Building an Audio PC

I'm hoping to build a new PC to track 16-24 channels at once, what should I be looking for in a processor and mobo? High Clock Speed or would I benefit from prioritizing core count?

Attached: Studio_Computer.png (882x664, 986K)

Other urls found in this thread:

youtube.com/watch?v=_MK__lPE5z8
techreport.com/review/32642/intel-core-i7-8700k-cpu-reviewed/14
youtube.com/watch?v=CA7InQKsBT8
twitter.com/SFWRedditVideos

I have been researching the absolute fuck out of this lately myself (audio is all I do), gotta be real clear about what you're doing with it. Do you use VSTis? Or are you just tracking real instruments? Are you mixing or JUST tracking? Do you usually prefer to make powerful single channel chains or use more send effects? What's your interface, does it have low latency input mode? Are you wanting to use that or do you want to handle low latency in the DAW?

The way most DAWs are coded dictates that how you set up your session and what you're doing with it is going to determine what you need.

How it works (basically) is each track cannot be processed by more than one core. This is very important, how you set things up will affect the processing load. And obviously the lower the buffer, the lower the latency but the more demanding the processing.

One channel with a very complex chain where delays and reverbs are being used as inserts after a VST synth at a very low buffer (say 64) will require a very powerful single core rating. But if you were using those FX as sends on seperate channels each then now you're using 3 different threads and distributing the load.

If you want ultra low latency in the DAW more, you will need CLOCK SPEED. But thread count will allow for more seperate lanes of processing. As you can imagine, in mixed use operation a balance of the two is going to be better than being overkill in either direction.

Some other things to consider is that there is discussion that the current larger core AMDs like threadripper do not perform well under ultra low latency because of how the infinity fabric functions but I am not experienced there so I cant personally weigh in. I have heard that the AMDs work fine for mastering guys who only work at a high buffer. Oh and there's the other factor of whether or not you want Thunderbolt (either for an interface or addon ecosystem like UAD >inb4 shitposting).

Of course if you want thunderbolt, Intel makes a lot more sense, I *think* AMD has thunderbolt now?? But who knows if it works and what hell you go through to get it to work lol. Shame because its royalty free this year I believe but I doubt AMD adoption will increase.

Any anons with an AMD audio rig wanna sound off on how it runs/purchasing tips?

Oh and before I forget, from what I understand Quad Channel RAM is king for VSTis and Polyphony, to the point that 16 gigs of quad might be better than 32 gigs of dual

Oh and adding to there's only one mobo with built in tb3 where you dont need a header card and it's a mini itx fatal1ty board by ASrock with only 2 ram slots and 1 m.2 (it is a good board I hear though, sucks they dont make a bigger one with the same stuff).

Not sure what else about mobos applies to audio in terms of chipsets, but I would assume that buying a generally higher end mobo is going to be a good call for audio since the mobo and cpu are the only components that really matter

Get something that runs Haiku

lol just buy a macbook pro

kek honestly this, the sum of my research was that unless you're an orchestral composer and want monster templates, mac is still the way to go, it just werks.
>inb4 massive trolling

I track with a kabylake dual core i3 nuc, but all i do is track upto 18 simultanious audio channels and sequence. I do no effects processing in the box.

You dont really need a powerful computer to track audio or to sequence a hundred tracks, you could honestly do it on a 10 year old duo, its only if you are using plug ins or effects on dozens of channels at once that you need a good processor and lots of ram.

Im using a 18i20 focusrite usb interface with a focusrite octopre dynamic adat expansion plugged in via fibre optic

The 18i20 allows me 8 inputs i have running through alesis rackmount compressors and rack effects , + 1 stereo spdif coming from one of my rackmount guitar effects units.

The adat expansion has 8 compressors before the 8preamps which are sent over the fibre optic into the optical input of the 18i20. My outputs are cloned and go to monitors, samplers, chromatic tuners etc.

youtube.com/watch?v=_MK__lPE5z8

This. Get a 2015" 15" Macbook Pro for like what, $1,000?

Learn how to play an instrument and stop making shitty techno faggot

techreport.com/review/32642/intel-core-i7-8700k-cpu-reviewed/14

I'd wait and see if TR will do their DAW benchmarks on the 9900K

Using Reaper currently, and I try not to get too complex but typical projects look like 50-60 tracks, many of which with an eq and a comp or two from a UAD Quad. Usually use sends for delays and verbs, and a tape sim across a few busses

Seems like I want an i7 8700

what techno artist is tracking 16-24 tracks at once? I play 4-5 instruments as it is...

I know people who track 8 at once doing drum and bass just with an akai sampler.

Depends on use case really, some people essentially use a mixer to sum the inputs down to stereo ins, some record everything on a seperate track so they can process later.

I tend to like having as much as possible in its own track as its great for seperate eq and compression of just say a snare alone or a bass drum, its a nightmare trying to mix problems with recordings of drums when there is bleed or other stuff going on the channel. You can use gates i suppose. but its just a pain in the ass.

But imo, and im not a pro, just a long term hobbyiest, shit in shit out is a real thing, if you can get good recording/sound in it makes post so much easier.

There is a reason gear like sure sm57s and 58s, alesis 3630s, sennheiser hd280 pros, fender strats etc etc are industry standard, because its cheap realiable gear that works well.

The old macs are better for music and graphic design was a thing over a decade ago, but not so much these days.

I've used everything from atari st cubase, to macs, to pc.

>what techno artist is tracking 16-24 tracks at once

Most are not i give you that, certainly not people working in the box.

I can output 16 channels at once from one of my akai samplers for example, if i have it setup with say channel 1-8 as seperate drum sounds, 9-10 as basslines, 11-12 as lead and 13-16 as strings/pads you'd be suprised.

I honestly probably need a large format mixer again, probably an 8 bus with 32 channels if not 48 again, and some patch panels but i swore years ago im done with analogue mixers, and they are expensive because they are somewhat rarer these days.

Its oft easier to just sum it to stereo and track the seperate parts sequenced from samplers onto seperate channels via overdubs/rerecords, saves me having to plug and unplug shit, its just a pain to have to rerecord seperate parts on seperate channels 10 times over.

So basically thats a really long way of saying "none"

cool!

>i7 8700
Yeah if you're on UAD already stick with intel because I know their hardware is built with that in mind. Imo, the 6-core is the perfect sweet spot for audio, its top of the line when needing low latency, but the extra 2 cores means when you back the buffer off to mix, you'll have enough power under the hood to spare. More cores than that is better only if you are strictly mixing or have monster orchestral templates

>But imo, and im not a pro, just a long term hobbyiest, shit in shit out is a real thing, if you can get good recording/sound in it makes post so much easier.
Oh and you're right on the money with this. I've apprenticed in real top studios in LA, I've sat in on seminars with two of the biggest mixers in the industry in their studios and "shit in/shit out" is 1000% the facts. You cant polish a turd but so much. Whatever kind of music it doesnt matter, good music is like good food, it starts with good ingredients.

Just get an teenage engineering op-1, and op-z.

I don't want to hear you play 4-6 instruments shitty. Get good at one and do something impressive. And if you play the memetar your a retard.

>memetar
お前はもう死んでいる
youtube.com/watch?v=CA7InQKsBT8

>covering another man's song

Let's be honest the only instrument we're gonna see your ass playing is the skinflute.

don't listen to shills
any quad core or better cpu, 8+ gigs of ram, and ssd will the be good enough for general use
combine with a decent audio interface and speakers and you're set

this
t. actual musician that makes money from making music

Agree, especially on the ssd you will notice a huge difference

I actually don't have an ssd. I don't think it's a big deal.

*shrug* I have huge 100s gigs of piano libraries to load, life before ssds is a distant memory

you fit 100 gigs of sample libraries into the ssd? aren't ssds by nature small?

16 to 24 audio channels could be very tricky. You might need a 10 year old computer verses 20 years.

If its just for tracking audio (and not using lots of VSTis) you shouldn't need a very powerful processor in the first place. Recording and Playback of samples are not very CPU intensive, but working with lots of VSTi's and effects can be.

by nature that humans are poor, yes. however there are ssds bigger than the biggest spinning rust, they're just expensive.

Get a Mac. Asio is garbage and just adds even more latency to how windows already handles ram like garbage. Also no good daws

you can literally do all of what you say on a 10year old computer or even older.

you don't need a powerful computer to edit audio just a bunch of ram.

Even if that was compiled on the most fucktarded version of the ICC, there's no way any program ever will see more than a 2x performance difference between an OC'd 1700 and a stock 8700k. The 8700k's boost is 4.7GHz, a 15% difference in singlecore clockspeed and a 10% MAXIMUM IPC difference doesn't translate to 100% performance differential. What the fuck is wrong with that software? Even if it were using AVX512 I wouldn't expect to see this kind of performance delta.

This. I/O is going to be your primary bottleneck 9/10 times when dealing with audio. Max out your RAM slots, boot off an Optane drive, and overclock the tits out of your RAM. If you can run 128 gigabytes @ 3600MHz, do it.

>track 16-24 channels at once
I can do 128 @ 48/24 using an usb madiface xt on my 2012 samsung ultrabook with no problems at all. audio itself is no workload at all