It’s an option! The prices on the base M1 Studios refurbished are pretty good, and they are still very good computers. The only thing that puts me off is how long Apple will support them with OS updates now that the M3 is out. It’s why I favour an M2 Pro Mini.
Both AMD and Intel have workstation chips with eight channels of DDR5, which is well over 200GBs.
Nobody knows but my hunch is around 7-10 years just like the Intel machines.
Another measure is per core bandwidth, which AFAIK Apple is way ahead of everyone else.
From when they were released or from today? 7 years from release seems fairly standard from Apple, longer for the Mac Pros and perhaps for the Studios to.
I get on just fine, as a fellow hobbyist mind you, with BBCSO Pro on M1 16GB RAM. Of course it depends on how "heavy" you mean: if I wanted several mic signals on every instrument in a large orchestral piece, I would expect to freeze tracks... I just haven't needed to.
I said you could avoid preloading as much in the first place. I didn't say you could fit 64GB of data into 24GB of RAM, and I didn't say anything about swapping. Next time you speak to everyone, could you ask them how they define "the same"?
I really wish people would read what I wrote, rather than imagining I said anything about swapping.
Seriously, I appreciate that fast swap is a common explanation for why Apple silicon requires less memory for general applications, but it's absolutely not the argument I was making. Swap should be irrelevant to sampled instruments (and if it is happening, something's probably broken.)I think your info may be out of date? Also, is the Apple M3 Max actually equivalent to a "desktop" processor from Intel or AMD, or is it more equivalent to, say, a Xeon or Epc? I think the Xeon is around 300 and Epyc is around 400 now. But maybe those are clusters of processors or something - I don't follow that closely.
M2 Max MBP user here with 96 gigs of ram. BBC SO pro template with all instruments loaded and three mics enabled on each instance takes a little under 20gigs of total ram. Have preloads set as low as they go. Loads quick and have had no issues. While the ram can’t hold more than what you have, lowering the preloads on all the sample players can make a big difference. My OT Berlin template that used to run at roughly 105 ish gigs on my 2013 Mac Pro runs around 60ish gigs on the the MBP.
Where do you store your samples? I’m currently using external SATA SSDs over USB.
No, preload buffers are currently default. However I’m running on a 2018 mid range Mac Mini that streams from external SATA drives over USB.
That’s helpful. Are those low preload buffers dependant on samples being streamed from the internal SSD?
No. I'm streaming from NVMe Thunderbolt (2 lane) drives, running Kontakt minimum preloads, never a problem. Projects load quicker that way, too. I'm on M1 Mac Studio, DP11.
So, your current RAM requirements are more substantial than they will be with min preload. Try it with your current setup. You may be pleasantly surprised. And you can use teh data when calculating what you'll need in your next computer.
BBCSO is on external SSD (Samsung T7) which connects as USB 3.1; this is an M1 from 2020, though, so you'll be able to do better today. Preloading 1/4 the Spitfire default works fine, and I've run with lower - especially if using larger rendering buffers. (I do a lot of experiments, so please do corroborate numbers with others in case I've mis-remembered.)
I think that's an underestimation, as a *lot* of software developers use Macs these days, and 16GB of RAM is nowhere near enough for anything serious. It's still likely a minority, but Apple know that the employers will 'pay up' for the additional specs, and so don't care.
Software developers must be a tiny minority of users surely?
Probably... although you'd also include any content creators as well. Any professional use, basically, which was previously mentioned, but it seems (from my perspective) that far more businesses give their employees high-spec macbooks, than I remember 10+ years ago.
Whoa there
Not sure what counts as serious or what you work on yourself, but I've worked on big data applications, simulation, signal processing, etc, all on 16GB or less. They might use 100s GB in production, but it's rarely good practice for devs to work against full-sized data-sets at all, and certainly not on their own machines.This does happen all the time, but IME that isn't evidence the specs are necessary: it's evidence that most firms haven't the slightest clue what they're doing when issuing IT to employees. (Clueful firms may still issue over-specced hardware, since aggressively standardising usually saves more money than buying cheaper models for some roles and more powerful units for others.)
