Mac Studio announced - Apple's most powerful computer yet

[sQeetz]

likely these limitations can be superceded by turning all security off. I haven't tested that theory yet.

windows with a sound card asio is faster. anything with thunderbolt or usb isn't as fast as a pcie connection mostly because that every connection is tied a serial in/out connection proxy firewall and security setting, rather also than routing that one connection to various outs in parralell; it's designed in a paranoid forcing re-iteration. I assume that would prevent any kind of malware from changing the sound itself, it does seem bizzare though, should that happen. of course when tpm's on processors are not serial and have a layer for every core any latency that encryption once imposed will be a thing of the past.

using encryption and forcing a serial route does away with any added speed that a multi-core processor can achieve. it's a simple thing.

maybe though i was thinking that the newer mac m2 or whatever have considered this advance. it doesn't seem rocket science. i notice that cole guy on youtube just compared pro and studio, maybe that will be useful.

Quoting the nonsense in its entirety just for the sake of having it on record.

[BertKoor]

> it's designed in a paranoid forcing re-iteration. I assume that would prevent any kind of malware from changing the sound itself

Yes, assumption is the mother of all feck-ups.

[rod_zero]

Haven't seen a case of dunning kruger sohard in a while, also his "problem" is something so common in music production that if it was really a problem people wouldn't be using DAWs xD

[jamcat]

Anyone know if Apple Silicon cores process a single thread at a time or two threads per cycle like Intels?

[sQeetz]

Anyone know if Apple Silicon cores process a single thread at a time or two threads per cycle like Intels?

Hyper Thread? I think it's an Intel thing... Although I'm lead to believe that paralleization of instructions might not be such an issue with fixed width instruction sets like on RISC processors. But I'm just as knowledgeable on such an issue like any other who does not know exactly about CPU architectures

[sQeetz]

... but no there's no error. I have over 20 years experience with setups.

to quote rod_zero just a few posts above:

Haven't seen a case of dunning kruger sohard in a while

[jamcat]

So I looked into it, and it looks like M1 does not support simultaneous multithreading. It seems it’s power and memory inefficient. Intel is also apparently moving away from it too.

[machinesworking]

I won't indulge in replies to this again, but no there's no error. I have over 20 years experience with setups.

Not trying to pile on, but I would bet more than half of us on KVR have over 20 years of experience with setups. Experience does not automagically mean actual knowledge on the subject. I consider myself mediocre in terms of knowledge and that's far above the average 20+ year user of a DAW. Most people are set it and forget it. Nevertheless I've helped troubleshoot at least a dozens peoples setups in the last year, all top notch professionals.

We all learn something new if we're open to it. I recently learned that samplers like Kontakt are not really helped by super fast SSDs, the load time is as much about your CPU as it is about the drive you're using.

[sQeetz]

there IS NO error! No fu****G way!
I'll make a t-shirt of it.

[machinesworking]

So I looked into it, and it looks like M1 does not support simultaneous multithreading. It seems it’s power and memory inefficient. Intel is also apparently moving away from it too.

I'm the dork that read all about that for months when the beta M1 minis came out. It's not really surprising M1 doesn't "support" it, multithreading is a great trick to get more out of the x86 design, but in the end it's a useful hack. Any chip can run multiple operations at the same time, apparently using multithreading allowed x86 style chips to take a percentage more out of each core. [I didn't get solid numbers on this, but at least 10-15%] So if Apple do it "right"? they shouldn't need to do this.

It also doesn't do AVX, but apparently it doesn't need to, even running in Rosetta, according to NI.

Please note: MASSIVE X requires an Intel processor with AVX or an Apple Silicon processor. On Apple Silicon machines, Rosetta 2 must be installed in order to install and use MASSIVE X.

[jamcat]

I had assumed that SMT executed one thread on the low signal and one thread on the high signal of the clock, but apparently it has nothing to do with that, and actually just works by duplicating some parts of the CPU, so it’s more like a partial second core. Not really much magic there.

The reason I was wondering is so I could know how many dedicated audio tracks you will be able to do without sharing processing with other tracks. SMT gives you 2 per core.

The problem with Intel of course is Windows handles multiprocessing poorly so you get some cores barely being used while others are maxing out. Apple seems to handle thread distribution WAY better.

[kperry]

Plenty of RISC chips use SMT in some shape or form (IBM Power chips for example).

[machinesworking]

I had assumed that SMT executed one thread on the low signal and one thread on the high signal of the clock, but apparently it has nothing to do with that, and actually just works by duplicating some parts of the CPU, so it’s more like a partial second core. Not really much magic there.

The reason I was wondering is so I could know how many dedicated audio tracks you will be able to do without sharing processing with other tracks. SMT gives you 2 per core.

The problem with Intel of course is Windows handles multiprocessing poorly so you get some cores barely being used while others are maxing out. Apple seems to handle thread distribution WAY better.

Therein lies the rube though, it's duplicating "some parts of the CPU", the whole design is to maximize a single core with two threads, if they could flatly market it as separate cores they would. We will see, but so far Apple is not using this technique to get more juice out of their chips and IMO it's because they don't have to.

It's always amused me to some degree how Logic of all DAWs still has issues with CPU piling up on the initial core, it's easily solvable by stopping and starting the sequence, but it's still a WTF? moment to watch Apples DAW pile all processes on a single core until you stop the sequence and it adjusts itself.

[gearwatcher]

Actually ARMs are, in a way, VLIW processors (although so are the SIMD units in x86 CPUs) capable of decoding and executing some of the instructions in the ISA in parallel (2 or 4 at the same time).

The design of the processors doesn't need SMT ("Hyperthreading" in Intel marketing parlance) to achieve maximum utilisation of it's pipelines, doubly so with Apple Silicon and it's obscene memory throughput.

[2DaT]

Actually ARMs are, in a way, VLIW processors (although so are the SIMD units in x86 CPUs) capable of decoding and executing some of the instructions in the ISA in parallel (2 or 4 at the same time).

This is wrong use of terminology. Just because x86 and ARM use uops doesn't make it VLIW. Although you can say that between scheduler and execution units it runs "like a" VLIW, schedulers are extremely complicated. Real VLIW is about simple schedulers.

It doesn't really matter all these acronyms (RISC,CISC,SIMD, VLIW, SMT, etc.), unless you actually write code for the platform. For anybody who is not a programmer, it's absolutely meaningless.

End users should compare real performance of real software and not reason about things they don't understand.