You are here

Audio Interface - Low Latency Performance Data Base :

For anything relating to music-making on Windows computers, with lots of FAQs. Moderated by Martin Walker.

Re: Audio Interface - Low Latency Performance Data Base :

Postby Goddard » Fri Aug 23, 2013 12:23 pm

Pete Kaine wrote:Well I was going to try and keep away from this, but seeing as Vin's being called out and Sam is being questioned...


Pete, nobody's called Vin out here and Sam's review was certainly not being questioned (unless kudos have now become questions?). I was merely disagreeing about whether Roland develop in-house or buy-in.

You trying to stir up trouble or something? :tongue:

Now, if you have a Roland AI about, you could pop the lid to see whether there's any XMOS or other 3rd party chip in there (sorry, my VS-20 is packed away at the mo else I'd oblige).

Pete Kaine wrote:
Goddard wrote:
Perhaps Roland should seriously consider their participation now in view of Sam Inglis' assessment of the touted "industry-leading low latency USB performance" of their newest Studio Capture interface in the current issue of SOS.

Kudos to Sam and SOS for telling it like it really is (er, isn't?)!


Which issue no. is that exactly, I don't see it in the one in front of me and I'd really like to read it right now!


Sam's Studio Capture review


Pete Kaine wrote:
Goddard wrote:Something is nagging at me about all this. Roland certainly know how to design USB interfaces (iirc they were the first out with USB1 and USB2 interfaces) and have the resources and expertise for firmware and driver development.


I'd contest every statement you made in that sentence. Their drivers on everything I've had my hands on have under performed to the point of being unusable. I've an open case that's been going on for 3/4 months now where their Japan support has buck passed and failed to resolve it time after time.

Uk support is superb, but if you hit a problem that requires Japan to fix it, you can go whistle.


Hey, you calling me out? :tongue:

Well, iirc they were very early out (if not the first ones out) with a USB1 interface (with onboard dsp fx to boot) back when, and also with a USB2 model. But that's not really relevant other than to suggest they did their own development rather than wait for a 3rd party solution to become available.

What's nagging at me is the feeling that Roland should still be capable of producing exemplary gear (and I've a good bit of their stuff boxed up for a move), and yet they apparently haven't. I certainly appreciate that they've kept up driver support for my long-OOP MIDI interface and not just for current production gear, but it's worrisome when development for high end gear like the VS-700 appears to have been dropped or at best put on the back burner and not kept pace with the Sonar platform for which it was designed.

Hearing of open cases going unresolved is troubling. Reminds me of the brouhaha with their Sonic Cell VST editor and VST3 a while back. Perhaps the folks in Hamamatsu are backlogged.

Pete Kaine wrote:
TAFKAT wrote:
Its all inconsequential whether Roland use 3rd party or do in house, IMO, its all in the delivery or lack there of.


Exactly that.


Agreed, what bits get used is not really important, merely of passing academic interest as perhaps indicative of development expertise and resources. Don't really care about who made the pudding, only how tasty it proves to be...
User avatar
Goddard
Frequent Poster
Posts: 999
Joined: Wed Apr 04, 2012 12:00 am

Re: Audio Interface - Low Latency Performance Data Base :

Postby Pete Kaine » Fri Aug 23, 2013 1:56 pm

Goddard wrote:
Pete Kaine wrote:Well I was going to try and keep away from this, but seeing as Vin's being called out and Sam is being questioned...

Pete, nobody's called Vin out here and Sam's review was certainly not being questioned (unless kudos have now become questions?). I was merely disagreeing about whether Roland develop in-house or buy-in.

You trying to stir up trouble or something? :tongue:

Errr... not a whole lot.. :D

Sorry, when I read " Something is nagging at me about all this. " I thought you were questioning the the testing or the results rather than where the controller came from. Miss-understood mate, carry on.

Goddard wrote:
Now, if you have a Roland AI about, you could pop the lid to see whether there's any XMOS or other 3rd party chip in there (sorry, my VS-20 is packed away at the mo else I'd oblige).

Lattice FPGA with a AD Sharc sat next to it. It's not off the shelf in this instance.



Thanks, but I let me subscription lapse a few months ago and it appears someone has pikey'ed the office copy, so I'll have to try and track one down somewhere.

Goddard wrote:
Hey, you calling me out? :tongue:

With you... I honestly wouldn't dare! :tongue:

Goddard wrote:
Well, iirc they were very early out (if not the first ones out) with a USB1 interface (with onboard dsp fx to boot) back when, and also with a USB2 model. But that's not really relevant other than to suggest they did their own development rather than wait for a 3rd party solution to become available.

What's nagging at me is the feeling that Roland should still be capable of producing exemplary gear (and I've a good bit of their stuff boxed up for a move), and yet they apparently haven't. I certainly appreciate that they've kept up driver support for my long-OOP MIDI interface and not just for current production gear, but it's worrisome when development for high end gear like the VS-700 appears to have been dropped or at best put on the back burner and not kept pace with the Sonar platform for which it was designed.

As it happens this started off with a VS-700 issue and the fact it won't talk to the X79 chipset. That's the point I leant that they hadn't tested on anything newer than a X58 rig and expected it all just to work. It doesn't and the more testing I did the more problems surfaced. I'm seeing problems on the mobile chipset as well currently, it truely is shocking. Every avenue I looked down had people being told it's their configuration or some other piece of kit... i had to recreate various isses across 3 or 4 different generations of chipset before they'd consider listening. As for Sonar it picked up various bugs in X2A (and even X2) that don't appear to be getting resolved all that quickly, but at least in those cases the development team are working on them and really I'm happy about that. It's when they throw their hands up and go "yes it's broken, unlucky" I get annoyed.


Hearing of open cases going unresolved is troubling. Reminds me of the brouhaha with their Sonic Cell VST editor and VST3 a while back. Perhaps the folks in Hamamatsu are backlogged.

Do some reading on V-vocal if you want to see something that has never worked properly for a lot of people.
User avatar
Pete Kaine
Frequent Poster (Level2)
Posts: 3172
Joined: Thu Jul 10, 2003 12:00 am
Location: Manchester
Kit to fuel your G.A.S - https://www.scan.co.uk/shop/pro-audio

Re: Audio Interface - Low Latency Performance Data Base :

Postby Goddard » Sat Aug 24, 2013 11:11 am

Pete Kaine wrote:As it happens this started off with a VS-700 issue and the fact it won't talk to the X79 chipset. That's the point I leant that they hadn't tested on anything newer than a X58 rig and expected it all just to work. It doesn't and the more testing I did the more problems surfaced. I'm seeing problems on the mobile chipset as well currently, it truely is shocking. Every avenue I looked down had people being told it's their configuration or some other piece of kit... i had to recreate various isses across 3 or 4 different generations of chipset before they'd consider listening. As for Sonar it picked up various bugs in X2A (and even X2) that don't appear to be getting resolved all that quickly, but at least in those cases the development team are working on them and really I'm happy about that. It's when they throw their hands up and go "yes it's broken, unlucky" I get annoyed.


May I presume you've been dealing with this issue as posted over in the CW forum?

If so, was a discrete USB controller card ever tried in lieu of the onboard USB ports?

The reasons I ask are because, firstly, the OP in that CW forum thread never responded about that, and, secondly, as you are no doubt aware, Intel rather radically changed their chipset USB controller topology when moving from their X58 platform (having a separate I/O controller hub (ICH 10) from which all USB connectivity was provided) to their X79 platform (in which the PCH now integrates all former ICH functions including USB), replacing their ICH's discrete UHCI/EHCI duality and port switching logic with a new PCH "rate matching hub" tier and no discrete UHCI. This change also affected Intel's 6-series (Sandy Bridge) and newer PCHs as used on desktop and mobile.

In this regard, note that the X79 USB controller's "rate matching hub" appears to software as an external USB hub (tier). Note also that Roland do state in the VS-700 manual (Memo on p.27):

The VS-700 may not be recognized correctly if you’re using a USB hub. In this case, connect it directly to a USB port on your computer.


Hmm...

It is rather shocking to hear that Roland hadn't tested on anything newer than an X58 rig (and thus wouldn't have become aware of issues arising on newer platforms), even if X58 had been their original development platform (along with Sonar 8.5).

I'm aware that a number of issues have been encountered using the VS-700 with the newer Sonar X1 and X2 versions. Dunno whether those are USB-related, although considering that all data transfer including all control messages between the VS-700 (and Octa) and PC are via USB, that's probably something to look into (along with all the USB-related errata detailed in Intel's X79 spec update).

Although it's odd that another poster in that CW forum thread had a very similar VS-700 and X79 setup using the same make mobo as the OP and reported things worked at low ASIO buffers.

Just to be clear (don't want you to think I'm calling you out!) I'm not coming to Roland's or CW's defense here, only trying to get a better grasp on things as I also use Sonar and so have a keen interest in this situation.
User avatar
Goddard
Frequent Poster
Posts: 999
Joined: Wed Apr 04, 2012 12:00 am

Re: Audio Interface - Low Latency Performance Data Base :

Postby Pete Kaine » Tue Aug 27, 2013 10:59 am

Goddard wrote:
May I presume you've been dealing with this issue as posted over in the CW forum?


You may presume correctly.

Goddard wrote:
If so, was a discrete USB controller card ever tried in lieu of the onboard USB ports?


One is fitted within the system and it shows no benefits above and beyond the native controller solution.

Goddard wrote:
The reasons I ask are because, firstly, the OP in that CW forum thread never responded about that,


We managed to get hold of one of the CW developers around that time that the OP had dealt with previously and we choose to persue that angle.


and, secondly, as you are no doubt aware, Intel rather radically changed their chipset USB controller topology when moving from their X58 platform (having a separate I/O controller hub (ICH 10) from which all USB connectivity was provided) to their X79 platform (in which the PCH now integrates all former ICH functions including USB), replacing their ICH's discrete UHCI/EHCI duality and port switching logic with a new PCH "rate matching hub" tier and no discrete UHCI. This change also affected Intel's 6-series (Sandy Bridge) and newer PCHs as used on desktop and mobile.

In this regard, note that the X79 USB controller's "rate matching hub" appears to software as an external USB hub (tier). Note also that Roland do state in the VS-700 manual (Memo on p.27):


Yep, a point I spent a few months trying to make before someone got around to building a rig to test it on and realising that it was the case.

The VS-700 may not be recognized correctly if you’re using a USB hub. In this case, connect it directly to a USB port on your computer.


Hmm...


Hmmm... indeed.


Although it's odd that another poster in that CW forum thread had a very similar VS-700 and X79 setup using the same make mobo as the OP and reported things worked at low ASIO buffers.


And that tripped us up for a period that had me banging my head against any solid object that was in reach.

The inital rig that went out had a graphical glitching error within CW that came down to the X2A update and windows 7. We chased our tails trying to resolve that and as we did we discovered different bugs in different configuration setups that seemed to get more and more crazy as we went along. We managed to get the inital graphical issue replicated with CW support and it's being looked at, but otherwise it was fairly usable as a set up and may not have been a game breaker for some users. Over the course of this we've found, documented and submitted half a dozen bugs and issues and in most cases it was questioned regarding the configuration or platform, so at one point I had X58, Z77, &79 and AM3 rigs all set up side by side with matching window images (well as close as I could get) just so I could replicate problems across multiple platforms and prove it wasn't a "configuration problem".


Just to be clear (don't want you to think I'm calling you out!) I'm not coming to Roland's or CW's defense here, only trying to get a better grasp on things as I also use Sonar and so have a keen interest in this situation.


I may have been a little wound up and jumped in prematurely on Friday tbh. I've been dealing with this issue since around March now and it's one of those situations where my inital thoughts were close to the mark but the "configuration" card kept getting played. Mentally it's absolutely drained me and the clients suffered due to serious studio downtime, it's nice to get some varification that I'm not loosing the plot here.

*Edit* Just read Sam's article and it's an interesting read in that it backs up my thoughts on it. I'll just add that my client is working at 48,000/96,000 most of the time and the problems become far more obvious if your working at more the higher more intensive rates.
User avatar
Pete Kaine
Frequent Poster (Level2)
Posts: 3172
Joined: Thu Jul 10, 2003 12:00 am
Location: Manchester
Kit to fuel your G.A.S - https://www.scan.co.uk/shop/pro-audio

Re: Audio Interface - Low Latency Performance Data Base :

Postby TAFKAT » Wed Sep 04, 2013 10:07 pm

Just a quick update.

Image

I managed to get in front of a Mackie Onyx Blackbird this week , which like Focusrite and Presonus , etc, use the TC Applied Dice FW controller and base driver. I am happy to report that the unit has been updated to also be using the TC applied V4.x driver that has dramatically improved performance as it did for the previously mentioned.

The Mackie Unit has lower latency converters than the tested Presonus and Focusrite units , resulting in a lower RTL , placing the above those in LLP Ratings, just below the UFX Firewire.

Results were a little down perhaps because of the 4.1 version driver , but overall a very impressive result which has elevated the unit IMO to a very attractive option.

I know a few reading in were not happy about my previous reports on the Mackie interfaces tested, but I can only report as the units perform on the reference system , hopefully this will ease some of the qualms.

Peace

V.
TAFKAT
Regular
Posts: 244
Joined: Wed Jan 08, 2003 1:00 am
Location: Australia
AAVIM Technology
DAWbench.com

Re: Audio Interface - Low Latency Performance Data Base :

Postby Martin Walker » Thu Sep 05, 2013 3:29 pm

Thanks Vin 8-)


Martin
User avatar
Martin Walker
Moderator
Posts: 15658
Joined: Wed Jan 13, 2010 9:44 am
Location: Cornwall, UK

Re: Audio Interface - Low Latency Performance Data Base :

Postby Goddard » Mon Sep 09, 2013 1:38 am

Thanks for the info, V. Interesting.

FYI, Ploytec are offering a hardware RTL meter now:

http://www.ploytec.com/latenc-o-meter/

And another one for the DAWbench LLP wish-list:

http://www.directout.eu/en/products/exbox.uma.html

:D
User avatar
Goddard
Frequent Poster
Posts: 999
Joined: Wed Apr 04, 2012 12:00 am

Re: Audio Interface - Low Latency Performance Data Base :

Postby TAFKAT » Mon Sep 09, 2013 2:30 am

Interesting G.

Not sure who would be paying that amount of money for a hardware box that does exactly what the Oblique RTL Utility does , and it only has RCA connectors ?

The DAT stuff is interesting, but the website is frightfully slow at my end for some reason today , so not able to have more of a read :-(

V.
TAFKAT
Regular
Posts: 244
Joined: Wed Jan 08, 2003 1:00 am
Location: Australia
AAVIM Technology
DAWbench.com

Re: Audio Interface - Low Latency Performance Data Base :

Postby Goddard » Mon Sep 09, 2013 5:39 am

Hey V, yeah that Ploytec RTL meter is a bit pricey for what it offers, probably because of low volume production.

Here's a USA vendor site page for that Direct Out USB-ADAT/MADI interface, might load faster down there.
User avatar
Goddard
Frequent Poster
Posts: 999
Joined: Wed Apr 04, 2012 12:00 am

Re: Audio Interface - Low Latency Performance Data Base :

Postby TAFKAT » Mon Sep 09, 2013 10:12 pm

Hey G,

I did finally manage to navigate around the DA site eventually, and realised I had been shown the converters last year by a local rep ( who ended up not going with the product line ). I know a little back history of the company which isn't overly important, but with any of these new companys , the actual interface performance comes down to the controller/driver. This will be just another XMOS with a choice of one of the 3rd party drivers I suspect.

V.
TAFKAT
Regular
Posts: 244
Joined: Wed Jan 08, 2003 1:00 am
Location: Australia
AAVIM Technology
DAWbench.com

Re: Audio Interface - Low Latency Performance Data Base :

Postby Goddard » Mon Sep 09, 2013 11:22 pm

Hey V, dunno what you heard but DirectOut do the "i64" slot boards for RME (the founder of D-O was RME's digital guy).

The D-O Exbox UMA uses Ploytec's USB audio/MIDI platform and Class 2 Windows drivers and afaik, another microcontroller (Atmel), not XMOS.

XMOS and Archwave are hardly the only USB audio solution players. CEntrance, Ploytec and Thesycon (amongst others) are also in the USB audio interfacing solutions game, supporting a range of microcontrollers, DSPs, FPGAs, etc. C-Media are in there too, with embedded RISC microcontrollers/logic cores and 3rd party (Thesycon's "TUSB" typically) firmware/stack and drivers, just like XMOS.

Thesycon's TUSB Spy Tool is handy! ;)
User avatar
Goddard
Frequent Poster
Posts: 999
Joined: Wed Apr 04, 2012 12:00 am

Re: Audio Interface - Low Latency Performance Data Base :

Postby TAFKAT » Tue Sep 10, 2013 10:39 am

G,

All you have done is cited the 3rd party driver providers I referred to in my earlier post.

My point still stands, the USB Drivers/Controller ( what ever it is ) for the unit, I doubt will be anything to get overly excited about.

Yes I know who the founder is , but as I mentioned earlier its not overly important as this endeavour has no connection or correlation to RME.

V.
TAFKAT
Regular
Posts: 244
Joined: Wed Jan 08, 2003 1:00 am
Location: Australia
AAVIM Technology
DAWbench.com

Re: Audio Interface - Low Latency Performance Data Base :

Postby Goddard » Sun Sep 15, 2013 3:22 am

No argument that the drivers (and USB stack) matter.

But my point was that, contrary to popular belief, not all current (nor, I would venture, even most) current USB audio interfaces use XMOS' (or Archwave's) silicon. Tenor (GFEC), Freescale (Motorola), T.I., C-Media, Atmel and others' silicon can be found in there as well.

Now, given that 3rd party drivers are available for XMOS's USB audio platform from CEntrance, Thesycon and Ploytec (Propagamma), it would be interesting to see how their performance compares. Maybe something along the lines of what was done years ago concerning the different available bus-mastering IDE drivers.

Another thing which gets glossed over imo is USB Audio Class 2 (UAC2) compliance (and kernel support or lack thereof). Might be interesting to see how the performance of RME's USB interfaces when in "class compliant mode" compares with their performance when using RME's drivers.

;)
User avatar
Goddard
Frequent Poster
Posts: 999
Joined: Wed Apr 04, 2012 12:00 am

Re: Audio Interface - Low Latency Performance Data Base :

Postby TAFKAT » Sun Sep 15, 2013 11:38 pm

G,

Sorry Mate you are losing me on this.

Whether there are other controllers available apart from Achwave/XMOS is of no relevance, it comes down to the performance that the manufacturers are delivering. Also directly comparing the 3rd party drivers is futile unless the manufacturers are willing to change to another driver, which I seriously doubt. Also, I don't have the time and energy to do that level of background investigation, and to be honest , I have no idea how it would even be accomplished.

Re Class Compliance, not getting the relevance in the context of low latency performance for DAW's using ASIO ?

V.
TAFKAT
Regular
Posts: 244
Joined: Wed Jan 08, 2003 1:00 am
Location: Australia
AAVIM Technology
DAWbench.com

Re: Audio Interface - Low Latency Performance Data Base :

Postby Sam Inglis » Tue Sep 24, 2013 12:37 pm

Well...

I was just getting ready to feel smug about having exposed the Roland Studio Capture's poor low-latency performance, when Roland got in touch. They have sent me a second unit for testing and provided some additional guidance as to how to get best results. I have updated the online version of the review here if you want the details, but the nitty-gritty of it is that the driver is set up by default for operation under WDM. To achieve best ASIO performance you need to tick a small box labelled 'Reduce CPU Load'. This isn't enabled by default, and the documentation makes no mention of it having an effect on latency (whenever I've encountered options like this in other interface drivers, they invariably make latency worse!).

With this box enabled, ASIO latency is roughly comparable with many other USB interfaces, and unlike some, I could get the Studio Capture to work at a 32-sample buffer size in my system.
Sam Inglis
Moderator
Posts: 2637
Joined: Fri Dec 15, 2000 1:00 am

PreviousNext