1 Topic by rmemius (edited by rmemius 2023-09-22 02:35:21)

  • Registered: 2022-06-02
  • Posts: 21

Topic: Welcoming comments about TM FX (naming) + adding the FX to HDSPe line?

Happily, a few seasons ago I recall discovering way(s) to deduce which interfaces HAVE the effects chip in TotalMix. (including USB, TB, etc.)

Additionally, some don't have the compressor, like my Babyface pro, though have EQ, reverb and Delay.


I understand it's a chip, I understand most of the realities at this point.

What have I missed I am wondering. What might future bring? ~> ALL HDSPe cards HAVING the effects chip?


1) do you find the naming TM FX confusing?  EX. all but one HDSPe cards are without effects ( TM FX?).

2) Are you disappointed that ALL the other HDSPe cards are missing the FX portion (delay, reverb, compressor, EQ) ?

3) Would you be much more interested in an HDSPe card IF it had the FULL FX portion?

ONE GLARING HOLE IN MY UNDERSTANDING is whether the number of possible tracks a card COULD deliver has something to do with why the HDSPe cards are missing the FX (effects) portion?


-------
Yet, the HDSPe MADI FX card does hundreds of tracks, and presumably it can have the ACTUAL EFFECTS of totalmix on every channel? NOT that I'd ever need to. Maybe there's a limit to the number of effects possible with HDSPe MADI FX - I will read manual though will post this as is for now.
-------



Perhaps you're like me and have reconsidered your interest in Thunderbolt.
I now believe that HDSPe is the way to go for most stability and performance. All else aside, the MAIN BENEFIT is ~~>>

NO CPU hit like there is with USB  (which is often waved away as not important in debates - "USB is 'good enough')

Anyways, for likely critical live performance work ~>  I'm now all about needing HDSPe cards. (stability - latency)

Should we be asking for the full compliment of effects to be in the next version?

AS AN ASIDE: (thank you) I'm most interested in the entire  Dante vs AVB, (AND also AES/ANALOGUE!!) -- and how one can build, perhaps, a live MONITOR rig with JUST RME cards and Steinberg's VST LIVE PRO.  Without a digital mixer onstage, using splitters/breakout stage boxes, Analogue i/O, via AVB etc.   

I currently believe ~> We NEED the  FX part of TotalMix FX in the next releases of HDSPe cards?  :-))

make the card bigger if necessary ?

COMMENTS WELCOMED!-)

best wishes to all
RME FOREVER -  just a fact that!

2 Reply by ramses (edited by ramses 2023-09-22 06:14:23)

  • ramses
  • DAW Mastermind
  • Offline
  • From: UTC+1
  • Registered: 2012-10-14
  • Posts: 13,184

Re: Welcoming comments about TM FX (naming) + adding the FX to HDSPe line?

I am not against it, don't get me wrong, I am only trying to explain to you the situation.

I would be glad if features like FX chip (and maybe also DURec) could be added to any recording interface.
On the other hand, this will have a price impact.
Now where everything in the world has become pricier, this might lead into the wrong direction for customers, who possibly couldn't pay even higher prices for the products (some products became already pricier).

Regarding TotalMix FX, this has been discussed already years ago. The name TotalMix FX is the new name after a big overhaul of TotalMix. I doubt that this will ever change. Imagine the efforts (changes of web server, product packaging, manuals).
I think it's better for us customers, that RME spends the time working on the things that bring us more benefits, than renaming (plenty of) products.

I think the FX chip was never designed to support all FX on all channels. Most people use and prefer the FX of their DAW or 3rd party FX, or are using external hardware anyway. So why make a product much pricier for a feature that merely anybody would use? One for sure, this would require a more powerful FPGA (programmable/flashable CPU, a 2nd one for full implementation of FX).
I noticed this myself when checking it on UFX, UFX+ (on UFX III I didn't try). But I can tell you that I never needed so much FX power. Currently, maybe 3% are being used in my setup.

During my comparisons with different UFX products (UFX+, UFX III) I never experienced any problems with "CPU hits".
The system load (CPU utilization) was comparable between RayDAT (PCIe based) and UFX+/III (USB3).
As a side note: the system is from 2014 (same Supermicro server mainboard since then), when I started the tests it was still using a 6-core CPU, Intel Xeon E5-1650v3. Over the years (9!) I upgraded to E5-1650v4 and last recently to E5-1680v4 with 8-cores, more overall performance, but with slightly less single thread performance.

That for 400 tracks and 803 VSTs at the smallest ASIO buffer size, for single speed 32 samples, for double speed 64 samples.
https://www.tonstudio-forum.de/blog/Ent … cks-de-en/

I think the key here is to have a system with good drivers. Also, PCIe and TB are no guarantee for best performance and stability if the CPU cores are being blocked by bad drivers that occupy CPU cores for too long (-> DPC latencies).
The problem is that Windows and macOS are no real time operating systems.
Drivers (low-level routines) are not under the control of the process scheduler for data consistency reasons.
They have to detach themselves from CPU cores based solely on "programming conventions". Some companies are possibly too eager to get better benchmark results by using the CPU a little longer, no issue for applications without real-time demands. For audio processing with real-time demands, not so well.
A bad driver might block a CPU core for too long (-> DPC latencies). If an audio related task with near-realtime requirements is scheduled to run on such a core which is blocked by such a bad driver, then it is likely to get audio drops. Especially if you are using small ASIO buffer sizes. The system needs to perform I/O more often, which increases also the "interrupt load" on a system, which means even more context switches for the CPU and more often the execution of low-level routines.

On a well-designed system with good hardware and drivers, you will have no issues using RME recording interfaces with a combination of good drivers and FPGA based communication for any transport to the PC (USB, FW, PCIe).

The current situation is that Intel sadly didn't release the promised TB chip and according to RME the latest TB chips are too complex and too expensive to be used in audio interfaces. I am not so convinced that this will ever change, and it looks as if other vendors currently simply use those TB chips that they have on stock. Once the stock is empty, they will have the same situation. The advantage of RME, they have the knowledge and excellent drivers, so that USB "feels" like PCIe/TB.
If you look at the RTLs, there is no significant difference.
https://www.tonstudio-forum.de/blog/ent … cts-en-de/

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

3 Reply by rmemius (edited by rmemius 2023-09-22 06:39:23)

  • Registered: 2022-06-02
  • Posts: 21

Re: Welcoming comments about TM FX (naming) + adding the FX to HDSPe line?

ramses-!!
Great to receive one of your always cool, deep, and straight-forward replies!
Thanks - will read now!

-------------- bit of a quick ramble below -------------


TOPLINE: copied from bottom to top ~~>>


ALL TO SAY: Geez, add that effects chip to the HDSPe cards and sales will go way Way WAY UP!!!

AND, have to add - NOBODY much cares about another 50 or 100 bucks/euros for the effect (FX-) chip in the final HDSPe product. That's just NOT an issue in my view.

SO, DOES anyone else wanna see the effect (FX-) chip in FUTURE HDSPe cards?

------- ------- ------- ------- ------- ------- -------

OK, have done a quick read now.
And, i will read again in a bit after posting what will be below.

1) I am no longer that keen on thunderbolt - however thunderbolt 5 is out soon - so all the thoughts about no chips is puzzling.

For live performance I, as well as many many others (percentage-wise) i believe, would do best with a direct card IN a PCIe slot.
That ONLY the Madi FX HAS the 'actual' effects i am again puzzled.

2)   TOTALMIX actual EFFECTS:   Does that mean only some channels in totalmix can have effects on them i am wondering? that would be fine though! (i have not yet read the manual, sorry)

3)  My post was on two main subjects, with an additional, bit of a poke-) at the whole, USB is "good enough" debate.

   USB RME audio interfaces means the CPU can not get past about 80% usage ROUGHLY.

[DIFFERENCES BETWEEN PROGRAMS]

I LOVE CUBASE now,  due to their astonishing progress with VST LIVE PRO - which has made gigging possible like i never thought would be possible in 2023.

I was shocked that I went back to Steinberg after almost 2 decades away.

With Cubase and my superbly optimized win11 i9-9900K the CPU craps out regularly -and no-  it's not plugins or any of the usual things folks miss. 

CPU crap out using VST INSTRUMENTS !!!   Not just audio streaming.

I'm the obsessive type and never much assume I've "got it figured"
READ: I doubt myself until I'm positive enough i have it right enough.

I think 'you know that one'.

------- OPTIONAL READ - RAMBLE -------
I will always love REAPER and support the DEVs (must use the "I LOGIC V3.90 ULTIMATE" theme though!)

-- CUBASE gets music completed for me. Less fiddling about. I miss tons from oh-so-lean+brilliant REAPER and use it for batch exporting and more.

In reaper the CPU is 'currently' vastly more efficient (in my experience) - yet delay compensation is very different between CUBASE and REAPER "perhaps".

CUBASE doesn't have it's MIDI start a bit late after recording it etc.   WHO KNOWS. I never did figure that out much with REAPER and delay compensation - I could be WAY off.
------- ends of ramble -------

ANYWAYS: RE: CPU and USB ~~>> ABSOLUTELY for definite, and pretty much missing in RME's "Why still USB" video. ~>

Is that somewhere around 20% of the POSSIBLE CPU power will be lost, literally, due to the USB crapping out at around 80%, -- instead of CPU being usable right up to 98% as with a PCIe card.     A FACT (pretty much heh-heh)

4) My RME naming convention puzzlement:   I have wondered why RME calls it TOTALMIX FX when the FX-effects (which are a 'relatively' new thing) are only in some interfaces. it's confusing. And, as none, but one, HDSPe card HAS the effects chip. (MADI FX) Well, it's just confusing to display interfaces as having TOTALMIX FX.   

just a finer point and me, hoping to be helpful.


ALL TO SAY: Geez, add that effects chip to the HDSPe cards and sales will go way Way WAY UP!!!

AND, have to add - NOBODY much cares about another 50 or 100 bucks/euros for the effect (FX-) chip in the final HDSPe product. That's just NOT an issue in my view.

Best to you ramses!


SO, DOES anyone else wanna see the chip in FUTURE HDSPe cards?

4 Reply by rmemius (edited by rmemius 2023-09-22 06:56:45)

  • Registered: 2022-06-02
  • Posts: 21

Re: Welcoming comments about TM FX (naming) + adding the FX to HDSPe line?

Actually, I can convince myself that with all the benefits of an HDSPe card (for me) -- I should be able to overlook that the effects are missing in the HDSPe cards...


EXCEPT. grrr, the LEAST attractive HDSPe card, (for me) the MADI, HAS THE EFFECTS !!!!!!!!    hehehehehe



ALSO, totally fine with the name Totalmix FX - though some people might think they're getting EFFECTS. that's all.

OBVIOUSLY a known aspect and I appreciate the time-line of the transition to FX. ALL IS GOOD!

SIGH, as effects ARE IN the MADI FX HDSPe card - I hope sales of it make it logical, so that then circumstances SHOULD see the other HDSPe cards WITH the effects eventually.


HOPE THE MADI ONE sells well - though i doubt it would be as popular maybe as the other HDSPe cards IF they had effects.

Though fact is, MADI/AVB/DANTE are new to me - so i still have learning to do. 

IF even THE RAYDAT card (for example) had effects it should fly out the door I'm guessing.

ACTUALLY - I'll bet 50 it's right around the corner! (by Spring/Summer 2024)



[PERFORMANCES AND attainable IN-EAR MONITORING systems ARE GETTING VERY POPULAR]

Like a ton of people since this year ~~> MY GOAL IS LIVE IN_EAR MONITORING SYSTEMS  -- hopefully/perhaps WITHOUT the need for a digital mixing desk.  USING primarily Steinberg's VST LIVE PRO using the programs "peers feature" for band members controlling own in-ear mix --  AND using RME INTERFACES - D/A  A/D etc.  AVB or DANTE I expect.   

MADI is coax!?   yikes!  ummm, ewww. heh-heh ~> showing my ignorance about MADI - yet if it's only coax that's not as useful as CAT5/6/7 cable i'd think. (For me, and band performer, that is. I could be missing a lot though)

Not having to use dozens of XLR cables is HUGE - I'm still at the beginning understanding phase.

So, IF NO need for MIXING DESK - those TOTALMIX effects FX should be INCREDIBLY useful to save CPU for non-critical. 
WHILE special effects VST effects would be done via Steinberg's VST LIVE PRO and more CPU power.


(VST LIVE PRO is getting weekly updates!)


I should mention how important the CPU "headroom" is-- ABLE to go to 98% usage is in a LIVE situation using Steinberg's VST LIVE PRO program.




THANKS ALL. COMMENTS WELCOMED

5 Reply by ramses (edited by ramses 2023-09-22 06:37:27)

  • ramses
  • DAW Mastermind
  • Offline
  • From: UTC+1
  • Registered: 2012-10-14
  • Posts: 13,184

Re: Welcoming comments about TM FX (naming) + adding the FX to HDSPe line?

80% CPU is a severe system load, I wouldn't recommend having such a high load on systems for audio processing and especially when having to use low (ASIO-) buffer sizes to get low RTL's.

Regarding your audio issues on your system. Without knowing your system and the project, it's impossible to make any claim or statement performance wise. Just a few points:
Your
- VSTs or VSTi's might need more processing power
- system might have too high DPC latencies or a too low single thread performance for your projects
- project might be too complex creating too high demands for processing power / single thread performance

You might not have read all I wrote about thunderbolt chips.
"The current situation is that Intel sadly didn't release the promised TB chip and according to RME the latest TB chips are too complex and too expensive to be used in audio interfaces."
If TB4 created already such issues, guess what will happen with higher TB standards, do you really think it will become less complex and less expensive? It would be nice, but I bet, that reality is different.

> Does that mean only some channels in totalmix can have effects on them i am wondering?
> that would be fine though! (i have not yet read the manual, sorry)

Many, but according to my experience with UFX, UFX+ not all, as I said, didn't try with UFX III.
With UFX and I think also UFX+ I tried to enable as many as possible, but it is so long ago, sorry I do not remember.
All I can say is, that you can't enable FX for all channel, but I forgot for how many.
Currently my FX load is quite low (see screenshot).
My demand for FX in terms of the number of channels, where I enable it is generally relatively low.
But I would definitively miss it, if FX wouldn't be there.

https://www.dropbox.com/scl/fi/nfk537bnna7mig1wpo2oa/2023-09-22-TM-FX-FX-load.jpg?rlkey=vgyslp2r5adf80kz3aqy2eyey&dl=1

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

6 Reply by rmemius (edited by rmemius 2023-09-22 06:48:57)

  • Registered: 2022-06-02
  • Posts: 21

Re: Welcoming comments about TM FX (naming) + adding the FX to HDSPe line?

80% CPU high?    tell that to my super old lovely pair of 9632's which happily run/ran well over 90% no crackles/dropouts!  heh-heh

--will read your kind write now --

as everyone's somewhat different - perhaps you and me are superb examples*-) 
the world should be like us! COMMUNICATING - DISCUSSING - FIGURING

(NRN) thanks! NoReplyNeeded - also means write anytime of course!
"I'll be back"

7 Reply by ramses (edited by ramses 2023-09-22 07:00:41)

  • ramses
  • DAW Mastermind
  • Offline
  • From: UTC+1
  • Registered: 2012-10-14
  • Posts: 13,184

Re: Welcoming comments about TM FX (naming) + adding the FX to HDSPe line?

Re-Edited, wrong image link

Do you mean peak load or sustained / average load?

For comparison: here a Screenshot of my "artificial cubase load" project:

400 tracks, 803 VSTs, 44 kHz, ASIO buffer size: 32 samples (lowest with MADIface driver 0.9827)
CPU load: 2-4%, Average: 3% (!)

Windows 11, 22H2
CPU: Intel Xeon E5-1680v4 (8-core) from Q4 2016, current pass mark rank #731 (!)
https://www.cpubenchmark.net/cpu.php?cp … mp;id=2869

https://www.dropbox.com/scl/fi/t6ukx3qltagx9pq37a0x0/2023-09-22-CPU-load-400-tracks-at-44.1-kHz-and-32-samples-ASIO-buffersize.jpg?rlkey=us89yrtgd26whm4ktnp3x4azz&dl=1

VSTs ....

https://www.dropbox.com/scl/fi/u4bg23blj11foqweicvnn/2023-09-22-CPU-load-400-tracks-at-44.1-kHz-and-32-samples-ASIO-buffersize-VSTs.jpg?rlkey=0ifuhnui2kqzmc4vwccq62x34&dl=1

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

8 Reply by rmemius (edited by rmemius 2023-11-08 07:09:19)

  • Registered: 2022-06-02
  • Posts: 21

Re: Welcoming comments about TM FX (naming) + adding the FX to HDSPe line?

HI ramses !

I'm back in town now. Sorry for late comment(s).

1) THANK YOU for your great efforts -- your replies to my now 3 various post areas (posts started or replies) 

~> with me always bemoaning (politely) that there's NO THUNDERBOLT interface(s).


2) I notice USA-SYNTHAX doesn't date this article, unless I've missed it.

where they boldly state: Because it is dead.

https://www.synthax.com/rme-why-usb-and … underbolt/

THUNDERBOLT 5 is already out somewhat. And, UAD is doing rather well with thunderbolt 3 interfaces.

3) Noting your kindly posted image above. You don't use VST instruments much?
Maybe try a few VST instruments and watch what happens at over 80% CPU load. Depending on much of course.

I STAND BY MY STATEMENT as it's FACT.

RME PCIe cards can be run right to 98-ish percent CPU usage on a well managed windows PC.

RME Thunderbolt I would think no crackling/drop-outs until over 90%

USB 2/3 craps out somewhere around 80% or less, depending
~> (guesstimate though you get my point and/as you've be great about all my comments)



I want RME to be the best in ALL AREAS.

I do NOT want to get anywhere near a UAD APOLLO

I am now wanting RME PCIe cards and am not even interested much in thunderbolt

THE NEW PROBLEM IS THAT THERE IS ONLY THE PCIe MADI card interface that has the EFFECTS chip.

One needs to be careful as well as the naming TotalMix FX suggest effects chip -
and yes, I've managed to find the, for me, not so easy to find website page that shows
which interfaces  DO NOT have the effects chip - for, uhhh... TOTALMIX FX

sigh - smile

bit of a rush here

THANKS AGAIN - best wishes to all

9 Reply by torbenscharling (edited by torbenscharling 2023-11-08 18:16:31)

  • Registered: 2019-02-14
  • Posts: 148

Re: Welcoming comments about TM FX (naming) + adding the FX to HDSPe line?

I agree 100% with rmemius.

It is far too common for people to be commenting back a bunch of excuses or steering in the other direction. rmemius points out FACTS that are KNOWN by ANYONE that actually WORKS with AUDIO on a REGULAR basis and WANTS to BETTER their PRODUCT.

If a CUSTOMER is giving PROACTIVE advice, suggestions, wants and needs, it is not up to other CUSTOMERS to throw those ideas away, as if they hold no merit or as if they speak for the company/other consumers. RME is already overpriced/expensive so really I think it's laughable to suggest adding FX would be too expensive for their customers.

The fact is, all Pro's with experience know how crappy USB is to work with regardless who writes the drivers, it is a bottle neck protocol more so than PCIe. That is a fact. Price of adding DSP chip is negligible. TotalMix should not be marketed as TotalMix FX when the product(s) does not have any FX.

10 Reply by ramses (edited by ramses 2023-11-09 19:04:28)

  • ramses
  • DAW Mastermind
  • Offline
  • From: UTC+1
  • Registered: 2012-10-14
  • Posts: 13,184

Re: Welcoming comments about TM FX (naming) + adding the FX to HDSPe line?

RME has adequately explained why there are currently no Thunderbolt-based recording interfaces and probably won't be any in the future.

As for my background, during my extensive career, I was involved with major enterprise and datacenter clients in the server and network area, dealing with various aspects, including sizing, capacity planning, and end-to-end troubleshooting.

I would still not recommend pushing a computer to 100% CPU utilization, especially when critical applications are involved and performance is crucial. This is especially true for applications with near-real-time requirements, like audio processing. If it works for you, that's great, but it shouldn't be generalized. Moreover, the performance can vary significantly from one system to another. Different HW, Setup, drivers, thermal throttling, specific workloads and load by other system activities.

For normal security considerations, it's wise to plan with a certain headroom and thoroughly test it on the particular system, rather than assuming anything. Each system has different hardware and drivers, and their performance can vary. I assume you are familiar with the topic of DPC latency.

Now, I've become curious about these claims that you can fully utilize a system up to 100% CPU load ONLY with PCIe, while USB can only go up to about 80% CPU load.

My test setup:

Hardware

UFX III via USB, MADIface driver 0.9827.
PC hardware: https://www.tonstudio-forum.de/blog/Ent … mponenten/

CPU: https://www.cpubenchmark.net/cpu.php?cp … mp;id=2869
Passmark score: around 14,000. Single-thread performance: 2281. Overall Passmark Rank: 743.

The system is based on a 9-year-old server motherboard with a 7-year-old CPU. Also functions as an office, gaming and NAS system (RAID-0 and 10 Gbit LAN). Certainly, not "up-to-date" compared to the CPU performance of current systems, but based on a powerful 8-core CPU with sufficiently high single thread performance and additional reliability by ECC-RAM. All in all, a flexible, powerful and high-performance silent-system with a high upgrade potential.

The DAW project

Cubase 13 Pro, an "artificial" load-test project with 400 tracks, 803 VSTs, 2 VSTi.
ASIO buffer size set to 64 samples @44.1 kHz.

The CPU load remained relatively low at 1-4% (memory usage 18 / 64 GB).
So I had to activate the CPU-Z CPU stress test to reach a sustained system load of 98-99%.

Result: there were no audio dropouts over USB3 with a permanent CPU load between 98-99% (mainly caused by CPU-Z Stress Test on top of the DAW load).

Perhaps this information will provide some inspiration to reconsider the topic of system load and USB.
Also not every vendor can handle 70 tracks of I/O (in and out) via USB2...

https://www.dropbox.com/scl/fi/u0eezf2gmzg3qifsnjchm/2023-11-08-Load-Test-99-CPU.jpg?rlkey=ctjwl8truqeanj0aft7ylfj74&dl=1

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

11 Reply by rmemius (edited by rmemius 2024-01-02 00:08:20)

  • Registered: 2022-06-02
  • Posts: 21

Re: Welcoming comments about TM FX (naming) + adding the FX to HDSPe line?

Hi ramses, hi everyone!

[On how the interface effects are allocated]
i certainly have gaps - for one, I've long wondered how a PCIe card with hundred of channels could have effects on every channel.
I have now discovered that the EFFECTS chip has a display in the top right of totalmix showing how much of the effect processing power is used, and, still available!

I am hopeful that ALL future RME PCIe cards will come with the fullest effect chip. Certainly no-one cares about more cost vs less features generally.


For LIVE performances the RME effects chip makes computer based solutions sensible!!!   (ex. Steinberg VST LIVE PRO)


1) Myself, I'm now only interested in PCIe cards.

2) i am uncertain, though studying, MADI - DANTE - AVB - AES - etc. Hoping to decide on how best to use system for LIVE and STUDIO. NOTE: not a question, just where I'm at with thinking through.

3) As a few other manufacturers - antelope, etc. are putting out Thunderbolt interfaces, and are having lots of complaints about drivers - I'm guessing that RME is wisely accepting that their deservedly TOP reputation USB drivers might not be the same with thunderbolt drivers for technical reasons.  Not going to attempt more on that wondering. (see No.1 above?-)

4) as noted in top post here - when a card says it comes with TOTALMIX FX i figure it should best clearly say that it does NOT have FX. hehehe. I know there's a chart on the site somewhere. Yes, as you pointed out - it's the name of totalmix.
Just a finer point that - and, my babyface pro doesn't have compression - which i knew though.

-->> 5)   people often love to track vocals with a bit of reverb etc. and for LIVE performances,, some not-CPU-taxing effects, would be nearly essential.  (STEINBERG VST LIVE PRO is outrageous and new code --> with an RME PCIe interface that HAS the effects chip --> NO DIGITAL MIXER NEEDED-)

Yes, there's the MADI FX card - though time are a changin' - AVB - DANTE - etc.  The dust hasn't settled as they say*-)



Thanks for all the great charts and comments on my few threads I started. (Firewire & USB FORUM and HDSPe FORUM)

https://forum.rme-audio.de/viewtopic.php?id=35699

https://forum.rme-audio.de/viewtopic.php?id=36484

(THIS THREAD) https://forum.rme-audio.de/viewtopic.php?id=38106



Regarding your CPU usage chart/images above -

Wondering if you've tried realtime played VST instruments in your stress test above *-)

[running in the 95% region]
I was always able to get my pair of OLD PCI cards, synced with the RME supplied jumper cable between them running happily waaay up in the 90's without hiccups, crackles and the usual.

AND just for fun here (sigh) ->

LOVE my BABYFACE PRO for light work

Yet, I don't see myself ever using a USB interface for live performances and therefore can only be interested in PCIe cards or just maybe,
a thunderbolt interface --
though I'm now moving to firmly believing that thunderbolt drivers may never be that stable. (?-dunno-?)