Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Marsellus Fariss]

So I've been testing out Waves Multirack during some non-mission critical shows lately and the latency is proving to be a problem. I'm using it with a M-Audio Profire 2626 at the moment cause it's what I have (until I buy something else) and although that interface sounds good and has tons of i/o the lowest round trip latency I can achieve with stability is around 16 milliseconds which I'm not at all happy with. I'm running at 24/96 on a dual core 2.3 gig machine and only running a few plugs using less then 15% or so of the cpu so I'm sure it's the interface.

Is anybody else using Multirack and what kind of round trip latency numbers are you seeing with which interfaces?

I can see myself getting comfortable with using it but I need an interface with better latency numbers and the only way get good info on that is from other users as far as I can tell.

Latency management within the software between the different plugs is less than easy can be problematic as well. A "rack" has to be assigned to a group and you have to have the group preferences window open to even view what latency the plugs are causing. I honestly think it's a big mistake to sell this software without better latency management features and without some sort of hardware qualification program. A lot of users could get themselves into trouble not being aware of the latency the system is creating. I was talking to an engineer a few weeks back before I had tried the software who was using it mixing at a local club and I asked him about latency problems and he said he was getting around 1.5ms. He was obviously totally unaware of the latency the hardware was causing and that figure came from what the software was reporting for the plug ins.


 

[Marsellus Fariss]

This free tool can be used to calculate round trip latency:

http://www.centrance.com/products/ltu/

It requires you to patch an output to an input of your interface so be careful about feedback loops.

[George Dougherty]

Newer RME gear will do 32 sample buffers for ~1ms round trip.  MOTU 424 cards supposedly can do 16 sample buffers.  The smaller the buffer size though, the less time it gives the CPU to process and move on to the next set and the more horsepower you need under the hood.  The higher the sampling resolution ie, 96K vs 48K, the smaller the amount of time a 32 sample set represents.

[Marsellus Fariss]

Indeed but there's no way for the consumer to know how even an RME interface w/the waves software will perform. I could buy a $1500 RME interface tomorrow and still have too much latency to be comfortable using it live. I could not gain any advantge over my current interface. Probably would but no idea for sure. And once again the only way for me to know how much latency I've got at any guven time is to test it myself with a third party utility and manage it myself in my head.

I guess nobody else around here is using it to report.

[George Dougherty]

Indeed but there's no way for the consumer to know how even an RME interface w/the waves software will perform. I could buy a $1500 RME interface tomorrow and still have too much latency to be comfortable using it live. I could not gain any advantge over my current interface. Probably would but no idea for sure. And once again the only way for me to know how much latency I've got at any guven time is to test it myself with a third party utility and manage it myself in my head.

I guess nobody else around here is using it to report.

Very true.  Question is how much latency the Waves product itself generates. 
We deal with this question in the SAC community a fair amount and a 2x64 buffer setting including AD/DA steps comes out to about 7ms of latency.  2x32 cuts that to 3-4ms.  I was under the impression that all of the waves plugins for the MultiRack were zero latency, like SAC forces you to use.  Seems like most plugins that add latency add enough that I'd never consider using them live.  Otherwise, 3-4ms on a channel insert or an aux fed reverb should hardly be noticeable, especially as a FOH rack.

I know of several people in other forums that use them and absolutely love them.  If I were going this route, I'd go MOTU with a preference for RME simply due to the stability of the drivers.  I've played around with a number of interfaces and SAC.  The only other one I've found tolerable is the M-Audio ProFire 2626.  Most of the firewire and lower grade interfaces gave me fits during testing.  I can possibly imagine my inserts dropping out live without a panic attack, but personally I'd put as much importance on it as most do with SAC rigs that handle the full mix.

[George Dougherty]

Sorry went back and read the first post and noticed you're testing with a 2626.  I can reliably run mine at 2x64 buffers in SAC packed out with 24 channels of processing on an IBM 2GHz Core Duo laptop.  Round trip is about 7ms.  1x64 should be workable if the MultiRack code is halfway decent.

Another suggestion from the SAC world.  Find and download dpc latency checker.  It'll measure background processing latency for your system.  If you see spikes into the yellow or red, then you've got stuff stepping on the CPU in the background.  Wireless networking, unused network cards with no cable plugged in, APCI battery management, all kinds of things that could be causing problems.  I had a Toshiba laptop that couldn't run SAC reliably unless I disabled every piece of unused hardware, all networking features and it still gave me occasional problems.  From experiences on the SAC forums, some systems will never play nicely.  Graphics cards can also be problematic.  If you're using a laptop you're SOL.  Otherwise you can try swapping out the card.  Most find ATI cards are less problematic than NVidia, but there are plenty of people using NVidia cards in SAC rigs as well.

I'm guessing you're running Windows.  If so, another item to look for online are the BlackViper guides to streamlining Windows.  Basically you turn off all the flashy UI features and disable all unnecessary services.

The biggest lesson was that multiple cores are not a help in realtime processing.  With current hardware architectures, only one core can access the same bit of RAM at any one time.  You might try booting with just one core enabled if you can.

[ProSoundWeb Community]