occasional audio drops of around 200ms every 10–20 minutes (not fix time interval). DPC latencies looked well, but such issues.

It turned out to be caused by energy saving on my graphic card, nVidia RTX 2070 Super, which could be fixed by turning it off using a tool from a Russian developer called powermizer. Use extended forum search and search for powermizer and my name ramses. I didn't detect it immediately after GPU change because I used at that time Windows 7 and, if I remember right, this issue popped up only under Windows 10. At the support end of Win7, I had to change to Win 10.

> Have you experienced pops and clicks while doing normal things such as Youtube with USB 2.0 RME devices?

No. I have otherwise a very solid PC from 2014, no drops even in higher load situations (meanwhile upgraded CPU, GPU, USB3 cards).
https://www.tonstudio-forum.de/blog/Ent … mponenten/

This is my "artificial" load Test using a big Cubase project with 400 tracks and 800+ Steinberg VSTs.
No drops at lowest ASIO buffer sizes of 32 samples at single speed and 64 samples at double speed.
https://www.tonstudio-forum.de/blog/Ent … cks-de-en/

BR Ramses - UFX III, 12Mic, XTC, ADI-2 Pro FS R BE, RayDAT, X10SRi-F, E5-1680v4, Win10Pro22H2, Cub13

Re: HDSPe Raydat vs RME UCX II converters, EQ, latency.

Try disabling the e cores of your CPU, in the bios. If this fixes your clicks you know what the cause is.

Vincent, Amsterdam
https://soundcloud.com/thesecretworld
BFpro fs, 2X HDSP9652 ADI-8AE, 2X HDSP9632

Re: HDSPe Raydat vs RME UCX II converters, EQ, latency.

LiftedSeven wrote:

Great measurements in that picture you linked, but I wish you did UCX II in there too.
Since you have not included UCX II in that ranking, what is approximate percentage difference between it and Raydat?

I am not looking for exact here, but generally what would you say is the percentage between them (if you know) ?

Here are some RTL (Round Trip Latency) measurements for UCX II for various combinations of sample rate and buffer size:

https://www.audiosciencereview.com/foru … st-1455790

9 (edited by LiftedSeven 2023-06-02 09:53:03)

Re: HDSPe Raydat vs RME UCX II converters, EQ, latency.

Muffin wrote:
LiftedSeven wrote:

Great measurements in that picture you linked, but I wish you did UCX II in there too.
Since you have not included UCX II in that ranking, what is approximate percentage difference between it and Raydat?

I am not looking for exact here, but generally what would you say is the percentage between them (if you know) ?

Here are some RTL (Round Trip Latency) measurements for UCX II for various combinations of sample rate and buffer size:

https://www.audiosciencereview.com/foru … st-1455790


Thank you, I actually saw that measurement at this exact post couple of days ago. I am part of that forum as well.
But, it is so vague, there is no comparison to other RME units, and we have no idea what hardware (cpu) the person is using.

Kind of a blind test.

10 (edited by ramses 2023-06-02 10:36:10)

Re: HDSPe Raydat vs RME UCX II converters, EQ, latency.

> there is no comparison to other RME units, and we have no idea what hardware (cpu) the person is using.

RTL has nothing to do with CPU performance.
You can expect that all RME solutions have similar "performance" regardless of the driver.
Even USB and PCIe are on par.
See here a comparison of different products that I owned.
https://www.tonstudio-forum.de/attachme … es-v2-jpg/

With newer products also converter latency is very low and much lower compared to communication over USB/PCIe where additionally buffer sizes are needed for the communication between recording interface and PC and depending on the application / system load.

A comparison of converter latencies you can see here in my Excel comparison sheet:
https://forum.rme-audio.de/viewtopic.php?id=35156

Can you describe your use case and formulate latency requirements for it?

Please also be aware of, that RayDAT is a digital working card, for e.g. comparing RTL you would need to connect AD/DA converter to it. The final RTL then depends on your product choice, how fast the converters are.

BR Ramses - UFX III, 12Mic, XTC, ADI-2 Pro FS R BE, RayDAT, X10SRi-F, E5-1680v4, Win10Pro22H2, Cub13