Re: Laptop Looping

[Mark Sottilaro <sine@zerocrossing.net>]

Thanks for this! Great info!  I just watched what Kris did and  
learned...

Sent from my iPhone

On Aug 24, 2007, at 7:44 PM, kkissinger@kevinkissinger.com wrote:

> Quoting midifriedchicken@comcast.net:
>>
>> Audiomulch seems to be PC only, correct? Ive come to the  
>> conclusion  that working with music on a laptop isnt for everybody.  
>> The constant  fear of upsetting the great and almighty CPU is very  
>> frustrating.  Everything you could ever  want is soooo close, but  
>> just out of  reach. Heaven forbid you ask too much of the great  
>> overlord CPU. You  are then punished with a mighty CRACK, BUZZ, SNAP!
>>
> This email is about my battle with CRACK, BUZZ, and SNAP and what I  
> am doing to win the battle!  Hope this isn't too long-winded -- I  
> hope this saves someone some frustration.  Here is my tale:
>
> 1) Choose a laptop
>
> Well, I don't know much about laptops and PC's however I had heard  
> that for audio applications, get as fast a CPU and as much RAM as  
> you can.  So, I found me a nice laptop with a 3.2Ghz Pentium 4 and  
> upgraded it to 2gig.  And when I added a Presonus Firepod and Cubase  
> SX3 and had it working nicely up over 60 tracks with no glitch, I  
> felt like I had ARRIVED.  I dutifully used the MSCONFIG utility to  
> turn off non-essential services (don't need to run a virus checker  
> while music making, etc).
>
> 2) The harbinger of interesting things to come
>
> All was working well.  I did a multi-track project with a stereo mix  
> -- even used VST plugin compressor and reverb.  The only problem was  
> an occasional pop during playback while sizing windows.  Not really  
> a problem -- a minor annoyance.
>
> 3) What you don't know will hurt you
>
> Confident of my equipment, I applied for a performance slot at the  
> electro-music festival.  I decided to delve into surround sound, so  
> I purchase two additional rear monitors.  After all, my Firepod has  
> plenty of outputs and Cubase SX3 has a surround bus.  I'm cookin'  
> now!  And, I decided to include a couple of works that utilize live  
> looping!  Oh... what I got myself into!
>
> There were plenty of things just waiting to BITE me!
>
> 4) Live-looping and latency
>
> I created a work that included some pre-recorded tracks to play in  
> sync with loops (provided by a number of VST delay lines).  Now I am  
> getting a lot of pops, and crackles.  And... I couldn't seem to play  
> in time!  No matter what I did, my rhythm was SLOP.  I thought to  
> myself, "Jeez -- my computer is crackling and I can't play."  And I  
> was kinda torqued because I had tons of memory and CPU power and a  
> firewire port to boot!  Why is the darned thing crackling?
>
> Well, I did some experimenting and found out that the reason for the  
> sloppy rhythm was system latency.  And, I found out that the pops  
> were due to an incompatability between the Firepod and my video  
> chipset!
>
> 5) Throw money at the problem
>
> ... the American way!  An friend on the electro-music forum  
> suggested that I purchase an RME800.  Well, I did and have never  
> regretted it.  No more clicks and pops and latency below 10  
> milliseconds.  Now I'm smiling.
>
> 6) Well, I auditioned for Y2K7 and promised a live-looping set with  
> no pre-recorded elements.   Emboldenned by the Electro-music  
> experience, I incorporate more hardware fx into my sonic repertoire  
> and tie up some more inputs on the RME800.  And, much to my chagrin  
> I have to increase the buffer sizes (and thus the system latency) to  
> avoid crackles and pops.  Of course, the rhythm goes to mush with  
> over 10ms of latency.  I thought to myself, "Dang it!  I thought I  
> had this problem licked!"
>
> 7) Finally, learned more about the PCI bus.  We have all heard of  
> PCI cards, etc.  Well, it turns out that the Firewire port is  
> connected to the PCI bus, along with a lot of other things.  Thus,  
> the cpu and memory may show little utilization however the PCI bus  
> may be straining under the load.   The solution was to create a  
> seperate hardware profile wherein I disable anything connected to  
> the PCI bus that is non-essential.  For instance, I disabled the  
> modem, wireless internet card, ethernet card, internal soundcard,  
> and a few other things.  Also, I reduced my video bit depth to  
> reduce the demands on my meager video card's memory.  Now I can work  
> at latencies <10ms without any glitches.  *whew!*
>
> 8) Enlightenment (for the moment)
>
> A firewire port has a tremendous amount of bandwidth and obviously a  
> CPU that runs at over 3Ghz with 2Gig ram should handle a lot.   
> However, the PCI bus can get overloaded.  Not all drivers and  
> chipsets are created equal.  For instance, the Video chipset in my  
> computer (a Toshiba P35) has limited memory compared to to others  
> and it tends to hog the PCI bus.  Also, the less expensive Presonus  
> unit apparently does more work in the driver than more expensive  
> units that do more of their work in their firmware.  Also, some  
> devices violate PCI standards which can wreak havoc with high- 
> throughput real-time applications such as audio.
>
> The latency thing?  Well, in multi-track recording, the application  
> (Cubase in my instance) compensates for the latency.  The only time  
> one notices any latency is if one is monitoring from the computers  
> output stream (the analogy would be to monitor a recording from the  
> playback head).  (Zero latency monitoring bypasses the computer  
> processing).
>
> Live looping is an application where you are both monitoring the  
> output data and playing in sync with the computer.  So, the latency  
> that is unnoticable in a recording situation becomes a big factor  
> for live looping.
>
> 9) Moral to the story
>
> There are many potential bottlenecks as data travels through a  
> computer and a computer system has numerous devices that compete for  
> resources.  We tend to think of CPU speed, memory, and disk- 
> throughput.   The PCI bus is fast, however beware of devices that  
> are robbing the PCI bus of bandwidth.  In fact, it doesn't hurt to  
> "shop the application" -- that is, choose your software and hardware  
> interface first, check the documentation for known incompatabilities  
> (such as the video chipset), then choose a computer that does the job.
>
> 10) to be continued
>
> I'm sure there will be more chapters to this saga yet to be  
> experienced.  :)
>
> -- Kevin
>