Dumb testing question

[moff]

I was having a discussion with a colleague about whether this would be a reliable test of a young customer's home system. We know that 8 simultaneous tracks should be OK with this rig, but the topic came up of how to demonstrate that.

Gear:

Focusrite Scarlett 18i20 USB interface
Late 2011 MBP i7 with 16 GB RAM, external USB 7200 RPM drive
ProTools 10

We want to test whether the system will handle recording 8 simultaneous audio tracks, say a drum kit, or the whole band "off the floor".

So, here's the question:
No time or space to set up a bunch of gear and/or 8 mics. Would simply attaching the interface to the MBP, with no mics attached, record-enabling 8 new audio tracks in PT10, and recording the noise floor coming from the interface into the 8 tracks tax the system accurately, and show whether it would work?

In other words, does recording the "silence" of the interface's noise floor require the same amount of CPU processing as signal from 8 mics? In my mind, audio is still audio, and the CPU has to process it, whether it's coming in as transient-laden drums at -3 dB or the noise floor at -40.

Thoughts?

[JFreak]

Yes, it is a valid test.

You can also double the tracks to record 16 tracks of silence, or more, which would prove the system would have juice to record that many tracks if only the interface had enough discrete inputs.

I do stress tests like this all the time, when a major update is happening. Try 96 tracks of 96kHz audio for example, if you want to have some fun

How long would you have to record to prove your point? Don't know. I do lots of live gigs so I often need 3-4 hours of recording without a single hiccup. So that's what I do, I set a 4-hour recording for the night and go to sleep. If in the morning I have 4 hours of useless junk, I just hit undo and smile. For you? If a 5-min song is longest you will ever record then perhaps 10min or 15min of troublefree operation would prove it?

[JFreak]

BTW, I noticed you're located in Nanaimo. I've been there once, beautiful place...

[moff]

Thanks. Much appreciated.

My colleague was convinced that recording mic inputs at the "normal" level would be different; my position was 1s and 0s are 1s and 0s, whether that was noise floor or somebody singing...

Yes, Nanaimo's not bad ;-). Lots of scenery, and we've had a fairly warm winter until a day or two ago, when it went down to -9c. Still, better of than those in the middle of the country or further east.

[JFreak]

There is absolutely zero difference whether the AD converter produces all zeros or all ones or anything in between. Only difference is that its lights blink differently

DAW doesn't care what happens in your music. It's just ones and zeros organized in (usually) 24 bit samples. PT is only recording a sample after a sample after a sample and it just goes on until you stop it or there's a problem. Even drawing waveforms is pretty trivial. PT doesn't care if you record silence or square waves, it needs to draw the waveform anyway. This is totally reliable test, what goes in is irrelevant.

Even if you had one input you could set it to input of many tracks and simulate what would happen if you recorded many tracks at once. And I urge you to try! First test with 8 tracks that you likely record in real life. If it's a success, duplicate those and test again with 16 tracks. If it's a success, add 8 more and 8 more and 8 more until you face trouble. Then you know how much your system can do.

And it is likely not the CPU or memory that you run out first. Recording many tracks requires a lot more from the hard drive you're using. For lowly 8 tracks it shouldn't be a problem, but once you go over 32 tracks or so, one not-so-fast spinning drive may have a problem with it, in which case you need more drives and round-robin allocation for the files.