• Registered: 2007-05-02
  • Posts: 316

Topic: The hole in RME's PCIe lineup

I'm ready to upgrade to a better converter with more features, better conversion quality, etc, but I'm stumped. I like (and need!) extreme low latency and the best low latency is still connecting directly to PCIe. There is always a penalty for connecting to Firewire or USB, and I've avoided those RME offerings mainly for that reason, but also because in the past these interfaces have shown hardware dependent performance and operational issues. I look at something like the UFX and its feature set is so much what I could need; it would otherwise be so perfect. Dammit.

So what is there that connects to PCIe that I could improve to? There's the high road: MADI, AES, RayDAT, ADI-8 DS/QS, MicExtacy... And there's the low road: Multiface II, or Digiface to conversion. Why is there nothing in between? BTW: I work in stereo and 5.1 so a minimum 8x8 analog I/O is necessary.

It seems to me there would be a place for a complete package like the UFX that connects to HDSPe PCIe. That to me is the hole in RME's lineup and sadly, it's causing me some grief. 

Thoughts? Suggestions?

PC1 = HDSPe PCIe: DF-ADI-8 DS / HDSPe PCIe: MF2
MBP = HDSPe Expresscard: MF1
  • Registered: 2007-05-02
  • Posts: 316

Re: The hole in RME's PCIe lineup

No thoughts? No suggestions? I'm sticking with RME that's for sure... they got me supporting my Multifaces for 10 years+ smile

PC1 = HDSPe PCIe: DF-ADI-8 DS / HDSPe PCIe: MF2
MBP = HDSPe Expresscard: MF1
  • From: Germany
  • Registered: 2007-04-10
  • Posts: 9,770

Re: The hole in RME's PCIe lineup

undertone wrote:

There is always a penalty for connecting to Firewire or USB, and I've avoided those RME offerings mainly for that reason,

You shouldn't have... ;-) Particularly with the RME USB drivers, there is no inherent "penalty" or loss of performance. Your assumptions are based on prejudice against USB from the days of yore... cool

but also because in the past these interfaces have shown hardware dependent performance and operational issues.

Every audio interface or card depends on the hardware surroundings, i.e. the computer it runs with... This is no different for PCI/e based systems.


Regards
Daniel Fuchs
RME

Regards
Daniel Fuchs
RME

RME Support's Website

  • Registered: 2007-05-02
  • Posts: 316

Re: The hole in RME's PCIe lineup

Thanks for replying!

And yet if you take a look at this thread, the hard data seems to disagree. None of the USB/Firewire intefaces can offer 32 sample buffers and with a 64 sample buffer the performance figures aren't quite good as the hard wired interfaces. Also, USB actually trails slightly behind Firewire in performance. Things however level off as higher buffer sizes are used.

I'm now debating whether I should go "Some Card" (MADI?) to ADI-QS or UFX, but neither is a perfect choice; it's either inconvenient quality or convenient compromise. Tough choice!

PC1 = HDSPe PCIe: DF-ADI-8 DS / HDSPe PCIe: MF2
MBP = HDSPe Expresscard: MF1
  • From: Houston, TX
  • Registered: 2007-05-11
  • Posts: 1,191

Re: The hole in RME's PCIe lineup

At this level, even the latency of your AD/DA conversion comes into play!!!

AFAIK - The DawBench report hasn't been updated to include the new UCX.

I believe the UCX will easily best the UFX, and will likely be right on the heels of the "Benchmark" HDSPe/ADI8QS setup (the UCX uses similar ultra-low-latency AD/DA converters as the ADI8QS where the UFX is using slower AD/DA converters that have more conversion latency).

Taking conversion latency into account - and presuming you aren't planning on dropping $3000 on a single ADI8QS - the small buffer discrep from 32 Samples (with HDSPe PCIe) with "standard speed" AD/DA converters vs 48 Samples over USB is easily negated by the faster AD/DA converters of the UCX smile

I'm not sure if the UCX would fit your I/O needs w/o an additional ADAT AD/DA...

I'm plenty happy with HDSPe MADI and SSL AD/DA conversion.  The SSL Alphalinks aren't the fastest converters around, but with the 32-Sample ASIO buffers, it is splendid!  Plus I can use the HDSPe MADI's Low-Latency Headphone Out for at least one performer's mix (The HDSPe MADI card uses a similar low-latency Headphone DAC as the ADI-8QS and I believe same as the UCX).  Allows me to get at least one of my mixes down a little lower with the faster DAC (generally the drummer's mix)...

cool

MADIface-XT+ARC / 3x HDSP MADI / ADI648
2x SSL Alphalink MADI AX
2x Multiface / 2x Digiface /2x ADI8
  • Registered: 2007-05-02
  • Posts: 316

Re: The hole in RME's PCIe lineup

Thanks for the heads up on the UCX, Randyman. The single ADAT lines do spoil the I/O requirements for me @ 96kHz but the rest of the package and the latency benefits could have fit the bill. Again, it's the high road or the low road: card + ADI-QS > $4k or one of these all in ones. I'm going to reasses what I can do with my current group of RME cards and see if I can mix and match me a solution with a newer box.

My two concerns other than latency are being able to transparently work at 44.1/48 and 96kHz without wiring issues (I don't really like ADAT for that) and of course conversion quality.

The most startling personal experience I had with conversion quality was when I got the original Nuendo-branded ADI-8 (44.1/48) about 5 years ago and drove it from my Multiface 1 ADAT. The improvement in definition, detail and positioning was really noticeable. I could never go back to less than that now and I'm concerned that the cheaper all-in-ones may not match up. Any comments on that?

BTW, thanks to all who are contributing to my public shopping experience.., wink

PC1 = HDSPe PCIe: DF-ADI-8 DS / HDSPe PCIe: MF2
MBP = HDSPe Expresscard: MF1
  • From: Houston, TX
  • Registered: 2007-05-11
  • Posts: 1,191

Re: The hole in RME's PCIe lineup

I'd HAVE to assume the UCX's AD/DA's are light years ahead of the original MF-I.  As far as a subjective SQ comparision of the UCX against the ADI-8 Pro (48K) - that will be down to your ears and your specific environment as published specs alone will not show the improvements you might expect to see/hear from either unit.  I'm not thinking you will get an appreciable $3000 improvement over the UCX by going with an AES HDSPe card and an ADI-8QS - but some people hear way more than I do :-)

Also - As long as having your CH 7/8 Outs available as an unbalanced TRS pair on the front panel is OK, then the UCX will indeed give you 8 I/O @ 96K w/o any additional boxes (according to the specs).

The UCX seems like a killer modern option - and your get TM-FX as a bonus!!! :-)  It's also easily portable - can be used on any computer you can access (unlike a PCIe card).  If I only needed 16 I/O's or so, a UCX and/or UFX would be high on my list - and I've always been a PCI/PCIe guy from day one!!!

cool

MADIface-XT+ARC / 3x HDSP MADI / ADI648
2x SSL Alphalink MADI AX
2x Multiface / 2x Digiface /2x ADI8
  • Registered: 2007-05-02
  • Posts: 316

Re: The hole in RME's PCIe lineup

In have to say that the UCX certainly looks really good, but I do need at least 10 channels of balanced outputs to handle my needs. I could get those additional 4 through a single ADAT at 96k with a 4 channel converter... but that's also missing from the RME roster! It feels like I'm falling through the cracks looking for a solution. BTW, ADAT seems so archaic to me these days...

Anyways I've got enough food for thought now. I'm going to mull things over, possibly look at what I can rummage elsewhere too. I updated my sig to give you some idea what I'm currently running. Thanks for the tips!

PC1 = HDSPe PCIe: DF-ADI-8 DS / HDSPe PCIe: MF2
MBP = HDSPe Expresscard: MF1
  • Registered: 2008-01-01
  • Posts: 3,025

Re: The hole in RME's PCIe lineup

undertone wrote:

None of the USB/Firewire intefaces can offer 32 sample buffers and with a 64 sample buffer the performance figures aren't quite good as the hard wired interfaces. Also, USB actually trails slightly behind Firewire in performance. Things however level off as higher buffer sizes are used.

Last time I checked the internal safety-buffers (which cannot be changed by users) of the USB driver were smaller than the FW ones (depending on OS and input vs. output). So you need to take a look at the effective latency in milliseconds, not just audio buffer size in samples. wink

10 Reply by undertone (edited by undertone 2012-06-14 13:10:07)

  • Registered: 2007-05-02
  • Posts: 316

Re: The hole in RME's PCIe lineup

Timur Born wrote:

Last time I checked the internal safety-buffers (which cannot be changed by users) of the USB driver were smaller than the FW ones (depending on OS and input vs. output). So you need to take a look at the effective latency in milliseconds, not just audio buffer size in samples. wink

Hi Timur, that's also clear from the information Randyman provided. The UCX is clearly in the lead in this department I've started working with VEP 5 and it's going to greatly simplify my hardware routing needs; the idea of simplifying my audio chain is very appealing and if I'm not mistaken I could temporarily use one of my MF1's via ADAT for the extra 4 outs I need @ 96k till I find something better.

However, strictly speaking, the specs of my old ADI-8 are still better than the UCX (which actually has THD figures closer to the MF2). This is still of some concern to me because even if I did eventually get a better 8x8 box, I would never be able to get all 8 I/O @ 96k with the UCX. I'm not sure how RME can call these "Mastering Grade" converters...HeadScratch

PC1 = HDSPe PCIe: DF-ADI-8 DS / HDSPe PCIe: MF2
MBP = HDSPe Expresscard: MF1
  • Registered: 2008-01-01
  • Posts: 3,025

Re: The hole in RME's PCIe lineup

HDSPe AIO + analog expansion board makes for a total of 6 analog inputs + 8 analog outputs.

12 Reply by undertone (edited by undertone 2012-07-10 23:00:08)

  • Registered: 2007-05-02
  • Posts: 316

Re: The hole in RME's PCIe lineup

Just a quick last word form me on this thread. After some thought I've decided to simplify my setup and stay mostly with what I have, continuing to work at 44/48 for now, and leveraging the flexibility of VEP5 for inter-computer audio.

I would love to see a PCIe-based multifunction box like the UFX or UCX, and I personally think that's a still a hole in the RME lineup. But I've decided to wait and see if anything does pop up in that direction, or for a time when my budget will allow the evolutionary step up to better gear.

Thanks to all who contributed to this thread! :-)

PC1 = HDSPe PCIe: DF-ADI-8 DS / HDSPe PCIe: MF2
MBP = HDSPe Expresscard: MF1