Track-Count on MacBook Pros

[dix]

In the early days of DAWs track-count was always the no 1 topic. These days track-count usually far exceeds the needs of most productions, but this situation is a bit different:

I'm trying to put together a mobile rig for an on-site surround mixing project - I'm thinking a 2015 MBP will cover it. The playback will occur in a one-of-kind multi-channel theatre, so I will need to create the final mix in the venue. My time working in the venue is limited so I'm planning on doing as much sub-mixing as I can in my 5.1 studio and bringing the stems to the final in the venue.

My question is, what kind of track-count can one expect from the latest MacBook Pros (playback only). With any luck I'll only need to run minimal EQs on the a few tracks. No VIs or reverbs (I'll create multiple options of reverbs and ambience stems to try). I'll have an OWC Thunderbay 4 (striped to Raid 0 probably). With the MX, FX and DX I could easily hit 200 tracks (24bit 48k). I know it's a rather specific scenario, but can anyone tell me if a setup like that would have a problem on the latest MBPs?

Thanks!

[mikehalloran]

My question is, what kind of track-count can one expect from the latest MacPros (playback only)

Do you really mean MBP?

[dix]

Oops. Right. MBP. ...editing op. thx.

[mikehalloran]

I should think that an iMac and a MBP should be roughly the same with similar amounts of RAM and processors. You want an i7 Quad, of course, plus as much RAM as you can handle.

200 audio only with little processing? I don't know but you would want the largest output buffer setting.

Are you going to mix live tracks with this? If not, why wouldn't you just play back a 5.1 mix?

[bayswater]

I was wondering about track count too. So I did a test project on my MBP, early 2010 with i5 and 16G RAM, which would have a zillionth of the power of a new MBP.

There are 192 tracks with a few bars of empty soundbites at the start (to see if silence uses less power than sound)

After that I added soundbites with a signal in them (tapping on a mic). I staggered their start times so more would be playing as playback progresses. They are set up so about 100 are generating a signal at any time. See picture below. The pattern shown extends downwards to 192 tracks.

The first 24 tracks have the 4 band MOTU EQ with three bands activated.

There are 18 Auxes, each with the MOTU Dynamics plugin, routed to Main out.

All 192 tracks are routed to one of the 18 Auxes.

Buffer is set to 1024.

On playback the Audio performance meter stays below 25% and is at 15% to 20% most of the time.

So I guess the answer is, a new MBP will be half asleep most of the time.

(I might just try this on Logic to see what happens

[dix]

Oh, man. Thanks bayswater! This is exactly the info I was hoping for.

Good that you tested with Auxs too. That's something I forgot in the scenario I described. The busing on the project might get extensive, which definitely eats some cpu.

I'm assuming your test project is on a FW external HD - I don't think 2010 Macs had Thunderbolt - so the tb raid will be even better I bet.

So, yes. It looks like an i7 MBP should cover it easily.

...this tells me what I need to know and then some, but if it were me I'd be curious just how far it could go. what happens if you Duplicate All, so 396 tracks?

I did find this 2009 SOS article on SSDs where Logic could play back 255 96k 24bit files (Logic's max track count) from the, then new, internal SSD. So, again yes. A new MBP should cover it. http://www.soundonsound.com/sos/dec09/a ... 09.htm#Top

Thanks again for taking the time to test this out. It's appreciated.

[bayswater]

I used the internal drive, (I usually do audio on my iMac, not this MBP) but the samples are quite small, so I assume they are all loaded in RAM. Later I'll be doing something with bigger and more varied soundbites. That will probably bog things down a bit, but with a TB disk, I'd bet things will be fine. At this point, it looks like the number of tracks per se is not much of an issue. And for live stuff, it's hard to imagine needing more than 16 submixes.

[bayswater]

BTW, that internal drive is a SSD.

Anyway, to see if this performance is simply a matter of having only a few physical sound files being accessed many times, I modified the project by taking a basic audio test file and making 64 physical copies. I put these into the project in the same staggered pattern so that at the peak, there are over 150 tracks playing audio at once. This would consist of each of the 64 audio files being played twice, and some being played three times simultaneously.

I did this on my 2009 iMac, with i7 and 16 G RAM with buffer at 2048. Again, probably not as powerful as a 2015 MBP. Audio is on one FW 800 drive.

At the peak, the 8 processor threads show the following % CPU load: 37, 3, 24, 1, 21, 1, 29, 3 (average 13%) At idle, the numbers were: 17, 0, 5, 1 ,4, 0, 3, 0 (average 4%). So again, simply playing back a lot of audio with a few plugins barely moves the meters.

From what I can tell looking Activity Monitor, disk activity was about a third of the maximum throughput level measured by Blackmagic.

TTTT, I'm amazed at what this old Mac can do.

[bayswater]

Just to compare: Same project in Logic, ported via AAF, using a 1024 buffer (2048 not available), using stock Logic plugins to replace stock DP plugins, idle CPU average is 4%, same as DP. Peak CPU average is 8% versus 13% for DP.

Bring on 9.1 !!!

[dix]

Cool! The i7 iMac test might be closer to what I could expect from a 2015 MBP. I'm not even sure I'll need the SSD raid now. Amazing that your test project on the MBP was running from the internal SSD. I didn't even think that was legal!

Thanks again for all this testing. I'd say you can stop now, but it sounds like your having fun

[bayswater]

I started this a while ago and wanted to get it finished. Your post prompted me to get on with it.

On using an internal SSD, I don't know why there would be a problem. I thought the issue leading to recommendations to use dedicated drives was contention for bandwidth between the OS, the audio streams, and sample streams. My internal SSD is 8-10 times faster than the original spinner, so it would seem to remove the need to have dedicated drives for samples and audio.

As a closer, I copied the project back to my MBP, and played it with Activity Monitor running. The system CPU was at about 5%, and DP CPU usage peaked at 23% with 156 tracks active.

[dix]

Are you going to mix live tracks with this? If not, why wouldn't you just play back a 5.1 mix?

[sry. missed this post the other day] The venue will have a 26.1 channel system, with speakers above and below and a 180º image (it's an amusement park ride essentially). I really can only create mix options in my horizontal 5.1 studio. Placement and panning will take place in the venue. And since the speakers in the installation are TBD (Meyer's probably. Don't know what model(s?)) I'm sure I'll need some EQ.

[dix]

Well, the aforementioned project has ended up being over 220 stereo tracks, so I think I'll be okay with a 2015 MBP. I don't have access to the newer MBP so I tried running the project on my 2011 MBP with DP8 installed and, amazingly it worked! It was sluggish, but played back with the Playback performance at around 20%.

One weird thing is that that the waveforms didn't draw. I'm not sure if that fact makes the test less valid or not. Maybe they would have eventually, but there was no Background processes going so I don't know what was going on there. A DP8 thing m'be...since it was only a rough test I'm not inclined to trouble shoot.

Anyway, hooray for technology! The me of 1996 is very jealous of the me of 2016