1 (edited by rmemius 2022-07-30 02:39:13)

Topic: Enjoyed: 'Why we don't need Thunderbolt' video--No mention of CPU hit?

RME always, and for life - literally.   Since the VERY first cards.

SO, "sigh/smile"

Been waiting for a TYPE-C TB4/USB4 interface and sadly, will continue to wait.


Correct Me, Please and Thank You.

QUESTION: was there a mention of CPU hit in the video?

MY POINT:  I've always found that crackles/dropouts occurred WELL before 90-ish% of CPU usage WITH USB interfaces.


Been delaying this post for weeks as/and I'd LOVE to be mistaken.


I can say I'm really really wanting a TB4/USB4 Type-C interface.


best wishes to all - lucky to know and love RME

2 (edited by ramses 2022-08-01 12:19:18)

Re: Enjoyed: 'Why we don't need Thunderbolt' video--No mention of CPU hit?

Welcome to the forum.

Sorry, but your expectations / statements sound wrong.

You can't expect to have no audio drops at very high CPU/system loads.
And a high-speed interface like USB4/TB4 would not make any difference.

It depends on many other things, whether audio processing can happen in time so that no audio drops occur.

You will have more advantages if RME uses only the computer interface that is needed in terms of bandwidth, and that depends on how many channels need to be transmitted at the same time.

RME can transmit 68 channels IN/OUT via USB2.
You will also have price advantages, the recording interface will not be so expensive and
EDIT: also the maximum cable length of USB2 (5 m) is higher compared to USB3 (3 m) and Thunderbolt (2 m).

Performance and reliability of RME USB2/3, FW400/800, TB drivers are anyway much better compared to the competition due to the implementation of communication in the FPGA which can be reprogrammed/fixed by firmware updates. No use of 3rd party communication chips, which can be buggy.

It's completely unnecessary to raise a demand for USB4/TB4, it only increases total costs for no reason.

BR Ramses
UFX+, 12Mic, XTC, ADI-2 Pro FS R BE, RayDAT, X10SRi-F, E5-1650v4, Win10Pro22H2, Cub12Pro

3 (edited by rmemius 2022-07-30 02:12:32)

Re: Enjoyed: 'Why we don't need Thunderbolt' video--No mention of CPU hit?

Thanks! Good to know.

My experiences, being somewhat vast with RME PCI and PCIe cards during 20 plus years, are the reasons I'm hoping for eventual clarification, or more facts regarding CPU and USB.

Or, just a Type-C TB4/USB4 interface in a year or two.  (heh-heh)

Seems it wasn't too many years ago it was accepted that USB and PCIe were "different" and shouldn't be expected to perform similarly.  (all said with jokingly friendly smile~)

Yep, I understand enough about that.  I'm on about CPU "max" here and not the sample setting.

I recall my PCI/PCIe cards regularly running at 90% CPU plus, without any issues. Well, none that I noticed audibly. 

That's why I'm longing for TB4/USB4.  Just to know~}

TB4 = as good as PCIe pretty much, I've heard. ~~>>

With delicious, state-of-the-art, TB4/USB4 ~> all in one type-C, full backward compatibility, and a tad of future-proofing!  (why i didn't go for the TB2 RME interface)


Anyways, my BabyFace Pro hasn't had much of a workout on my 2013 i7 PC.

I've just purchased an i9 system and am sure I'll be able to do all I wish.

Not a big deal, my CPU/USB puzzlement, though I'd sure love to see TYPE-C  TB4/USB4.

Thanks again!

Any other comments welcomed of course.

4 (edited by ramses 2022-08-01 12:18:43)

Re: Enjoyed: 'Why we don't need Thunderbolt' video--No mention of CPU hit?

As a side note … You also have advantages in terms of maximum cable length: USB2 (5 m) vs USB3 (3 m) vs Thunderbolt (2 m) and that most computers have USB2 and USB3, not USB4 or Thunderbolt.

If you compare different RME solutions in terms of RTL, you will find that they perform at the same level.
https://www.tonstudio-forum.de/attachme … es-v2-jpg/

I also through in the past that PCIe should be more efficient compared to USB because of the speed and because the protocol overhead is not high. For this reason, I bought a RayDAT PCIe card and used the UFX+ as Preamp connected to the RayDAT. I found out that there was no real difference in terms of RTL and stability.

Then I upgraded to UFX+ and USB3 runs so rock solid that there is also no reason to complain.

I created a large Cubase project with 400 tracks and two Steinberg VSTs in every track and stereo sum, in total ~802 VSTs.
Each of the tracks is a 5-6min wave file. DRAM consumption is around 20 GB in Total and CPU load between 22-30%.
This project runs at lowest ASIO buffer sizes (32 samples at single speed, 64 samples at double speed) up to 96 kHz without audio drops during playback, no matter whether I use UFX+ with (USB3) or RayDAT (PCIe). Furthermore, the CPU consumption is comparable.
Even if PCIe should be a bit more efficient from the protocol, I see no difference even in such large “artificial/benchmark” project, see:
https://www.tonstudio-forum.de/blog/Ent … cks-de-en/

More important is to have a good system with good drivers and settings to have low DPC latencies. Then your CPU cores can quickly react to a workload when CPU cores (where audio related tasks are scheduled to run on) are not blocked too long by badly written drivers.
A while ago, I resolved the last issues on my system … Although DPC latencies were low without any big spikes, I had random audio glitches without any heavy CPU/audio load. It turned out that it was the energy saving of the nVidia GPU, where you needed a special tool powermizer from a Russian developer to be able to turn it off.

BR Ramses
UFX+, 12Mic, XTC, ADI-2 Pro FS R BE, RayDAT, X10SRi-F, E5-1650v4, Win10Pro22H2, Cub12Pro

5 (edited by rmemius 2022-07-30 06:20:31)

Re: Enjoyed: 'Why we don't need Thunderbolt' video--No mention of CPU hit?

Great Tips ramses!  Nice write.

Additionally:
I was aware of graphics drivers and potential issues... though certainly not
"powermizer from a russian developer"

Should be most useful that.


If you're ever running your CPU fully above 80 percent i think maybe USB will show what I've experienced maybe.

I appreciate greatly RME's points too!   AT THIS TIME.


------- WONDERING -------
I'll bet 100 Euros/Dollars virtually (hehe) right now, that within 24 (36?) months RME announces a TYPE-C  USB4/TB4 interface. More motherboards, more standardization then i guess.

Then again -  I've missed the other products, Dante etc. and what they do, so maybe USB2 will be for regular musicians like me until 2030.

My BabyFace Pro will be great for many more years.

And so...  'sigh/smile'   

Many thanks again!

6 (edited by ramses 2022-08-01 11:48:23)

Re: Enjoyed: 'Why we don't need Thunderbolt' video--No mention of CPU hit?

RME's strategy is to use an interface which is sufficient to deliver the bandwidth and which is available on most computers.
As the communication is being done inside the FPGA (self coded, no 3rd party communication chip) a high bandwidth interface like USB4 would require more CPU power from the internal FPGA. This will be avoided, if not really needed.

BR Ramses
UFX+, 12Mic, XTC, ADI-2 Pro FS R BE, RayDAT, X10SRi-F, E5-1650v4, Win10Pro22H2, Cub12Pro

7 (edited by rmemius 2022-07-30 06:33:57)

Re: Enjoyed: 'Why we don't need Thunderbolt' video--No mention of CPU hit?

UFX+ II  ??

was just looking at the UFX+ Anniversary unit - and stumbled across the UFX+ being back-ordered for some months.

Didn't look for more than the one, 'stumbled-upon' mention of the back-order status, and of course who knows!

All is Good!

Re: Enjoyed: 'Why we don't need Thunderbolt' video--No mention of CPU hit?

rmemius wrote:

UFX+ II  ??

What is the question?

BR Ramses
UFX+, 12Mic, XTC, ADI-2 Pro FS R BE, RayDAT, X10SRi-F, E5-1650v4, Win10Pro22H2, Cub12Pro

9 (edited by rmemius 2022-07-30 06:44:56)

Re: Enjoyed: 'Why we don't need Thunderbolt' video--No mention of CPU hit?

Oh, sorry.. yes, just speculating that perhaps there's a UFX+ II next year.

I must be getting a bit sleepy (i work until i need a nap)... hehehe

Thanks for the prompt!

10 (edited by rmemius 2022-07-30 06:59:49)

Re: Enjoyed: 'Why we don't need Thunderbolt' video--No mention of CPU hit?

MY OVERSIGHT: (correcting a bit of my first post above)

at 5 minutes and 30 seconds onward -- a mention of CPU and Direct Memory Access for Thunderbolt IS noted in the RME video.   Most interesting!

YOUTUBE:  Why You Don't Need Thunderbolt For Professional Audio

I guess it's OK to post the link
https://www.youtube.com/watch?v=dSIf4QGYs-c

11 (edited by ramses 2022-08-01 12:05:34)

Re: Enjoyed: 'Why we don't need Thunderbolt' video--No mention of CPU hit?

He said, “In theory” … but at the end of the day, this little advantage has no big impact on the resulting “performance” when using it. I gave you already the link with the RTL of different RME products, all taken at the same sample rate 44.1 kHz:

https://www.tonstudio-forum.de/attachment/2343-ufx-ufx-raydat-latencies-v2-jpg/

As already mentioned: higher bandwidths have a negative impact on the maximum possible cable length and price.

It's more important, that every computer has USB2/3 and that the cables between computer and recording interface can be longer (5 m / 3 m) compared to only 2 m for thunderbolt.

For thunderbolt, it can be even the case that you need to buy the more expensive Apple cables. In contrast to that, you can take any USB2 / USB3 cable for a fraction of that cost.

It's also harder to find good PCs/mainboards with thunderbolt, and I, personally, regard it as a big misconception that thunderbolt can't be upgraded by PCIe cards.
I heard the reason is the integration of Video and you need to have a plug on the mainboard, and it has to be implemented in the BIOS.
Well, but then it's IMHO a far too inflexible and complex “crap design”. For me, it would have been sufficient to get thunderbolt only for data transport. But such capabilities (data transport only) are not covered by this (inflexible) standard.

Another thing is that Intel seems to make it still very hard for companies to implement, why you only see it rarely implemented. I am fully satisfied with USB.

BR Ramses
UFX+, 12Mic, XTC, ADI-2 Pro FS R BE, RayDAT, X10SRi-F, E5-1650v4, Win10Pro22H2, Cub12Pro

12 (edited by rmemius 2022-08-03 07:56:12)

Re: Enjoyed: 'Why we don't need Thunderbolt' video--No mention of CPU hit?

Me: online again here!


thank YOU - SUPERB effort


I don't disagree - I simply wish for both USB4 and TB4 

ALL are backwards compatible as i understand it.


USB CPU penalty and USB lowest possible sample setting penalty
(noticing again it's called asio buffersize)
- added to: the variety of USB motherboard circuitry 'problem potentials' maybe -

heck, i get it, though still betting big-time that by 2023/2024 latest we see a TB4/USB4 interface

simply the marketing points alone suggest it will happen...

uhhh,,,,, exactly at the time i need it.... hehehehehe

gotta laugh at all of it sometimes...


best to all
music

Re: Enjoyed: 'Why we don't need Thunderbolt' video--No mention of CPU hit?

Reality Check. There is no technical reason to implement USB4/TB4. More powerful FPGA would be needed. This in turn would unnecessarily drive up the price of the UFX+. It is already in a price range that not everyone can afford. This does not help anyone.

BR Ramses
UFX+, 12Mic, XTC, ADI-2 Pro FS R BE, RayDAT, X10SRi-F, E5-1650v4, Win10Pro22H2, Cub12Pro

14 (edited by rmemius 2022-08-03 08:35:21)

Re: Enjoyed: 'Why we don't need Thunderbolt' video--No mention of CPU hit?

MY EPIPHANY with respect (most genuine on the "respect" part-)


""" I no longer am guessing on an interface release announcement having TB4/USB4, by, the end of 2024."""


Thanks a-lot 'ramses' for your time and great graphics info and all, on all this!!


I figure my much better CPU and motherboard (i just got) will be brilliant with my BabyFace Pro for quite some while...  until at least Dec.2024-)))

Have yet to turn it on due to work etc.


UPDATE ~> likely 'for sure'  inaccurate~~>>     NOTE: haven't delved into the finer points, though TB4/USB4 i believe doesn't have quite the motherboard header requirements that TB2 did. Something like that.

<~Happy to leave that comment wanting.      Recalling My Epiphany!-)

15 (edited by vinark 2022-08-03 10:29:56)

Re: Enjoyed: 'Why we don't need Thunderbolt' video--No mention of CPU hit?

On the first original post, being able to run the cpu at 98% with pci/pcie is absolutely true. That is if the daw does not make a mess of it. With DAW bench which is all parallel tracks, no busses no fx sends, cubase could load up to 98% at lower buffers and 100% at higher from 1024.
I have not tried it with USB and my babyface fs (yet?). In practical projects, with busses, fx and instruments, PCI still has a small edge over usb, but difference is very small and only at super low latencies. And 96 buffer on usb performs better the 64 on pci and have the same latency. From 256 all equal.

Vincent, Amsterdam
https://soundcloud.com/thesecretworld
Babyface pro fs, HDSP9652+ADI-8AE, HDSP9632

16 (edited by audiogeek7 2022-08-03 23:22:54)

Re: Enjoyed: 'Why we don't need Thunderbolt' video--No mention of CPU hit?

ramses wrote:

...

Performance and reliability of RME USB2/3, FW400/800, TB drivers are anyway much better compared to the competition due to the implementation of communication in the FPGA which can be reprogrammed/fixed by firmware updates. No use of 3rd party communication chips, which can be buggy.

It's completely unnecessary to raise a demand for USB4/TB4, it only increases total costs for no reason.

Hi Ramses...
long time RME user, new to forums. I too have been pondering this video and the point you make about "No use of 3rd party comm. chips..." I would like to argue, because my Fireface UCX definitely has a 3rd party USB chip in it. More specifically a SMSC(which looks like it might be owned by a different company now) USB3250:

https://ww1.microchip.com/downloads/en/ … 3250db.pdf

Thoughts?

17

Re: Enjoyed: 'Why we don't need Thunderbolt' video--No mention of CPU hit?

That is a so called Phy (Physical Interface chip), not a complete interface solution. USB itself happens inside the FPGA. The same is true for other formats. USB 3, FireWire, PCI, PCIe - often such a chip is needed to get access to the physical carrier and protocol.

Regards
Matthias Carstens
RME

Re: Enjoyed: 'Why we don't need Thunderbolt' video--No mention of CPU hit?

MC wrote:

That is a so called Phy (Physical Interface chip), not a complete interface solution. USB itself happens inside the FPGA. The same is true for other formats. USB 3, FireWire, PCI, PCIe - often such a chip is needed to get access to the physical carrier and protocol.

Ah interesting. Thanks for the info!

Re: Enjoyed: 'Why we don't need Thunderbolt' video--No mention of CPU hit?

I've been asking for six months why the Thunderbolt driver is behaving worse on M1 macs than USB but it's been silence, seems contrary to what you would expect. USB2 seems to work fine though, USB3 isn't detected much so I don't use that.