Topic: RME & Linux

If I may, I'd like to confirm exactly what the RME options are with
respect to Linux users where (+) indicates Linux drivers do exist
and (-) indicates Linux drivers do NOT exist:

(+) Multiface II & HDSP CARDBUS
(+) Digiface     & HDSP CARDBUS

(+) Multiface II & HDSP PCI card
(+) Digiface     & HDSP PCI card

(-) Multiface II & HDSPe PCI card (pci express/pci-e)
(-) Digiface     & HDSPe PCI card (pci express/pci-e)

(-) Fireface 400
(-) Fireface 800

Just to summarize: no Linux drivers for any Fireface product (which
I suspect has more to do with lawyers than anything else).  However,
Linux drivers do exist for cardbus and PCI HDSP products.

Please forgive me if I overlook the obvious here, but is there any
chance us Linux folks will be able to use RME's HDSPe PCI card (PCI
Express/PCI-E)?  Since there will never be any Fireface 800 in Linux
users' future, will we at least be able to use our PCI-E slots?

I'm itchin' to pull the trigger on some RME product, but just wanna
get my facts straight before parting with any of my "hard earned."

Thanks!  :-P

Re: RME & Linux

rmeuser wrote:

Just to summarize: no Linux drivers for any Fireface product (which
I suspect has more to do with lawyers than anything else).

Umm... No. The reasons have been stated here numerous times. No need for such allegations.

Please forgive me if I overlook the obvious here, but is there any
chance us Linux folks will be able to use RME's HDSPe PCI card (PCI
Express/PCI-E)?

That really depends on if and when an ALSA programmer will take to this task... As you know, we do not develop Linux drivers.

Regards,
Daniel Fuchs
RME

Re: RME & Linux

Wow!  That was quick, Daniel.  Thanks for your feedback.

You were silent on my product list, so I think you are
agreeing that it is accurate and complete.  There are no
additions/deletions or changes needed to that listing.

About the Fireface issue: believe me when I tell you I know
this is a sensitive area right now.  I've been reading all
of the postings (some quite testy and antagonistic).

So I just want to be clear that I have no desire to get
involved in any arguments/debates about that.  Like I said,
all I'm trying to do is get the facts straight and not to
get people all riled up.

Now... with that said, I guess I also have to say that despite
reading all of the postings regarding the Fireface/Linux issue
I still don't understand exactly what the obstacle is.  I know
I am dense, but to me it's either a technology problem or a legal
problem.  In all my readings I was not able to identify it as
being a technology problem.  But I know you are burned out on
this issue and it's a major sore spot, so I will not pursue it
further with you.  (If anyone else could enlighten me, however,
I'd sure be grateful!)

And, yes, I know RME does not develop Linux drivers.  But you
mention that a PCI-E driver "depends on if and when an ALSA
programmer will take to this task."  To me, with my limited
understanding, this means no obstacle exists for open source
community to create a PCI-E driver.  Good news!

So, there is a difference between Fireface product which has no
hope of ever running on Linux and HDSPe (PCI-E) which has hope.
This is an important information that I did not know, so thank
you for sharing that with me!

For the life of me, however, I still don't understand the difference
between the reasons why Linux drivers for one but not for the other.
Can anyone else enlighten me?  hmm

Re: RME & Linux

rmeuser wrote:

For the life of me, however, I still don't understand the difference
between the reasons why Linux drivers for one but not for the other.

The reason is this: Firewire Audio Made by RME
An Open Source driver would mean sharing technical details we don't wish to share...


Regards,
Daniel Fuchs
RME

Re: RME & Linux

Thank you again, Daniel.  Very kind of you to give my post your
attention (and even kinder to reply).  It helps me understand
better and I really appreciate it.  Hopefully it helps some other
people, too!

Take care...

Re: RME & Linux

This link has info on RME cards under linux.

http://www.alsa-project.org/main/index. … Vendor-RME

Regards,
Jeff Petersen
Synthax Inc.

7

Re: RME & Linux

rmeuser wrote:

(-) Multiface II & HDSPe PCI card (pci express/pci-e)
(-) Digiface     & HDSPe PCI card (pci express/pci-e)

The PCIe cards are fully downward compatible, so can be used even with the oldest Windows drivers. So I expect Linux to work with them too.

Regards
Matthias Carstens
RME

Re: RME & Linux

Thank you very much, Matthias, for offering that intriguing bit
of information.

Far from being an expert on coding drivers, based on my experience
I just naturally thought the PCI-E card would need its own special
driver.  I wonder if it really is possible that the old PCI Linux
driver will work just the same with the newer PCI-E card?  It
would be great if it were that simple.

From the alsa-dev list, Mr. Clemens Ladisch mentioned that:

  "Yes.  On the software side, PCIe is compatible with PCI."
  (http://mailman.alsa-project.org/piperma … 03068.html)

However, Mr. Colin Stewart Campbell tells us:

  "Well, except for that the new PCIe cards do all the floating-point
   conversion in hardware and present 32-bit float instead of 32-bit
   integer data on the bus.  If the alsa driver handles this, then ok.
   If not then you'll have to wait. Those running OSX on Intel machines
   with PCI slots already got hit by this format issue."
   (http://mailman.alsa-project.org/piperma … 03070.html)

Perhaps the only way to know is to just buy one and see if it
works?  That seem like an expensive way to find out...  sad

Anyone at all have any further info?  Many thanks!!!

Re: RME & Linux

Hey rmeuser!
Well it looks like you hit it on the head here:

Perhaps the only way to know is to just buy one and see if it
works?  That seem like an expensive way to find out...  sad

However, if you want to be cautious find an outlet (close to you?) that has a good return policy.

Personally, I don't know anything about the PCIe situation. I have a computer which I know is solid and works well. As you probably know, it's not the fastest cpu/mobo which counts in audio, it's the one which has the most reliable/solid implementation and quality chipsets which work best. This will cause the least amount of grief in making your DAW run smooth (which is the goal--right?). If you want your (first?) RME GNU/Linux system to run smoothly then I suggest going down a more trodden path. Think of it another way, if 3x HDSP 9652s don't saturate a well implemented PCI bus, then do you need 500mb/s of bandwidth or whatever PCIe offers--right now?

Notwithstanding, I would have thought that the difference between PCI / PCIe would be taken up by the kernel chipset/bus drivers. Your quote above from the alsa-dev list seems to say there is alot more to the story. But at the end of the day, if your chipset drivers are compiled into your kernel then the first part of the question is already answered.

My own approach with GNU/Linux is always to be a little bit behind the cutting edge, same goes for audio hardware. You will have enough trouble sorting the good graphics drivers from the bad for instance (hint*, matrox is good). There will be some pioneering soul who will have the skills and patience to sort out a problem if there is a solvable one. As we know, RME gear is the top for GNU/Linux and that position is one that has been crafted over time allowing for some truly wonderful hard/software combos to be used. I doubt people will let that change easily.

GNU/Linux

10

Re: RME & Linux

The information from Mr. Campbell is out of context. Our (means RME's) new PCIe cards include some Core Audio software routines directly in hardware to reduce CPU load on the Mac. But this conversion is not active if not activated explicitely. For example it is not active in Windows.

The problem on Mac Intel with PCI had been that our Intel-Mac drivers no longer included the software routines (as the new PCIe have this in hardware).

Regards
Matthias Carstens
RME

Re: RME & Linux

Thanks for your comments, Independent!  As a minority, I think us Linux
folks are kind of the black sheep of the RME family and we can be a bit
of a pain in the behind for the majority white sheep.  We're very grateful
for and appreciative of their tolerance/patience/support.  (I notice that
the Linux posts here on the RME forum do get lot of hits, however!)

Perhaps the only way to know is to just buy one and see if it
works?  That seem like an expensive way to find out...

However, if you want to be cautious find an outlet (close to you?) that has a good return policy.

I believe it would have come down to that in the end.  I was just
trying to see what could be found out up front/ahead of time.  If
I decide to go that route, I'll let you know what my results are.

Also, I agree that PCIe is fairly new and as Linux user you want to
be a little "behind the (hardware) curve" to allow the code to catch
up.  Like you said, stable/solid performance is the goal.  You also
bring up a good point about the 3x HDSP 9652s not saturating the PCI
bus, so maybe PCI should "good enough."

Of course, on the other hand, if it's a brand new DAW you're building
then you naturally will want to try to go with the best technology
available.  I'm sure you are familar with the old adage "buy cheap,
buy twice".  I try not to do that.  I don't purchase a lot of gear
because I try to buy something that I am going to be able to use for
some time into the future and not need to constantly be upgrading ($)
all the time.  Hence my interest in RME.  smile

But, again, as a Linux user that goal will always be constrained by
hardware support; e.g., drivers, etc.  I may have been trying to be
that "pioneering soul" you describe, but now I'm not sure...

Take care!  cool

Re: RME & Linux

Matthias,

Yes, you are (of course) correct.  That was not directly related, but
I thought I would include it only for purposes of completeness.  Thank
you for clarifying for all of us.  We really appreciate your comments!

:-D

Re: RME & Linux

Hi,

well i did it & bought the pcie card for my multiface (the reason is that the motherboard doesn't has pci slots anymore), relying on my "outlets good return policy" and the statements that pcie should be downward compatible to pci. But unfortunately even with all alsa 1.16 packages the hdsp can't be loaded. With Alsa 1.13 this was still possible (hdsploader was succesful), we also saw audio playing back in the rmemixer, but nothing happend in the outputs (neither physically, nor at hdspmixer).

Hmm...so I guess I have to check my outlet guys and switch to the madi interface.

Greetings.

Re: RME & Linux

There was a problem caused by a modified channel mapping. It is fixed for a couple of months now and has nothing to do with the underlaying bus, be it PCIe or PCI. To sum it up: The PCIe variants work nicely.

Re: RME & Linux

I understand not wanting the technology to get into your competitors hands. So, why not follow in NVIDIA's footsteps and make a proprietary driver for the Fireface line? That would make a lot of us extremely happy:)

I would buy a Fireface 800 up in a heartbeat and have it running on a laptop for "on location recording". I would then add a new 800 to my collection, periodically, and build a nice home studio around Linux and RME. I bet RME would profit from it. I imagine RME would get tons more advertisement too, by fully supporting Linux. I don't think ya could loose there. I, for one, would advertise the bits out of RME. I'd fall in love;)

-Adam

16 (edited by meltoner 2012-02-04 13:15:10)

Re: RME & Linux

Linux drivers for the ufx would be fantastic.

For sure, it is a new market growing very fast.

(I am wondering if there is a way, method or mediator to convert - use windows drivers, for Linux.)

Re: RME & Linux

RME Support wrote:
rmeuser wrote:

For the life of me, however, I still don't understand the difference
between the reasons why Linux drivers for one but not for the other.

The reason is this: Firewire Audio Made by RME
An Open Source driver would mean sharing technical details we don't wish to share...


Regards,
Daniel Fuchs
RME

I understand you want to keep your secrets secret. But you must also understand than most musicians are musicians and not geeks, they want their hardware to work, and last but not least, as the linux users base is constantly growing, to have good ALSA drivers for your cards will help your company sell more cards. Note you can keep the source code for your firmware secret if you want to, but you must provide enough information for the driver to work.

Another solution would be you to write the ALSA drivers, like NVIDIA is doing for its driver. That would imply to share the source code at the exception of binary firmwares.

As last option, you can make binary only drivers, but you will have a lot of work because the linux kernel is constantly upgrading, and you will have to make new versions available for each single new kernel version. This is not a problem of the different distributions as stated, but of the kernel version.

This page on the ALSA website summarize very well this subject: Vendor Information

I will thank you for the efforts you made to support linux with some of your products, and stress you that it would be in the common interest of both your company to sell more cards and the linux musicians to have good drivers for more RME products.

Best Regards,
Dominique Michel

Re: RME & Linux

I'm new to this debate, but here's what I see so far.  It's not fair to complain that RME is unwilling to support Linux.  Nobody does, and the reasons seem clear.  On the one hand, no vendor wishes to divulge hardware details to independent device driver writers in order to preserve as much of their technological competitive advantage as possible.  That is understandable and justifiable.  On the other hand, they cannot be expected to write and maintain closed-source drivers that need to be revised and retested every time a new Linux kernel is released.  That's just a black hole of a committment to make (for a very small market share), and therefore a sure setup for failure (worse publicity than not supporting Linux at all), so it's no wonder none of the vendors even start down this path.

Maybe one solution could be to introduce (yet another!) abstraction layer on the system side, wherein the Linux kernel developers would present a stable API for audio drivers that would not break across kernel versions.  Companies like RME could then write closed-source drivers to that API with much less cost and risk compared to the present situation.  Maybe JACK can provide some foundation for such an effort.

Re: RME & Linux

the only truly balanced soulution to these issues (rme and non-rme related) is to truly know the full details without denying any aspect of what money, competition and 'commerce' are - including all law definitions and then to ask yourself honestly, is that a system you wish to maintain and do you wish honestly to keep 'secrets' from others in this way.

this presentation by the originator of the gnu system (more commonly known as linux) makes many valid points that many deny:
http://www.youtube.com/watch?v=E6LsfnBmdnk (part 1)

multiface I +pci card
gnu/linux mint debian edition 64-bit
http://www.infiniteeureka.com