• December 10, 2018 at 12:14 #5616
    no_user no_user
    Participant

    Hi all,

    right now I’m using the USB Card interface of the console which get’s me 5,5ms round trip latency.
    I’d like to get as low as possible, my aim is <3ms.

    So the possible solutions are interfacing over Dante, Madi or Soundgrid.

    #1: Dante Card -> Dante PCIe Card in a Sonnet chassis -> Thunderbolt -> Mac
    I read that Dante PCIe Cards can cause problems with LP2, so haven’t given this any more thought

    #2: Madi Card -> RME HDSPe Madi (FX) in a Sonnet chassis -> Thunderbolt -> Mac
    #3: Madi Card -> Madiface -> USB -> Mac (USB Controller / Bus, so potentially not as low latency as the other solutions)

    #4: Madi Card -> Digigrid MGO -> Soundgrid / Ethernet -> Mac
    I tried this once without success, but according to user MLCOOK19 it can work and it can achieve a ~2ms round trip latency, which would be ideal.

    Do you have experience with one or more of the interfacing methods and can share your thought / experience / RTL values? 😉
    What would you say is the way to go to have a stable and low(est) latency system?

    Thanks guys.

    December 10, 2018 at 23:52 #5617
    no_user no_user
    Participant

    I’m using liveprofessor with the new RME Digiface Dante and a A&H SQ5 with dantecard installed. it works very well and stable.
    The measured RTL at 96kHz is 2.875ms. So, for 2 roundtrips (1 time for groups and channels and 1 time for Mixbus) I get away with 5,75ms. With plugins added, I end up at roughly 7-8ms.
    I had a little problem with a certain plugin (vertigo vsc-2 by plugin alliance), causing clicks, but when removed all was clean and smooth again.
    I am using UAD, Waves, Softube, SSL Plugins right now on a dedicated hackintosh machine I built for this purpose.

    Before that, I also used the desks built in USB-Interface. Liveprofessor reported something like 5ms RTL, but it never felt that way. (Audible flam/delay) When I measured with this tool https://www.oblique-audio.com/tmp/beta.html it showed around 13ms for one roundtrip. I think you can not trust the driver reported latencynumbers.

    December 14, 2018 at 20:27 #5620
    no_user no_user
    Participant

    The best latency results i had so far, for a mobile/touring solution are with my RME HDSP AIO PCIe in a PCIe to Thunderbolt 2 chassis to a MBP and globally, most of Thunderbolt 2/3 (PCIe bus > USB or network) audio card are giving best latencies.

    Last day I wrote down my given LP RTL (i only usey 0ms latency plugins):

    RME HDSP AIO Thunderbolt 2 – MBP Mid-2014 i7 2,7Ghz ElCapitan

    96khz / 64 spl = 3,76ms RTL
    96Khz / 96 spl = 5,76ms RTL

    48khz / 16 spl = 3,85ms RTL
    48khz / 32 spl = 4,5ms RTL
    48khz / 64 spl = 5,85ms RTL
    48khz / 96 spl = 7,19ms RTL

    December 21, 2018 at 07:30 #5626
    no_user no_user
    Participant

    Thanks for your feedback both of you!

    Rapace were these latencies measured using the AES/EBU or the ADAT connection?
    As I am bound to using 48kHz (fixed Soundcraft) the 64smpl setting with your HDSP AIO isn’t faster then the USB Interface I am using right now. I didn’t expect that.

    I just ordered a Sonible ML MIO so I can make whatever Madi connection I need to – I’ll then try to test if possible all 3 RME Madi variants: Fireface UFX+ over TB, MadifaceXT/Digiface Dante over USB3 and HDSPe MadiFX in a TB Chassis.

    Both TB connections should in theory be superior as the go straight to the PCIe bus, the Digiface still has the usb controller that could cause more delay.
    I also looked at the AES Card, Dante (absolutly no information as to how the ASIO/Core Audio drivers perform, only network latency stuff on the internet) and different Madi-TB/PCIe interfaces like Antelope and Motu 112D.

    Do you have any more input on different protocols / connections that you might have tested first hand and why they weren’t satisfying?

    December 21, 2018 at 18:30 #5627
    no_user no_user
    Participant

    //Last day I wrote down my given LP RTL (i only usey 0ms latency plugins):

    As i said in my post those are the LP given latencies, shown in the audio panel (In LP <-> Out LP)

    If you use the 64spl buffer size, then most of the pro audio card will do the job.

    To me RME PCIe card are a matter of stability in a live real time processing situation 😉

    Try with 16spl buffer size to see if your computer/audio card combo got the balls !! 😀

    January 10, 2021 at 14:09 #6337
    no_user no_user
    Participant

    i think you guys will get much more if you will switch to pc. first you’ll get off of the pcie>thunderbolt conversion latency. secong the pc is better than mac (in all the ways and) for the low latency audio processing as well. antelope audio can get to the 1ms roundtrip via asio for example

    February 20, 2021 at 11:41 #6349
    wolferl
    Participant

    [quote quote=5617]I’m using liveprofessor with the new RME Digiface Dante and a A&H SQ5 with dantecard installed. it works very well and stable. The measured RTL at 96kHz is 2.875ms.[/quote]

    Almost the same here but with a Soundcraft VI1000.

    At 48kHz I got around 3,5ms.

    I’m on Mac. 😉

    • This reply was modified 1 year, 11 months ago by wolferl.
    April 4, 2021 at 03:08 #6400
    huub
    Participant

    Interestingly, although I have great faith in RME, I can get lower stable latency with my UAD TB3 solo than with my RME madiface USB.
    Higher samplerates (96khz or higher) and 64 samples runs absolutely fine on my macbook. Latency is almost none that way.
    This is not stable enough with my madiface.

    Huub

    April 15, 2021 at 16:09 #6412
    no_user no_user
    Participant

    That’s why i did precise RME PCIe card (AIO model), PCIe card in a Thunderbolt2 chassis on a mbp.
    Basically Thunderbolt2 is a PCIe port in this situation.
    USB port can hardly achieve such performance.

The forum ‘LiveProfessor 2’ is closed to new topics and replies.