Ableton Forum

Hi there! I am having some serious dropout trouble and I need your opinion.

I have a big live set that I play in gigs. It consists of about 14 tunes that use 8 tracks each = 110 tracks. When I play live I move diagonally (8 channels to the right and then down) from tune to tune using APC40.

I use this setup because each track of each tune has different fx and equalisation that i need to tweak while playing live, so 8 tracks for all tunes will not do.

The clips contained in the set are 48khz 24bit samples, mixed down to 8 tracks from my 60+ track arrangements for the purpose of playing them live. I also use a lot of dummy clips and complex programming to direct the flow. Overall I have 5GB of samples in this one set. The set lasts approximately 1 hour 20 minutes.

I play the gigs on a potent 2.8 Core 2 Duo MacBook Pro with 4GB RAM running Leopard (downgraded from Snow Leopard due to software compatiblity issues), using a NI Audio Kontrol audio interface. I use 48khz output, with a latency of about 13ms.

Ok, so this is the trouble I am having:

The set was playing smoothly until I incorporated the last 3 tunes (adding about 500MB and 20 minutes to the set). I started experiencing dropouts, especially when changing scene (either triggering it through the APC40 or pre-programming it). I got the output down to 44.1khz and I increased the buffer (and thus the latency) but the improvement was marginal, I still get dropouts. Or probably they are not dropouts, but "idle time", that is there is a quarter of a second pause between scenes every now and then.

Live's resource monitor and OSX's Activity monitor do not exhibit "spikes" when this happens. Also I don't get the Disk Overload icon on the top right hand corner blinking. I have deactivated all CPU-consuming settings (HQ pitch shifting, Complex time stretching) and effects (I play mostly with Live's fx, and only a few external plugins) but this has not eliminated the dropouts. I also tried different sets of samples loaded to RAM to ease out the hard disk, and also tried unloading all samples from the RAM. No effect .

Last week I played a gig and had a few embarrassing dropouts . I am now trying to make sure this does not happen again. I researched the forum and I came across the idea that Live has a "limit" as to the amount of information that it can handle in one set, even if not all information is accessed at the same time. Do you know anything about that?

Anyway here are some of the things that I am thinking about trying next:

1) Upgrading to Snow Leopard. Rumour has it it handles resources better, what with it being 64 bit and all.
2) Downsampling my set's clips to 44.1khz 16bit. That would be an enormous compromise on quality, and also a lot of work. (I would have to go back to the arrangements and mix them down to 8 tracks again, as 48khz to 44.1khz conversion would have crap results). Obviously this is something that I would like to avoid.
3) Splitting the set in two. That would introduce a 30 second loading pause in my gig and interrupt the flow. So I would like to avoid it also.
4) Getting the set back to the state before I incorporated the 3 new tunes. That would be a shame.

So, if you had the patience to read this far, you might have some informed opinion you'd like to share. Do you think I am stretching the machine too much, and I have reached its natural limit? Or the limit lies with the software? All I am asking is to be able to play an 1.5 hour gig with reasonable quality without compromising the complexity of my set .

I haven't had the time to contact Ableton's customer support on the matter because I work during their opening times, but I might do it soon.

Cheers in advance for your help

man, I have exactly the same problem.

MBP 2.66Ghz core2 duo, 8GB RAM. Should be beefy enough.

Ok, my live set also runs diagonally, but 159 tracks, 6 returns, lots of plugins and only 2.83GB and runs 16bit 44100Hz samples. I'm also running snow leopard.

I can't get more than 2.89 GB of real memory utilisation out of Live before it quits with an "out of memory" fatal error.

I know Live is a 32-bit program, but surely with a 64-bit OS and 8GB of RAM, Live should be able to use 4GB, right? The other processes - including the os - should be able to run in the other 4GB. Wrong.

I can't get a simple straight answer out of anyone as to why this would be the case.

With PT9 I can get to the full 4GB without any issue, so it's not my computer and nor would yours seem to be at issue.

I use a BCF and a Launchpad instead of the APC40.

My suggestion would be to bit reduce your samples if you can. 16 bit is totally fine as long as you don't need more than 96db of dynamic range. the 144db of dynamic range in 24bit will generally only help if you're looking to save your arse when tracking - lots of headroom is great at the tracking stage. I wouldn't go to 44.1kHz - the extra CPU use is negligible.

That still doesn't get to your real question, though.

As I've said, my wall is memory utilisation.

Splitting the set in two isn't a bad idea as long as you have two independent Live installations.

I use 8.2.1 for my main work (I'll keep upgrading until there's a stable release... Like a problem gambler, I am... Chasing a win that may never come) and the latest beta to always have the capacity to have multiple instances. Even if you don't want to rely on a beta for half of your set, maybe just use it in the middle for a single track while you load the second part in the non-beta instance.

It might be a bit busy on stage for that track, though.

I'd really like to be able to use the full 4GB potential. If you get any answers, I'd love to hear about it. I'll let you know the same.

Ian
SocialJusticeMan

I've hit a similar upper "limit" with a set contaning 110 tracks x 250 scenes and about 3.5Gb of clips.

Some suggestions: Definitely go to 44.1/16bit - I don't see this is a serious trade-off in sound quality for live performance AT ALL, there are many, many more things in a PA chain that will impact your sound quality before 48 -vs- 44.1, 24bit -vs- 16. Going to 16bit alone will cut down your set size by a third.

EQs: Do you need to have totally different settings on every track, that you will manipulate live? Print to audio the EQs that you will not be manipulating. In my live set the 110 tracks are bussed out thru 8 subgroups (kick, snare, bass, hats, perc, Insts A, B, C) and these sub busses are where I have my EQs which are assigned to my hardware.

EFFECTS: Print any and all fx that you will not be manipulating to audio. Also, do you really need any "glamour" 3rd party fx (especially reverbs) which could be substituted for a native Ableton effect for live performance?

INSTRUMENTS: Any AU or VST instruments that you aren't going to tweak live can be printed to audio.

MONO/STEREO: Do your kicks and basses etc need to be stereo? Every time you use a mono sample you cut down the RAM and CPU requirement by 50%.

2 SHOWLOADS: Can you split your showload into 2 Sets and play a CD or an iPod or an iTunes joiner to cover the loading time of the second Set? Probably about 60sec max time to cover.

As a general rule, try to "pre-trigger" as much of your set as possible before you go live. ie, jump through each scene once before your set. This can help with pre-loading or "getting ready" some clips and fx it seems, which can reduce dropouts later.

HTH!

Thanx for all the tips guys.

It is comforting to know that someone else is experiencing the same, this means that there IS a physical limit for Live. I would definitely like to have more info on the inside mechanics of how Live handles digital audio and what its limits are, so I can plan my sets in advance and avoid dead ends.

Indeed I have tried "freezing" many tracks that have redundant fx and instruments, the difference in performance was negligible. Also I had worked out the "preloading" trick before every gig. Switching to mono for certain samples is definitely a good idea, but I am not quite sure it is the answer to my problem. I have reached a limit and and I need a drastic solution.

Resampling everything to 16bit seems like a good option, I don't know about going to 44.1khz though, I CAN tell the difference between 44.1 and 48, especially on the high end of the sound.

Of course I was hoping that there was a "magic wand" solution that could take away the dropouts without having to compromise sound quality. I have paid a lot of cash on software and hardware and I tend to demand the impossible. But, oh well.

I am going to call Ableton and see if they have any tips. If not, I am going to start the process of upgrading OSX and downsampling my set, hoping it will all be up and working flawlessly before my next gig in early March.

One other thing to try is to remove the ghost transient markers - the grey ones - from all samples. each one takes memory, apparently.

Ian

Update:

Upgrade to Snow Leopard had no effect.
Splitting the set in two parts did the trick, all dropouts are now gone. I now have to deal with a 30 sec loading time in the middle of my set. Oh well. I probably prefer this than downsampling everything to 44 - 16.
Gig is up this Saturday, fingers crossed that technology doesn't get in the way again.

Schizophonix wrote:Update:
Splitting the set in two parts did the trick

Yeah, thats what we did too.