Are you planning to work in something like Dorico or Sibelius, or primarily in a tracking DAW?
If you’ll be spending a good bit of time in something like Dorico, skip AM4 and go with the latest generation of AMD or Intel, or perhaps even a Mac. Get as much CPU and memory as you can afford if serious scoring is in your game plan. In this case it’s not as much about the instruments as enjoying a speedy interface. While you can most certainly run Dorico on AM4 (I do it every day), I think you’ll appreciate a snappier system with some headroom.
With something like Dorico, even on the AM4 build, today you’d have little problem running a full orchestra template (100+ virtual players) using the instruments that come with Dorico Pro, or something like Note Performer 5. Don’t expect an uber high quality, realistic, surround sound mock up that’s ready for the cinema though. You can get quality sufficient to sit there and compose all day, and share a sketch that’s more than capable of expressing your ‘composition and arranging’ chops. It can still be musical and pleasing to the ear, but at the same time you’ll know right away it’s a machine and not a real orchestra.
Where going with the newest generation of mother board and CPU is of value there, is processing all the graphics and stuff in a score…give yourself room to grow. If you’re doing a large score that’s also ‘lengthy’…say a full Opera…one simple change in the score might require the software do countless recalculations to rebuild the entire thing, so you’ll quickly appreciate a snappy PC that has room to grow, be expanded and upgraded, etc.
Currently, unless you’re planning to get into stuff like spectral editing, stem separation, or AI tools, there’s not much to gain investing in a fancy GPU, or APU, but this could change between now and the next time you’re ready to upgrade or buy something new. If you go with latest generation AMD, Intel, or Mac, you’ll be better equipped to add the ability to use AI software that can use a GPU and/or APU.
If you’ll mostly be working in a tracking DAW, any of the builds you’ve suggested should cover you for a few years. I say this because if you’re on a tight budget for the hardware, you’re not likely to be laying out thousands of dollars on high end instrument libraries anytime soon. You’ll have plenty to explore and learn with the base rompler libraries that come with most DAWs, free stuff you can find online, etc.
Unless you are ALSO trying to track a lot of real instruments/singers with mics while using your instrument plugins…you should be able to run a 100 track template for orchestra, as you can set higher buffers and such to get more plugins going. There are tricks to render instrument tracks in situations where you might require small buffers and tight latency for things like tracking from microphones.
How many plugins you can run? It varies depending on the plugin. Some require a lot of dsp, and some don’t require much at all. Most orchestral libraries use very little dsp if you run them dry (disable internal effects, share fewer external effects), since they’re mostly just triggering samples. With those kinds of plugins think of in terms of ‘voices needed’ rather than ‘plugin instances’. Having more memory might help if some of your instruments have lots of layers of samples and articulations, and samples from multiple mic positions doubling up. Something like a sampled piano from two simultaneous mic positions might require several voices per key-strike (dampers, releases, misc pedal noises, and more), and access gigabytes of samples from hard drive or memory in the process. In contrast, something like a flute might never require more than one voice at a time, and may not need to access but a small handful of samples. Where a convincing concert piano might require several gig of samples and 30 or more voices of polyphony, you can also get a very fine flute or clarinet that fits in a few megabyte of memory and only needs 1 voice of polyphony at any given moment. So…it tends to average out over a project.
Instruments that do more synthesis may require more dsp, but very little memory. I.E. a ‘modeled’ piano might only need a sample set of a few gig, but will use synthesis to add overtones, resonance, different noise effects, and more that require more cpu cycles. VS. A fully sampled piano that could be several gig of samples that just need to be triggered and played on demand. The former needs more CPU, the latter needs more storage to host the library, and either more memory, or more bandwidth across the bus to ‘stream’ samples in real time from the storage media.
Since it’ll be pretty rare that you will ever have all 100 tracks of a 100 track project making sound, all that same time, you’ll probably be fine with any of the builds you’ve mentioned so far. Particularly if you go with more modest rompler style composer libraries (affordable, small, efficient). Just because a project has 400 tracks, it doesn’t mean they are ALL demanding resources at the same time. Most DAWs and plugins these days are pretty smart about putting things to sleep that are idle or not being used at the moment and freeing up resources until they’re needed.
If your primary goal is to simply be able to arrange and get instant feed-back on your composing, you should be fine. If you’re wanting to make cinema quality mockups that sound convincingly like a real orchestra from beginning to end, it’ll probably be years before you’ve learned enough to make that happen, so take your time. You can always invest in high end gear down the road if you get that accomplished at doing mock ups.
If learning to mock up is your goal. Keep it simple at first. Learn to do small groups. I.E. A Violin solo with a piano accompanist. A string quartet. A brass quintet.
Ease into it. Your computer can handle that kind of template no problem. You’ll have much to master learning to mix and get expressive performances at the micro levels of working with fewer, more exposed instruments. You’ll find yourself evaluating instrument choices, weighing if you want to sink money into something fancy that might save you time, or brute force your way through things with instruments of ‘your own design’.
I.E. It’s possible to get a really fine mockup using free stuff, modding the things that come with your DAW, bringing in real players for some tracking sessions, etc. The cost there is your ‘time’ and ‘effort’ in learning how to cobble all the pieces together yourself. In contrast you might invest in an expensive library that ‘does more for you’ with less manual intervention, and saves a bunch of ‘time’, but unless you’re made of cash, or making a steady income from this work to finance it out, is substantially out of reach.
By the time you’re ready to do much larger arrangements and mixes, and spend money on fancy libraries, you’ll know better what you need, and if more/better hardware is required.
With a low end or midrange build, do run some tests, get demos and such before investing thousands in some ginormous high end cutting edge instrument library.
Getting top quality orchestral libraries can get expensive fast, so take your time. Learn to get tolerable results with instruments that come with your DAW, free stuff, and take your time evaluating if it’s really worth it to you to sink big money in high end instrument libraries.
Also, choosing a solid audio interface is just as important as sweating over things like CPU and Memory. You don’t need to spend a fortune to get something to learn with, and be well served for years; but be sure to get something. You won’t be happy trying to get by with whatever comes on most motherboards! The stuff on most mother boards will be noisy, fiddly to get ‘stable’ as you increase dsp loads, and probably won’t come with native ASIO drivers.
When it comes to interfaces, you can find really high quality, but older used stuff all over the internet that’ll work with Windows for a song and a dance. I.E. Stuff some pro studio is dumping so they can have ‘the latest and greatest’. You might find something a little older with way more in/out capability than you’ll ever need, high quality preamps, and quality digital converters for what something ‘brand new’ with a fraction of the features and quality will cost.
The catch there is if you ever decide to go with a new Mac, might never find drivers for a lot of the older stuff. If you’re thinking of trying a Mac, you’ll have little choice but to go for newer audio interfaces (a few of the best of the best like RME might offer drivers for Apple silicon on older interfaces, but then again, people don’t part with those so easily).
. Feel free to rip the build to pieces; any feedback is helpful.
