identical performance after bouncing everything to audio

This forum is for discussing Reason. Questions, answers, ideas, and opinions... all apply.
Post Reply
Palmeira
Posts: 111
Joined: 18 Jan 2015

31 Jul 2017

Hi all

I've got an overly complicated song that I thought would be much easier to mix if I bounced almost everything to audio, which I did track by track. I've now got 63 tracks of audio and only 2 rack synth tracks left that are playing a not too complicated bassline and some synth chords for a vocoder on one track of vocals. I have removed entirely all devices and tracks that have been bounced and altogether this seems to me to be the equivalent of utilising the freeze function employed in other DAW's.

Apart from simplifying mixing I thought (quite reasonably - no pun intended) that this would also free up a great deal of audio performance and enable me to whack the buffer size down and record some additional guitar.

But no.

Performance seems to be identical and I still can't run the song unless I run at the highest buffer rate setting (4096 samples on a Balance interface).

Can anyone advise why I'm seeing this behaviour from Reason? It doesn't seem to make much sense but maybe I'm missing something obvious!

Cheers

Aiden

Hauser+Quaid
Posts: 147
Joined: 06 Jun 2017

31 Jul 2017

Hmmm... did you disable stretch on all of the audio files?

Palmeira
Posts: 111
Joined: 18 Jan 2015

31 Jul 2017

Hauser+Quaid wrote:
31 Jul 2017
Hmmm... did you disable stretch on all of the audio files?
HI mate. Took me a minute to realise where you were coming from with that but yeah disabled stretch on all audio and same result....

User avatar
miscend
Posts: 1955
Joined: 09 Feb 2015

31 Jul 2017

Did you print all your effects to the audio files? Sometimes it's the effects.

Palmeira
Posts: 111
Joined: 18 Jan 2015

31 Jul 2017

miscend wrote:
31 Jul 2017
Did you print all your effects to the audio files? Sometimes it's the effects.
Yep all effects printed and all inserts removed along with all devices and tracks!

User avatar
selig
RE Developer
Posts: 11685
Joined: 15 Jan 2015
Location: The NorthWoods, CT, USA

31 Jul 2017

Palmeira wrote:
31 Jul 2017
miscend wrote:
31 Jul 2017
Did you print all your effects to the audio files? Sometimes it's the effects.
Yep all effects printed and all inserts removed along with all devices and tracks!
How many audio tracks you talking about, and at what sample rate? Has your system ever struggled with that many audio tracks alone before?

Tip:
Best way to do this in the future is to bounce all tracks at once to a new folder (Bounce Mixer Channels; All Except Fader Section, Bounce Song (start to end marker), Bounce to Audio Files on Disk).
Don't Normalize, don't add Dither, keep files at 24 bit and the original (or destination) sample rate.

Then start a new empty (TOTALLY empty) song file, and drag all the files into that new file (also all at once).

Beats doing it one at a time, for sure!

Another tip - if you want all the files at the same reference peak level, then select all imported clips, normalize them, then set all of their clip gains to -12 dBFS. Instant way to know exactly what level all your tracks are at! I do this all the time with imported audio files, especially when they don't originate on my system (since I already keep all levels peaking around -12 dBFS). BTW - the original files are still on your drive. If you ever want to get back to them for any reason, just drag them back into the song file!
:)
Selig Audio, LLC

Palmeira
Posts: 111
Joined: 18 Jan 2015

01 Aug 2017

Hi Selig,

there's 63 tracks at 24 bit 44.1 Khz. Mostly I'll have around the same amount of tracks but a mix of audio and midi and would routinely have to up the buffer rate to be able to cope. I wouldn't usually bounce everything but thought this may be beneficial as things were getting overly complicated. I just wouldn't have really thought that the same song with most tracks being audio would have put the same strain on the computer as having synths, drum machines and fx playing the same parts.

I do get what you are saying about bouncing everything separately and using a completely new song too. I'll probably try that next time but I wanted to leave myself with a little flexibility with the vocal tracks and utilise the existing group channels etc rather than print everything.

Cheers for the tips and have a great day

Aiden

User avatar
selig
RE Developer
Posts: 11685
Joined: 15 Jan 2015
Location: The NorthWoods, CT, USA

01 Aug 2017

Palmeira wrote:Hi Selig,

there's 63 tracks at 24 bit 44.1 Khz. Mostly I'll have around the same amount of tracks but a mix of audio and midi and would routinely have to up the buffer rate to be able to cope. I wouldn't usually bounce everything but thought this may be beneficial as things were getting overly complicated. I just wouldn't have really thought that the same song with most tracks being audio would have put the same strain on the computer as having synths, drum machines and fx playing the same parts.

I do get what you are saying about bouncing everything separately and using a completely new song too. I'll probably try that next time but I wanted to leave myself with a little flexibility with the vocal tracks and utilise the existing group channels etc rather than print everything.

Cheers for the tips and have a great day

Aiden
I don't know if you're familiar with the Bounce Mixer Channels command, but it's quite flexible. You can exclude the vocal from the bounce simply by unchecking that channel, and choose to include the inserts or not, the fader/pans or not, etc.

There's no advantage in bouncing 63 tracks one at a time IMO!

One additional thing you can try is to delete any silence on all tracks to see if that gains you anything. Odd that CPU is as taxed with audio tracks - should tax the hard drive more and the CPU less (as I'm guessing you were expecting as well).

Hard to know more without seeing the song in question. These days I rarely get above 48 tracks or so (if even THAT many). Spent far too much time mixing the close to 100 tracks projects that are absolutely NO FUN whatsoever. I honestly don't think there's any advantage to that many tracks. I would suggest bouncing/combining tracks that belong together, something I'm used to doing coming from 4-24 track recording when I got my start! Used to run around the studio yelling "commit, commit, commit" when folks refused to print FX or combine tracks. Life's too short to spend micro-managing your mix IMO - I spend more time on the big picture (is this instrument even needed, is the song too long, etc.) and not worrying about one of my 23 vocal tracks being 1 dB too loud!!!
;)


Sent from some crappy device using Tapatalk
Selig Audio, LLC

per-anders
Posts: 224
Joined: 09 Jul 2015

01 Aug 2017

How fast is your HD?

Audio tracks are limited by how fast your HD can stream them. On your computer the HD is likely to be the slowest thing, so it's the limiting factor, 63 audio tracks is quite a lot of data, assuming a 24/32bit float per sample which takes up 4 bytes (24bit only makes sense byte aligned processors, you really want to align to the 64bit block on a 64bit processor for maximum performance) then at 44.1k that's 4x44100x63 = 11,113,200 bytes/second or just shy of 11Mb/s transfer rate.

Now that might sound like a pretty low figure and your HD may claim an STR of 200Mb/s or even more misleading it may claim transfer rates related to the interface of multiple Gb/s. Unfortunately even though that's listed as sustained it's a theoretical peak if you had only a single file that was totally contiguous. Without Reason optimizing the data into a single block for read/write what will happen is seeking, sections of a file may be all over the shop, reading/writing multiple files may occur on single threads in Reason but your HD only has so many arms, so they must physically move around, or if it's SSD then it must physically search for data in different locations. This fragmenting is what slows down your computer at every level (not just on the HD). Then if you're not using a separate scratch HD then Reason is competing with the OS and other applications for use of the HD slowing things down even further, and finally even if you are using another drive the bus that it's on may have any number of other devices also competing for it's bandwidth.

So 11Mb/s may actually be quite a lot of sustained traffic for your HD in the real world.

What are your solutions?

First step, if it's not an SSD then defragment it. This will at least help reduce fragmentation of the files which will speed up seek times.

Next try to reduce competition for bandwidth. Use a different drive for scratch work from your OS and programs drive. Check to see if you've extraneous stuff on the same bus, e.g. if you're using USB then remove USB devices you're not using etc.

Finally see if you can get a faster HD for the job, one with a bigger bus will help not just faster RPM or an SSD can do wonders.

Now having said all of this, it needn't just be the hard drive slowing you down. You may simply now have the correct setup for your soundcard, you could have the wrong driver, there may be other factors involved. Your comptuer may just not be suitable for the job.

In the future I would hope that Reason will offer smart track freezing because it's a major issue. By smart I mean including groups, so rather than streaming every single track it will just stream the parent group/bus and only deal with child channels/tracks if the user is editing them. This way a complex project could be handled and reduced to be extremely efficient and fast, and it would be like working on a project with the bare minimum number of audio only tracks.

jlgrimes
Posts: 661
Joined: 06 Jun 2017

01 Aug 2017

I think this is one of the benefits of using SSD drives and lots of Ram.

How much Ram do you have?

A lot of RAM (more than 8 gigs) might negate the need for a SSD or 7200 rpm drive.

If you are on a 5400 rpm drive (most laptop HHD drives), then you will typically have poor audio track playback performance, so in a nutshell it could be low RAM or a hard drive that is either:

1. Slow
2. Too fragmented
3. Almost full

per-anders
Posts: 224
Joined: 09 Jul 2015

02 Aug 2017

By the way don't forget to check out this great little guide to optimizing if you're on windows, some of this might seem dumb but it can have a surprising impact, the processor scheduling for me seems to be the inverse of what i'd assume to be correct, but it made a huge difference for me. I'm not even using a focusrite convertor in case you were wondering. https://support.focusrite.com/hc/en-gb/ ... Windows-10

Palmeira
Posts: 111
Joined: 18 Jan 2015

02 Aug 2017

Hi guys,

I've got a 2 TB 7200 rpm WD Blue HDD, 4 GHz I7 4790 CPU and 16 GB RAM. The song file is only 2 GB so I would have thought from a performance point of view it would make sense for Reason to copy all the audio (in fact the whole song file) into RAM rather than stream it from disk. SSD would only help here by making the initial opening of the song file faster although using the SSD as a page file may help - I don't know how Reason uses temporary disk storage particularly in light of the fact a great many people now have more than enough RAM to make virtual memory almost redundant. Having said all that I may just try an SSD just out of interest.

Thanks for all your suggestions I'm going to bounce down to stereo and import that into a new song file so I can complete the guitar overdubs at low latency and then these will be imported into the existing song which will have to be mixed while running the buffer at 4096 which shouldn't be a problem.

I'm still none the wiser though as to why playing the same number of tracks from audio should result in the same performance hit as producing the audio from devices. If this was to be expected none of the other DAWs would have a freeze function. I'm just hoping Reason's Freeze function when it comes is of the smart variety otherwise I'll likely have something else to moan about!

Cheers all

Aiden

User avatar
EnochLight
Moderator
Posts: 8405
Joined: 17 Jan 2015
Location: Imladris

02 Aug 2017

Palmeira wrote:
02 Aug 2017
Hi guys,

I've got a 2 TB 7200 rpm WD Blue HDD, 4 GHz I7 4790 CPU and 16 GB RAM. The song file is only 2 GB so I would have thought from a performance point of view it would make sense for Reason to copy all the audio (in fact the whole song file) into RAM rather than stream it from disk. SSD would only help here by making the initial opening of the song file faster although using the SSD as a page file may help - I don't know how Reason uses temporary disk storage particularly in light of the fact a great many people now have more than enough RAM to make virtual memory almost redundant. Having said all that I may just try an SSD just out of interest.
Well, for starters - the only thing Reason places into RAM are samples (think NNXT, Kong, Redrum, etc). All audio is streamed from your hard disc. So despite your HDD's size (2 TB), its read/write speed could very well affect your playback. Do you know what your HDD's read/write speeds test at?
Palmeira wrote:
02 Aug 2017
I'm still none the wiser though as to why playing the same number of tracks from audio should result in the same performance hit as producing the audio from devices.
Did you bounce your audio down into the same project? If so, even having instances of instruments and effects in your rack - sitting unused - can soak up precious CPU cycles. I would follow Selig's instructions and always bounce your stems down, close that session, and create a new one - importing those audio tracks only into a new session to continue working on (or mixing down).

That said, I have a slower CPU than you yet my DSP isn't too bad on large audio projects. I've done 100+ audio tracks without issue, but I also have an SSD (which is where my scratch disc is located).
Win 10 | Ableton Live 11 Suite |  Reason 12 | i7 3770k @ 3.5 Ghz | 16 GB RAM | RME Babyface Pro | Akai MPC Live 2 & Akai Force | Roland System 8, MX1, TB3 | Dreadbox Typhon | Korg Minilogue XD

Palmeira
Posts: 111
Joined: 18 Jan 2015

03 Aug 2017

Update - I copied the song file to a USB stick, took it to work and opened it on my iMac. I've taken the built in input and output down to a buffer size of 64 samples and added a Thor to check what the latency is like and it's non-existent. I'm now listening to the song played back from the USB stick at the lowest buffer setting and running absolutely flawlessly - no glitches at all. I'm even typing this into a browser while it plays in the background. I would have to conclude my other computer is shit so I'll be probably reinstalling Windows and all software and drivers to see if it can do any better. May try the SSD too and will in future take Selig's advice on the importing of any exported stems into a new song. Bummer!

Cheers all

Palmeira
Posts: 111
Joined: 18 Jan 2015

03 Aug 2017

:SOLVED:

I've noticed the computer fan getting louder for a while now so just on the off chance I opened it up and found the heat-sink between the CPU and the CPU fan was coated in a 3mm thick layer of dust so I got a knife between the fan blades and scraped the dust off and hoovered it out of there. I can now play the entire song at the lowest buffer setting with absolutely no glitches. The computer must have been throttling the CPU back due to the temperature. Who'd a thunk it? I'm now one very happy bunny.

Cheers for all your input guys - much appreciated

Aiden

User avatar
mcatalao
Competition Winner
Posts: 1826
Joined: 17 Jan 2015

05 Aug 2017

Palmeira wrote:
31 Jul 2017
Hi all

I've got an overly complicated song that I thought would be much easier to mix if I bounced almost everything to audio, which I did track by track. I've now got 63 tracks of audio and only 2 rack synth tracks left that are playing a not too complicated bassline and some synth chords for a vocoder on one track of vocals. I have removed entirely all devices and tracks that have been bounced and altogether this seems to me to be the equivalent of utilising the freeze function employed in other DAW's.

Apart from simplifying mixing I thought (quite reasonably - no pun intended) that this would also free up a great deal of audio performance and enable me to whack the buffer size down and record some additional guitar.

But no.

Performance seems to be identical and I still can't run the song unless I run at the highest buffer rate setting (4096 samples on a Balance interface).

Can anyone advise why I'm seeing this behaviour from Reason? It doesn't seem to make much sense but maybe I'm missing something obvious!

Cheers

Aiden
Maybe this is a stupid question, but did you delete the old devices? Re's, specially synths will still use A LOT of CPU even without any sound passing through them.

Palmeira
Posts: 111
Joined: 18 Jan 2015

05 Aug 2017

mcatalao wrote:
05 Aug 2017
Palmeira wrote:
31 Jul 2017
Hi all

I've got an overly complicated song that I thought would be much easier to mix if I bounced almost everything to audio, which I did track by track. I've now got 63 tracks of audio and only 2 rack synth tracks left that are playing a not too complicated bassline and some synth chords for a vocoder on one track of vocals. I have removed entirely all devices and tracks that have been bounced and altogether this seems to me to be the equivalent of utilising the freeze function employed in other DAW's.

Apart from simplifying mixing I thought (quite reasonably - no pun intended) that this would also free up a great deal of audio performance and enable me to whack the buffer size down and record some additional guitar.

But no.

Performance seems to be identical and I still can't run the song unless I run at the highest buffer rate setting (4096 samples on a Balance interface).

Can anyone advise why I'm seeing this behaviour from Reason? It doesn't seem to make much sense but maybe I'm missing something obvious!

Cheers

Aiden
Maybe this is a stupid question, but did you delete the old devices? Re's, specially synths will still use A LOT of CPU even without any sound passing through them.
HI mate, yep deleted em, it was an issue with dust accumulating on the CPU fan and the computer throttling down the CPU because it was getting too hot. All sorted - cheers.

Post Reply
  • Information
  • Who is online

    Users browsing this forum: No registered users and 27 guests