How much RAM is used for Garageband/VSTs (home recording)?

jcdamascenojr said:

Hi, first post here and yes, it is another "how much memory should I buy?" thread.

Well, I'm from Brazil and just ordered my first Mac (the i7 2.3 one, stock RAM/HDD), and I'm reallly excited about it. My girlfriend's sister is going to bring me it from US next week since it's a lot (think twice) more expensive to buy one around here. My question is: should I ask her to buy more RAM too?

It will be used for browsing, office, light photo editing, common stuff (for what I'm sure 4GB is plenty).
However I'd like to start recording some songs with a friend, nothing professional, just hobbyist. I just ordered an Apogee Duet 2, and plan to start messing with GarageBand. Since I have no prior experience with recording software requeriments, could some of the musicians out there suggest me how much RAM will I be using?

- It's just home recording, I have no plans on building a studio, recording other bands, etc.
- Plan to use GarageBand, may upgrade to Logic/ProTools if I find it too limiting.
- I plan to buy NI's Komplete 8, mainly for Guitar Rig, Kontakt and StudioDrummer (we play alt/indie rock). May as well buy some piano library (Ivory or Gallaxy) since I'm a pianist.
- Since my room has no treatment, we plan to record vocals and acoustic guitars on an real studio (maybe take the mini/duet and just use the acoustics there - is it possible? has anyone done it?)
- I've never recorded anything before, so I'm not not sure how many tracks, but I guess it will be around... 20? (drums, vocals, maybe 3 guitars, bass, piano or synth). Obviously I plan to run Guitar Rig for the guitars/bass, StudioDrummer/EZDrummer for drums, Kontakt for piano, along with EQ, reverb, you know, this things tracks usually have but I still don't know about


So, that's it. How much RAM should I buy/expect to use on a session?


And please, please just don't say "more is always better".
I don't want to spend $100 if I don't have to. I don't wanna spend $1 if I don't have to.
I don't wanna have 16GB if I'll be using only 4, just to watch those 12GB doing nothing.
I don't want to futureproof anything, I plan replacing the mini as AppleCare expires.
I won't be using RAM disk or VMs.
I don't wanna buy SSD for now.

Thnaks!

Click to expand...

you would be fine with 4 gb as a beginner, but 8 gb would be better. your mini would feel itself more comfortable.
16 gb is "too much" for you.
imho.

also... if your main focus is on recording, nothing real professional, the base i5 mini would be enough too (it is still waaaaaaaay more powerful than for example a top 2009-10 15" macbook pro model, which could run very big logic pro projects without ANY problem at all.
but... if you do HUGE projects with tons of software instruments (sample based or synthesizers), than the quad core processor has advantages (and more rams too - 16 gb).
so... "you should think first, befor you move" save money, man! i5 base mini + 8gb ram + external scratch disc (this is my configuration).

jcdamascenojr said:

Thanks for the reply!

So, it seams like 8GB is the way to go.

Are you really sure the dual-core is enough? What worries me is that on my current C2D notebook a single Guitar Rig instance easily uses ~50% of CPU... so I thought a quad would be best if I run, say, 4 or 5 at the same time.

BTW, do you know if multiple GR instances (on different tracks) use multiple cores? Or will they use the same one?

Finally, a scratch disk is an external drive where the recording will be made, right? Is there any reasons not to record at the internal drive (other than taking up space)?

Regarding space, how much would a project eat? Something like a 5min song with 20 tracks, 24/44.1? Is there a math for that? Do the amount of plugins matter?

Click to expand...

i suggest you to wait for a few more comments before you "move". that was only my opinion.
i think the i5 is very capable. just do a geekbench test to see, how strong your current notebook is. my base mini scored around 7000.
as i said before, you don't need that extra quad power, unless you have big projects. i mean, really big projects. software synthesizers can eat up your CPU, but this doesn't mean, that you can't work with a dual core cpu (actually the base mini support hyper threading ---> 4 cores). you have to use A LOT of VSTs at the same time to "kill" the base mini.
some people think that the base mini can't do anything and it can run only tetris with 5 fps, but that's just bulls##t.
actually, it's your decision. but it seems that you're a pretty conscious customer. and i find this great. i think most of the people has way too strong computers, even though they never really use that extra power.
it would be great, if you could test different minis before you decide to keep one, but as it seems, that's not possible. so... again: wait for a few more comments. I THINK the base mini would be great for your needs.
multicore usage: i'm not sure. it depends always on that certain program. for example the Native Instruments Maschine software can use only 1 core, which is pretty bad. i don't know how other NI products (Komplete?) perform. maybe garageband or logic can control these things...
it is always better to record on external HDDs, since your internal HDD has "always" something to do. this can influence the performance.
Space... that can vary.

I'd stick with the i7 quad-core. It seems like it is a big hassle for you to get the computer, and probably can't return it if it isn't cutting it for you. On top of a much faster processor, you'll get twice the hard drive space(500GB vs 1TB) to save your projects. I would purchase a minimum of 8GB of RAM. While I don't think you will fully utilize 16GB right now, you might in the future, and its not very expensive to pick up 16GB of RAM right now($60-$80).

memory's cheap. be glad you probably weren't old enough to be so g*ddamn happy when 32mb's dropped from $1,000 to $500. whoooooowheeeeee

why superior drummer alone will drop nearly a gb of samples into memory

jcdamascenojr said:

...a scratch disk is an external drive where the recording will be made, right? Is there any reasons not to record at the internal drive (other than taking up space)?

Regarding space, how much would a project eat? Something like a 5min song with 20 tracks, 24/44.1? Is there a math for that?

...Do the amount of plugins matter?

Click to expand...

It seems "scratch disk" is a sort of nebulous term; I would not put much stock in using that term for music recording. "Scratch" usually refers to a temp storage area. In music, you typically master to a permanent storage area, and keep the master so that you can always do mixdown tracks from that (which take less storage) without ever painting yourself into a corner (once mixed-down, you can't unring that bell and "un"-mix down). That implies that you will want a permanent copy of what you master preserving all of the tracks, assuming it is a keeper take.

But yes, the common practice is to use an external drive or other second drive. The reason has always been that you do not want the bus carrying non-media commands back and forth (nor the drive throughput) to be squeezed by the r/w of media, or vice versa. If you use two drives, try to place them on different busses, but that may not really be an issue with Thunderbolt.

Another reason to keep media separate is that if you are prolific you may want to have a lot of different media drives over time for different projects (since there will be an impetus to archive everything), and this allows you to manage the logistics of that without touching the boot drive.

I would recommend a WD Green Caviar drive, as they are designed for multiple tracks of continuous media, such as HD DVRs (they can carry 12 channels of HD video at once). The EURX model is probably the drive of choice; it is specifically designed for multimedia and has a 64 cache (the more tracks you attempt to r/w simultaneously, the harder the drive has to work jumping back and forth from track to track, and a large cache helps a lot). Check them out on Amazon, along with a ThermalTake eSATA enclosure, which allows swapping of bare drives fairly easily.

Video is most certainly done this way (separate drives) for the above reasons. Audio used to be done this way, but possibly that is now a legacy holdover and is not as important now that the drives and busses are faster. The good news is that you can try using the boot drive when starting out, and if that seems to stutter the playback or freeze your GUI, simply move to a x drive.

There is indeed math for that, and it is as simple as using the parameters you have stated. For instance, 24-bit digital words times 44,100 samples per second times 300 seconds (5 minutes) times X number of tracks will yield pretty much the exact number of bytes you expect it to. I recommend 48k/96/192k if you are going to try to do this professionally, and 44.1 or 88.2 if you are really just going to only eventually go to CD anyway. CD truncates 24-bit to 16, but if there are multiple layers of processing (and even just a single mixdown implies a generation and subsequent cumulative quantization error) using 24-bit or even higher for mastering minimizes any generational artifacts.

Now that resultant number refers to storage, or file size placed on the HDD. Plug-ins do not count there; they count against RAM. In a simplified example, if your OS and app take 2 GB of RAM and your plug-ins take 3 GB of RAM, that implies you must have 5 GB of RAM available to prevent paging, and for streamed media paging is the kiss of death, because no pauses in the operation could be allowed.

jcdamascenojr said:

...Since my room has no treatment, we plan to record vocals and acoustic guitars on an real studio...

...I've never recorded anything before, so I'm not not sure how many tracks, but I guess it will be around... 20? (drums, vocals, maybe 3 guitars, bass, piano or synth)...

Click to expand...

You might find that a tiled bathroom is ideal for acoustic guitars or vocals.

There are two ways to track. For instance, you can record an acoustic guitar or piano with one mic and then simply pan that mono source around in the musical space of a stereo mixdown so that it appears in the sound field localized where you want it. But a better way is to use two mics and create a full-center stereo track of the guitar (about 3 dB or more of panned separation between the mics), and then at mixdown pan the stereo track pair to localize that in the sound field. This gives instruments a better sound, to my way of thinking; more lush, more enveloping, more natural. But it takes 2 tracks (one stereo pair) per instrument.

One slight problem with that technique is that panning at mixdown will change not only where the instrument localizes in the stereo sound field, but the character of how the instrument actually sounds to some degree, since changing the relative levels of the two tracks as you pan essentially changes the relative levels of the two separate mics. The more they are tonally balanced or alike, the less of a problem this becomes, so generally speaking, when stereo tracking an acoustic instrument, use two identical mics.

Stereo tracking like this works really well for keyboards; I like to use stereo keyboard sources with higher keys panned progessively more to the right and lower keys panned progressively more to the left (center keys mostly in the middle); IOW, you can create a fully-panned stereo soundfield for a keyboard (every key has its own unique position in the sound field) if you use a keyboard sound element that does this and just lay that under everything else as the foundation for where you pan the other instruments. Where the pan location for the keyboard generally starts out will depend on what parts of the keyboard are used, and you can nudge the entire instrument to the left or to the right somewhat during mixdown, and then fill in the gaps in the sound field with the other instruments panned as you see fit. Again, just 2 tracks (a stereo pair) is all that is needed.

Drums, I like to have a single mic and a single track for every single drum, cymbal, and percussive instrument. Then, you can take the mono tracks of each of those and pan them into a stereo sound field at mixdown. If you use too many mics or try to track each drum with multiple or stereo tracking it would be overkill, and there is enough bleed between mics anyway, unless you are doing drums and percussion electronically, and in that case you may want to use the stereo tracking technique for those as well. But if you stereo track every element in an acoustic drum setup it will start to sound muddy, and that also makes it hard to leave "holes" in the stereo soundfield for the other instruments.

Vocals take a single mic per vocalist, IOW each vocalist should be tracked in mono; you don't want a lot of ambience in a vocal performance anyway, and what you do want can be done in processing. If you autotune a vocal mic and there are other, non-autotuned mics picking up ambience, that will start to sound like something from outer space, and not in a good way. You have to iso a vocal.

So, tracks start to pile up; you could use 12 to 24 tracks just for drums and percussion alone. I sometimes exceed 40 tracks for 5-6 instruments, drums and percussion, and accent instruments or percussion. But I track each instrument separately starting with a click track, so I don't need a lot of mics.

Generally speaking, for a precise track count, count every acoustic instrument or stereo electronic sound and multiply that by two, and count every mono electronic sound, acoustic drum, vocal, or percussion element and add that total to the first number. If your drum sounds are electronic, you may want two tracks per percussion element. If you avoid submixes of the drums or other instruments and dedicate one or two tracks to every element, that gives you maximum flexibility when you finally mix down. Anything else is a compromise you will likely wish you had not made.

A simpler approach is to count two tracks (a stereo pair) for every single element, acoustic or electronic, and then assume that you will use a few less than that because a few of the elements can be or will be mono elements. Be sure to track every element either as a mono track, or stereo but fully panned to center (other than the slight l and r pan you give that track pair to create that stereo source track). Tracking is a completely different process than mixing. In tracking, the goal is to get the highest volume available just below distortion, panned full center for stereo tracks.

When you do the eventual master mixdown you can pan each element or stereo pair where you want them localized in the sound field then. Track every element at full volume (and later when you mix, lowering a track to where you want it maximizes the final signal to noise ratio), and don't include silent tracks in the mixdown just because they are there and unused, because each track adds cumulative digital noise to the noise floor; eliminate those tracks unused instead.

TyroneShoes2 said:

It seems "scratch disk" is a sort of nebulous term; I would not put much stock in using that term for music recording. "Scratch" usually refers to a temp storage area. In music, you typically master to a permanent storage area, and keep the master so that you can always do mixdown tracks from that (which take less storage) without ever painting yourself into a corner (once mixed-down, you can't unring that bell and "un"-mix down). That implies that you will want a permanent copy of what you master preserving all of the tracks, assuming it is a keeper take.

But yes, the common practice is to use an external drive or other second drive. The reason has always been that you do not want the bus carrying non-media commands back and forth (nor the drive throughput) to be squeezed by the r/w of media, or vice versa. If you use two drives, try to place them on different busses, but that may not really be an issue with Thunderbolt.

Another reason to keep media separate is that if you are prolific you may want to have a lot of different media drives over time for different projects (since there will be an impetus to archive everything), and this allows you to manage the logistics of that without touching the boot drive.

I would recommend a WD Green Caviar drive, as they are designed for multiple tracks of continuous media, such as HD DVRs (they can carry 12 channels of HD video at once). The EURX model is probably the drive of choice; it is specifically designed for multimedia and has a 64 cache (the more tracks you attempt to r/w simultaneously, the harder the drive has to work jumping back and forth from track to track, and a large cache helps a lot). Check them out on Amazon, along with a ThermalTake eSATA enclosure, which allows swapping of bare drives fairly easily.

Video is most certainly done this way (separate drives) for the above reasons. Audio used to be done this way, but possibly that is now a legacy holdover and is not as important now that the drives and busses are faster. The good news is that you can try using the boot drive when starting out, and if that seems to stutter the playback or freeze your GUI, simply move to a x drive.

There is indeed math for that, and it is as simple as using the parameters you have stated. For instance, 24-bit digital words times 44,100 samples per second times 300 seconds (5 minutes) times X number of tracks will yield pretty much the exact number of bytes you expect it to. I recommend 48k/96/192k if you are going to try to do this professionally, and 44.1 or 88.2 if you are really just going to only eventually go to CD anyway. CD truncates 24-bit to 16, but if there are multiple layers of processing (and even just a single mixdown implies a generation and subsequent cumulative quantization error) using 24-bit or even higher for mastering minimizes any generational artifacts.

Now that resultant number refers to storage, or file size placed on the HDD. Plug-ins do not count there; they count against RAM. In a simplified example, if your OS and app take 2 GB of RAM and your plug-ins take 3 GB of RAM, that implies you must have 5 GB of RAM available to prevent paging, and for streamed media paging is the kiss of death, because no pauses in the operation could be allowed.

Click to expand...

you're right. the term "scratch disc" is not related to music production. it's related more to video production. i wrote my comment very fast and my native language is not english.:S
but he understood what i meant.)

So, after all of this time, how are things working out?

TyroneShoes2 said:

I would recommend a WD Green Caviar drive

Click to expand...

I respectfully disagree. These drives are designed to go into DVRs, not computers. They are also designed to save electricity and that feature has a significantly undesirable effect in PCs running modern operating systems.

I have two WD Greens. After a few months these started to have somewhat rare, but consistent problems. After a couple of years one failed completely and the other is so filled with errors that I refuse to use it.

I realize this is anecdotal evidence but in my research after these drives failed, I found many other people with the same problem and also came across articles about design flaws and how they shouldn't be used in computers.

To be clear, I was actually using these in a computer whose sole purpose was to be a DVR. This was a Windows Media Center computer dedicated to DVR purposes. The operating system was on a normal drive and the two WD Greens were recording drives, so the Greens were being used for their marketed purpose only. But it doesn't matter because modern operating systems do all manner of things with drives, even storage drives, that a DVR's firmware doesn't. This conflicts with the insanely aggressive power saving feature of the Greens, which results in a constant mechanical parking and unparking of heads. This combination leads to the drives exceeding their design life much, much faster than expected from use in an a simple DVR.

In order to help combat this problem, WD released a utility that changes the parking behavior. If you are currently using Greens in a computer, I suggest that you check their S.M.A.R.T. data for the Load_Cycle_Count (this is basically a counter for parking/unparking cycles). If your particular usage pattern isn't conflicting with the parking behavior, you'll see a low count and your use case is fine for these drives. If your usage pattern does conflict, you'll see an insanely high count and you should use WD's utility to change the parking behavior.

There is more information here if you care to read about it:

http://www.ngohq.com/news/19805-critical-design-flaw-found-in-wd-caviar-green-hdds.html

http://www.silentpcreview.com/forums/viewtopic.php?f=7&t=51401

TL;DR:
WD Greens are designed for DVRs and shouldn't be used in computers. If you are using it in a computer, there's a way to check if it's getting hosed or not. If it's getting hosed, or indeed just as a preventive measure, you can use a WD utility to change the Green's behavior so that it works okay with computers.

ActionableMango said:

I respectfully disagree. These drives are designed to go into DVRs, not computers. They are also designed to save electricity and that feature has a significantly undesirable effect in PCs running modern operating systems.

I have two WD Greens. After a few months these started to have somewhat rare, but consistent problems. After a couple of years one failed completely and the other is so filled with errors that I refuse to use it.

I realize this is anecdotal evidence but in my research after these drives failed I found many other people with the same problem, and also articles about design flaws and how they shouldn't be used in computers.

To be clear, I was actually using these in a computer whose sole purpose was to be a DVR. This was a Windows Media Center computer dedicated to DVR purposes. The operating system was on a normal drive and the two WD Greens were recording drives, so the Greens were being used for their marketed purpose only. But it doesn't matter because modern operating systems do all manner of things with drives, even storage drives, that a DVR's firmware doesn't. This conflicts with the insanely aggressive power saving feature of the Greens, which results in a constant mechanical parking and unparking of heads. This combination leads to the drives exceeding their design life much, much faster than expected from use in an a simple DVR.

In order to help combat this problem, WD released a utility that changes the parking behavior. If you are currently using Greens in a computer, I suggest that you check their S.M.A.R.T. data for the Load_Cycle_Count (this is basically a counter for parking/unparking cycles). If your particular usage pattern isn't conflicting with the parking behavior, you'll see a low count and your use case is fine for these drives. If your usage pattern does conflict, you'll see an insanely high count and you should use WD's utility to change the parking behavior. In the latter case you should WD's utility to change the parking behavior.

There is more information here if you care to read about it:

http://www.ngohq.com/news/19805-critical-design-flaw-found-in-wd-caviar-green-hdds.html

http://www.silentpcreview.com/forums/viewtopic.php?f=7&t=51401

TL;DR:
WD Greens are designed for DVRs and shouldn't be used in computers. If you are using it in a computer, there's a way to check if it's getting hosed or not. If it's getting hosed, or indeed just as a preventive measure, you can use a WD utility to change the Green's behavior so that it works okay with computers.

Click to expand...

Yep. Avoid anything with the word Eco or Green in the name. It's just greenwashing failure-prone products. I fell for it once. Never again. Get an SSD. You can thank me later. PS: Avoid those Hybrid drives, too. SSD all the way, but skip the Samsung 840 EVOs. Check out http://techreport.com/review/27909/the-ssd-endurance-experiment-theyre-all-dead/4