• Jayco
  • ADAT user
  • Offline
  • Registered: 2011-12-04
  • Posts: 12

Topic: Fireface UFX via Firewire... not quite what I had hoped for...

Hey all,

I'm brand new to RME equipment...  just got rid of a Presonus Firestudio 2626 in favor of an RME Fireface UFX.

I recently built a brand new machine in an effort to bring more usability into my studio.  My new machine consists of an Intel P67 motherboard, an Intell i7-2600K 3.4 GHz CPU, 16 GB of DDR3 1600 RAM and a Gigabyte Radeon 6770 graphic card.

The Firewire chipset on the motherboard is TI based and the UFX is the only device connected via Firewire.

I use Cubase 6 64-bit and run on Windows 7 Professional SP1 64-bit.

Via Firewire, I can't get Cubase to even work with the latency at 48 samples.  It immediately starts throwing errors in the RME Fireface software.  At 64 or 96, I don't get any errors, but I get pops and click and around 50% processor utilization if there are any VSTs in the project at all (even though they are 64 bit VSTs).

With USB, I can run at 48 samples, but it has pops and clicks.  64 and 96 work about the same as Firewire with about the same processor utilization.

It seems that I can't get anything reliable until I am around the 128 sample setting... which is not all that much less latency that I was experiencing with my Firestudio 2626.   One of the reasons I upgraded was to lower my latency.

Seems that with as powerful of a computer as I have, I should be able to be at a lower latency without taxing the CPU (multithreaded quad core!) as much as it does.

I've tried going to the legacy Firewire driver in Windows without any noticeable change.  I just feel like I must be missing something, as I expected much better performance from this system.

Thanks in advance for any advice.

Jim

  • From: Detroit
  • Registered: 2011-04-15
  • Posts: 67

Re: Fireface UFX via Firewire... not quite what I had hoped for...

I'd guess it's a system thing over the UFX.

Even on my FF400 I use 64 samples completely flawless on a MacBook Pro 13"

Your system shouldn't even budge at 64 with a UFX

  • Jeff
  • Jeff
  • Moderator
  • Offline
  • From: USA
  • Registered: 2007-05-21
  • Posts: 4,667

Re: Fireface UFX via Firewire... not quite what I had hoped for...

First, setting a 128 sample buffer size does not mean equal latency on two different interfaces, real input, output and round trip latency can and does vary quite a bit.

Second, have you optimized the computer? In Win 7 your Power Scheme should be set to "High Performance". Also, see the link below for tools to find DPC issues which will directly affect audio/video performance.

http://www.rme-audio.de/forum/viewtopic.php?id=1704

Finally, have you tried the unit via USB to see if this works better on your PC?

Regards,
Jeff Petersen
Synthax Inc.

Jeff's Website

  • Jayco
  • ADAT user
  • Offline
  • Registered: 2011-12-04
  • Posts: 12

Re: Fireface UFX via Firewire... not quite what I had hoped for...

ianhind wrote:

I'd guess it's a system thing over the UFX.

Even on my FF400 I use 64 samples completely flawless on a MacBook Pro 13"

Your system shouldn't even budge at 64 with a UFX

I looked at a latency monitor and didn't see anything other than a few required system items.  The DPC checker didn't show anything huge.  I will say that the latency monitor did show hard page fault counts.  I'm not sure what's driving that, as all of my drives are SATA 3 (system) or SATA 6 (audio and samples) and are Western Digital Caviar Black drives.

Jeff wrote:

First, setting a 128 sample buffer size does not mean equal latency on two different interfaces, real input, output and round trip latency can and does vary quite a bit.

My assessment was based on the latency time reported in Cubase for each interface.  At 128 or 256, they both report around the same milliseconds of latency.

Jeff wrote:

Second, have you optimized the computer? In Win 7 your Power Scheme should be set to "High Performance". Also, see the link below for tools to find DPC issues which will directly affect audio/video performance.

http://www.rme-audio.de/forum/viewtopic.php?id=1704

I have indeed set the power settings to high performance.  I've also set the Firewire driver to the legacy mode to see if it makes any difference.

Jeff wrote:

Finally, have you tried the unit via USB to see if this works better on your PC?

I did.  I was able to play at 48 samples with USB, but it had many crackles and pops.  When the crackling went away with USB, was also around 128 samples, so I wasn't sure that it made any real difference.  I have several USB devices (iLok, keyboard and mouse, Presonus Faderport, etc.) so I was trying to avoid adding more to the USB side in case it caused more issues.

I'm pretty sure my DPC latency reported much higher with USB as compared to Firewire, as well.

Thanks,
Jim

  • From: Houston, TX
  • Registered: 2007-05-11
  • Posts: 1,191

Re: Fireface UFX via Firewire... not quite what I had hoped for...

The DPC checker didn't show anything huge

What were your highest peak DPC's for a 5-10 minute period with miscellaneous activity going on?

While I've never ran USB or Firewire interfaces, I've always been able to run my RME PCI/PCIe cards at their lowest settings even back in the P4/DDR1 days - and of course on shiny new i7 hardware.

There's gotta be something going on IMO...

cool

MADIface-XT+ARC / 3x HDSP MADI / ADI648
2x SSL Alphalink MADI AX
2x Multiface / 2x Digiface /2x ADI8
  • From: Connecticut, USA
  • Registered: 2011-08-01
  • Posts: 376

Re: Fireface UFX via Firewire... not quite what I had hoped for...

Hi Jim,
What model Intel P67 motherboard do you have exactly?

latest BIOS?

Do you have the BIOS setup correctly?

I believe the firewire chip on the P67 runs off the PCI buss. I'd have to do some digging to see for sure.
If it does then it could be effected the PCI buss performance issue the P67 chip set design suffers from.

Every manufacturer implements 3rd party chip sets such as a TI firewire chip differently. Having it be a TI chip is no guarantee that it will work. It just increases the odd that it should work.

Some things to try.
make sure you are using the latest BIOs.
I would look into the Turbo Mode and C States settings in the BIOs.
(If you are not up on current BIOs settings then find someone that is before you go mucking about in it. You can really screw stuff up in there.) smile

Pick up a TI based PCI-e firewire 400 card and disable the on board.

Or you can always take the easy way out and just use the USB connection.


Thanks Chris

Chris Ludwig
North East USA Sales | Synthax/RME
www.facebook.com/RMEAmericas
Twitter @RMEAmericas

ChrisLudwig's Website

  • From: Houston, TX
  • Registered: 2007-05-11
  • Posts: 1,191

Re: Fireface UFX via Firewire... not quite what I had hoped for...

Not sure if this is the OP's exact MoBo, but according to http://downloadmirror.intel.com/19482/e … odSpec.pdf - (Page 13) - the Firewire chip is on a P67 PCH PCIe x1 lane.

cool

MADIface-XT+ARC / 3x HDSP MADI / ADI648
2x SSL Alphalink MADI AX
2x Multiface / 2x Digiface /2x ADI8
  • Knut
  • Knut
  • DAW Mastermind
  • Offline
  • Registered: 2007-04-06
  • Posts: 1,618

Re: Fireface UFX via Firewire... not quite what I had hoped for...

My assessment was based on the latency time reported in Cubase for each interface.  At 128 or 256, they both report around the same milliseconds of latency.

This shows clearly, why this is not a scientific value. Different audio interfaces use different safety buffers, so the real latency is not the same. Obviously Cubase calculates a latency time based on the buffer numbers, but don't measure the real (roundtrip) latency. The latest RME driver generation for USB and FireWire uses very small safety buffers, others use big hidden safety buffers on top of the given numbers. Many users measured the real roundtrip latency and found some interesting differences. We've got this discussion here several times in the past.

best regards
Knut

9 Reply by Jayco (edited by Jayco 2011-12-05 05:19:41)

  • Jayco
  • ADAT user
  • Offline
  • Registered: 2011-12-04
  • Posts: 12

Re: Fireface UFX via Firewire... not quite what I had hoped for...

ChrisLudwig wrote:

What model Intel P67 motherboard do you have exactly?

Chris,

I have the Intel DP67BGB3 motherboard, with the latest bios.

ChrisLudwig wrote:

Do you have the BIOS setup correctly?

I believe so, but is there something in particular that I should be looking for to check on this particular issue?


ChrisLudwig wrote:

I believe the firewire chip on the P67 runs off the PCI buss. I'd have to do some digging to see for sure.
If it does then it could be effected the PCI buss performance issue the P67 chip set design suffers from.

Do you know if that's the case with this motherboard?  It seems from Randyman's post that it is on the PCI-e bus, though that document didn't state if it was the B2 or B3 version of the board.

ChrisLudwig wrote:

I would look into the Turbo Mode and C States settings in the BIOs.

What should I be looking for in those settings?  C States and Turbo are enabled on the motherboard at the moment.

I'm quite comfortable in BIOS (and this motherboard has a BIOS recovery mode if all else fails).

I do have another Firewire card (TI based PCI-e) in my old machine, but I am hoping that I won't need to gut that one to make this one work.

Let me know your thoughts on the motherboard.

Thanks much!

Jim

  • Jayco
  • ADAT user
  • Offline
  • Registered: 2011-12-04
  • Posts: 12

Re: Fireface UFX via Firewire... not quite what I had hoped for...

Admin Knut wrote:

My assessment was based on the latency time reported in Cubase for each interface.  At 128 or 256, they both report around the same milliseconds of latency.

This shows clearly, why this is not a scientific value. Different audio interfaces use different safety buffers, so the real latency is not the same. Obviously Cubase calculates a latency time based on the buffer numbers, but don't measure the real (roundtrip) latency. The latest RME driver generation for USB and FireWire uses very small safety buffers, others use big hidden safety buffers on top of the given numbers. Many users measured the real roundtrip latency and found some interesting differences. We've got this discussion here several times in the past.

Knut,

Thanks for the response...  so in essence, though Cubase calculates the same, you think that my actual latency would be much less with the RME equipment?   I will be honest and say that I have not yet done a "real world test" with any of this... I've just been playing back files at different sample settings.

Thanks,
Jim

11 Reply by Randyman... (edited by Randyman... 2011-12-05 06:54:59)

  • From: Houston, TX
  • Registered: 2007-05-11
  • Posts: 1,191

Re: Fireface UFX via Firewire... not quite what I had hoped for...

I know I ALWAYS disable all C-States on my dedicated DAW's.  I'd probably disable Turbo as well, and see if your MoBo can run at the higher "Turbo" multiplier 24/7 (Asus BIOS allows this - at least on all of my DAWs).

You basically don't want the CPU to drop into "power savings" modes of any form when doing DAW work - ESPECIALLY when doing realtime low-latency stuff...

I believe the B2 and B3 revisions of your board would not affect where the Firewire controller connects to the system (can almost assure you the FireWire chip is on a dedicated PCIe x1 lane on your B3 revision).


When you said:

The DPC checker didn't show anything huge

Exactly what were your results (highest DPC peak over a 5 minute span)?  "Huge" is not very specific wink


Lastly, the DAW-Bench forms seem to indicate the Presonus interfaces are not huge "safety buffer abusers" - the reported latencies from Cubase seem to be in-line with their real-world DA/AD ASIO loop tests (they don't list the Firestudio 2626 specifically, but do list the StudioLive and other "DICE" based units.  While their low-latency CPU performance is horrid compared to RME, their round-trip latency seems fairly straight forward compared to what's being reported by the ASIO driver.

http://www.cubendo.com/showthread.php?1 … Base/page5

The StudioLive @ 128 Samples shows In/Out of 3.356/5.351; the measured DA/AD ASIO latency was 9.478ms (only a 0.77ms discrepancy - or around 32 samples - same as RME's safety buffer on the HDSP series).


I would certainly hope your nice Sandy-Bridge system is capable of running modest projects at the lowest buffers.  I wouldn't accept anything less (and never have with RME's PCI/PCIe cards!)

cool

MADIface-XT+ARC / 3x HDSP MADI / ADI648
2x SSL Alphalink MADI AX
2x Multiface / 2x Digiface /2x ADI8
  • From: Connecticut, USA
  • Registered: 2011-08-01
  • Posts: 376

Re: Fireface UFX via Firewire... not quite what I had hoped for...

Hi,

Randyman -

I know I ALWAYS disable all C-States on my dedicated DAW's.  I'd probably disable Turbo as well, and see if your MoBo can run at the higher "Turbo" multiplier 24/7 (Asus BIOS allows this - at least on all of my DAWs).

The newer Sandy Bridge Turbo Mode works fine and can be left on. You do need to adjust some fo the CState settings. There some non audio applications benefit from it being on.


I believe the B2 and B3 revisions of your board would not affect where the Firewire controller connects to the system (can almost assure you the FireWire chip is on a dedicated PCIe x1 lane on your B3 revision).

The board really needs to be B3. B2 had the defective SATA chips so that's bad either way.
It should indicate B3 on the white product sticker on the Intel motherboard box.

Having the TI be on the PCI-e bus "should" be good. I would still try the other PCI- card though to confirm.
It is not unheard of to have a motherboard break things with BIOs updates. smile

Jim -
Other than the info I PM'd you make sure you are running the latest firmware and driver for your USB 3.
Even if your not using the ports.

Chris

Chris Ludwig
North East USA Sales | Synthax/RME
www.facebook.com/RMEAmericas
Twitter @RMEAmericas

ChrisLudwig's Website

  • Jayco
  • ADAT user
  • Offline
  • Registered: 2011-12-04
  • Posts: 12

Re: Fireface UFX via Firewire... not quite what I had hoped for...

Randyman... wrote:

Exactly what were your results (highest DPC peak over a 5 minute span)?

Here's what I got for the DPC counts...

rspLLL64.sys      Resplendence Latency Monitor      856997
ohci1394.sys      1394 OpenHCI Port Driver            668523
ntoskrnl.sys        NT Kernel & System                    252891
dxgkrnl.sys        DirectX Graphics Kernel                250448
USBPORT.SYS    USB 1.1 & 2.0 Port Driver             87619
ataport.sys        ATAPI Driver Extension                 80387
ndis.sys            NDIS 6.20 Driver                          4517
etc...


ISR count was
ohci1394         1142883
intelsmb.sys    462903
hypkern64.sys 462894
hal.dll              429227
ataport.sys      354489
USBPORT.SYS  15084
etc...

I was running audio and trying to make it glitch at the time of recording this data (and it was).

One thing of note...  it seems that when I switch a file to 44.1, I can get it to run without issue @ 48 samples.  Yes, I know that's just reduction in the amount of work the computer has to do, but I thought it was interesting that I must be that close to being somewhat steady.

I also updated the BIOS, the USB3 drives and a few other (ME and Chipset) to make sure that they were the latest.  The BIOS and USB3 were not, but it did not make a difference.

Randyman... wrote:

I would certainly hope your nice Sandy-Bridge system is capable of running modest projects at the lowest buffers.  I wouldn't accept anything less (and never have with RME's PCI/PCIe cards!)

cool

Agreed... that's what I'm trying to get to!

Thanks,
Jim

  • Jayco
  • ADAT user
  • Offline
  • Registered: 2011-12-04
  • Posts: 12

Re: Fireface UFX via Firewire... not quite what I had hoped for...

To add a few more things...

I tried running of USB tonight and I can indeed run at 48 samples, but with several ASIO issues in Cubase (lights the ASIO warning light).  It stayed that way until I got up to 96 samples.  Firewire won't play at 48 samples (unless it's like a single audio file), but I can play at 64 samples fairly cleanly.

It should be noted that ALL test audio is 48k 24-bit.

Thanks,
Jim

15 Reply by phillip fee (edited by phillip fee 2011-12-06 06:50:31)

  • Registered: 2007-05-01
  • Posts: 24

Re: Fireface UFX via Firewire... not quite what I had hoped for...

Jayco wrote:
Randyman... wrote:

Exactly what were your results (highest DPC peak over a 5 minute span)?

Here's what I got for the DPC counts...

rspLLL64.sys      Resplendence Latency Monitor      856997
ohci1394.sys      1394 OpenHCI Port Driver            668523
ntoskrnl.sys        NT Kernel & System                    252891
dxgkrnl.sys        DirectX Graphics Kernel                250448
USBPORT.SYS    USB 1.1 & 2.0 Port Driver             87619
ataport.sys        ATAPI Driver Extension                 80387
ndis.sys            NDIS 6.20 Driver                          4517
etc...

The most useful number to convey to people here is the highest delay in milliseconds/microseconds you are seeing - these are the little vertical bars on the output graph and their height (read off the y-axis) gives the delay experienced in microseconds (from memory). It is these delays which cause clicks on audio when running low latency. The higher the spikes the higher you have to raise the buffer to avoid clicks.

  • From: Athens GA USA
  • Registered: 2012-06-13
  • Posts: 1

Re: Fireface UFX via Firewire... not quite what I had hoped for...

Hey, Jayco. Did you ever find a solution to your issue; I think that's pretty freakin close to what I am dealing with. I just set up my UFX yesterday and am set at 48 sample rate and dealing with the same tiers of issues as you from what I can tell. This so far is unacceptable... I run an 8-core computer with plenty of ram and specs to conquer performance issues. I am getting a CPU usage of about 30-35% in Sonar (even on a fresh project with one track armed). Keep in mind I hooked up my old Presonus FP10 and it played fine with low latency and hardly broke 8% even with VSTs running. There's def something wrong. I was running a Lynx Aurora 8 with no issues also via FW... IF you found a solution please let me know, I have a session in the A.M. and need to be back up and running all week, Blah!#$@# Thanks for any help in advance! -Anthony

guitarlunn's Website

  • From: Germany
  • Registered: 2007-04-10
  • Posts: 9,782

Re: Fireface UFX via Firewire... not quite what I had hoped for...

Does "48 sample rate" refer to the actual sample rate (48 kHz) or buffer size here?
If latency, what did you run the Presonus at, and what kind of device is this?

Regards
Daniel Fuchs
RME

Regards
Daniel Fuchs
RME

RME Support's Website