M1 Ultra vs M2 Max

[IFM]

My M1 Ultra studio is pretty rocking and won't disappoint. Go for the M1 Ulta with more ram/storage IMO because if you can keep your most often-used samples on the internal drive, you will find things running a lot faster overall. If you are coming from an intel chip, even a base M1 in the iMac will feel like a quantum shift in comparison.

 

[Phazma]

Good point. I guess it depends on where your bottleneck is. If you are constantly overloading a single core with a long plugin chain, M2 anything might be better. If you are running out of capacity on a large production, maybe more cores will be better. And it depends how much higher the M1 Ultra costs. If M2 is 20% faster, 12 M2 cores are 72% of 20 M1 cores, if all were equal. The choice might depend most on how you work.

Yeah, for sure a personal thing.
For me it’s most likely single core, as I get dropouts already when record arming a single track with Areia, Trailer strings and Pacific layered in an otherwise empty project, whereas I can have a full project, including that layered track, which also gives dropouts but doesn’t perform much worse (if record arm is disabled). That with an i7 2,2GHz Quad.

Still, I think any machine would solve all my bottlenecks. But what would be the next bottleneck in future is hard to predict. Still, the M2 Max is newer than the M1 Ultra (so at least a year longer support) and cheaper so I will go with that. But only time will tell if it’s the wiser choice and cost/performance ratio.

 

[Yamaoka86]

In terms of single vs multi core. It really depends on the project I’m working on.
My projects are most of the time hybrid scores with 50/50 synths/orchestra. Track count on my temp is around 400 in logic.

Oftentimes I need many UHE and NI massive instances. Something around 10-25 tracks. And of course the Omnisphere stuff.

One thing I noticed, first a single core overloads before my multi cores are in trouble. Therefore I would estimate single core power is still more important in my personal user case.

Sadly I cannot test any Silicon chips and must estimate my upcoming Mac purchase.

Still 50/50 m1 vs m2 for me…ughhh

 

[Yamaoka86]

another good comparison, he knows his stuff, but again no audio performance test.
The M1 Ultra had still excellent results compared to the M2 ultra. I was a bit surprised.

 

[TR909sample]

another good comparison, he knows his stuff, but again no audio performance test.
The M1 Ultra had still excellent results compared to the M2 ultra. I was a bit surprised.

I saw that video last night. Yeah, I agree. I like that guy because it was no-nonsense, straight to the point, he knows what he's talking about and he doesn't utilize click bait garbage thumbnails, unlike many other youtube channels. I tend not to take people seriously that make stupid looking faces.

Even though the comparison was based primarily around photo tools, I thought that there was a lot of good info to be gained from the video.

 

[gsilbers]

So, here https://9to5mac.com/2023/06/16/m2-m...and-m2-ultra-nearly-doubles-that-performance/

is an analysis of the tests in the video. They were GPU-heavy tests where the M2 Ultra saw the biggest performance jumps over the M1 Ultra. Here are a few quotes from the article:

"Despite the M1 Ultra literally being 2 M1 Max’s fused, the performance was never doubled."
"He also found that the M2 Ultra doubled the GPU performance of the M1 Ultra in these same benchmarks – a genuinely remarkable year-over-year upgrade."
"The reason for the massive performance improvement is that Apple added a memory controller chip to the M2 generation that balances the load between all of M2 Ultra’s cores – M1 Ultra required the ram to be maxed out before using all cores."

I don't know if the new memory controller chip helps with CPU-heavy work. It's very possible/probable that DAWs already balance CPU performance themselves across both of the M1 Ultra's "2 M1 Max's fused". (I'm not sure how to refer to them. The two processors of the Ultra? Maybe?) This is something DAWs have been dealing with for many, many years.

When the author compares the M2 Max to the M1 Ultra, he says, "While the GPU performance is similar, the M1 Ultra still has the advantage of far more CPU cores, and will outperform the M2 Max in CPU heavy workloads."

It sounds like an M1 Ultra might be the best for audio work. As they drop in price, they might become a real bargain.

very interesting. I dont know how this would translate to audio work. Im coming from a old mac pro 5,1 that both its single and multi core scores are terrible compared to any of the mac studios above. Needless to say i would be happy with any mac studio.

At the same time, here is the news that apple figured out something with its memory controller in the m2 that the m1 doesnt have and made a huge difference. There are some rumors that the m3 is going to be a big leap in performance. And im now starting to feel that maybe getting the best of the best like we once thought we needed to do with apple is no longer ncecesary and that steady incremental will be the norm ala iphone. Since ram or cpu or gpu cannot be upgraded then going all in into the most expensve might not be necessary in the audio world since already a lot of music producers have been very happy with the max.

On a side note, im on the mac pro 5,1 and mac mini intel dealing with hundreds of tracks in logic pro but i open pro tools ultimate and can barely load a dozen tracks before getting a cpu issue or errors for not being able to handle it.
In other words, the way things are coded might be different for different poeple in addition to the type of work. Large templates where even the maracas are in a single instrument track vs a "perc" patch with a bunch of small perc and so on.

My obvious ambivalence as to which mac studio to get might be due to the heavy change in workflow from mojave to ventura and installing and authorizing a decades worth of plungins that it would be dreadfull to have to say "I guess i should have bought the ultra instead of the max" and having to sell it and buy another mac and re do everything.

Ive learned that others "good" computer might be way too good for even the person claiming it. So many posts saying mvne/m.2 drive is a must on a thunderbolt port and you HAVE to get the best thunderbolt dock at $300 when im using a $10 usbc case with sata ssd drive of 8tb doing both audio drive and streaming without much hickups even w big projects. I start looking at some older posts and there's evildragon writing some math/specs on kontakt streaming voices vs disk speed and my assumptions where right, maybe we are over shooting it because we have become acustomed to buying and working a specific way and with apple "you have to buy the best that you will need" mantra but what you need is now a very blurry vision of specs vs workflows and new types of technologies like SOC.

Maybe the gap with SOC m series has widen so much than what we are used to that a pro max will be more than good for any profesional. Yet im still trying to see a video o post of someone with a max doing large template work.
Ive asked in the macbook pro m2 max thread but no one is answering. And still havent seen anyone working with the new macbook m2 max with large track count.

 

[Yamaoka86]

Just bought the M1Ultra/128/2Tb for 4700€. After three days searching the web the M1U still should be better than the M2max for audio production.

Had to pay 1500€ (30%) more for the same specs on the M2Ultra with 13%-19% more cpu power.

Maybe i took the 1500€ for the Studio Display (coming from the iMac Pro 5k) or a proper NAS (always postponed the latter).

 

[Nick Batzdorf]

So many posts saying mvne/m.2 drive is a must on a thunderbolt port and you HAVE to get the best thunderbolt dock at $300 when im using a $10 usbc case with sata ssd drive of 8tb doing both audio drive and streaming without much hickups even w big projects

Don't Thunderbolt docks divide the bandwidth anyway?

 

[HCMarkus]

When I was putting my Mac Studio rig together, the difference between SATA + USB3 SSDs and NVMe + Thunderbolt SSDs was so small that I just went NVMe. But, yes, SATA +USB3 is plenty fast enough for VI sample delivery.

For Audio and other files (which will be written and re-written multiple times as one moves from project to project), Thunderbolt does have the advantage of Trim support.

 

[Vik]

SATA +USB3 is plenty fast enough for VI sample delivery.

Could you tell more about how you came to that conclusion? How many 'pro' VI tracks do you run at the same time, how many mic positions/layers are active etc?

 

[gsilbers]

Could you tell more about how you came to that conclusion? How many 'pro' VI tracks do you run at the same time, how many mic positions/layers are active etc?

maybe its related but in usb audio interfaces rme mentioned a lot of voices can be in usb2 and the main engineer from rme posted the match somewhere i cannot find but showing how usb3 is already way above whats needed. I only found the usb2, which is plenty for audio interfaces.


https://www.rme-audio.de/rme-usb-technology.html



There was an old post by evildragon stating the math for streaming from disk and the speeds but below are some of the posts similar in nature. sata is more than good enough for streaming a lot of voices but m.2 loads faster the template. but sata/usb3 will be more than apt to stream a lot of voices.

I personally have usbc with sata ssd with all my libraries and its also an audio drive and can stream a lot of tracks. i get cpu issues before i get sample playback issues.
Ill still get the m.2 drive in the future as loading times drive me nuts.




s-c

 

[HCMarkus]

Could you tell more about how you came to that conclusion? How many 'pro' VI tracks do you run at the same time, how many mic positions/layers are active etc?

Please see gsiibers' post immediately below yours. EvilDragon's analysis is comprehensive.

 

[gsilbers]

Please see gsiibers' post immediately below yours. EvilDragon's analysis is comprehensive.

I could be wrong of him posting the math of this. Maybe it was someone else who added the math on how many voices would an orchestra have in kontakt sample instruments and the required drive type or raid or something along those lines. a very VI-centric thread but cannot find it out now .
Either way, ill be getting the nmve drives. Prices are very low so its a no brainer for me!.
What i WONT be doing is buying the mac studio with a lot of drive space cuz thats insanely expensive.

 

[Phazma]

There are currently several threads about SSDs, but as we are on topic here.. would a m.2 drive in a fitting enclosure over thunderbolt load Kontakt patches considerably faster than a simple, external USB 3.2 Gen2 nvme SSD?

Not sure if we are all on the same page, but with "Kontakt loading time" I mean the time it takes to complete this process:



Will a 2500MB/s read speed complete this process 2.5x faster than a 1000MB/s read speed?

Or will the bottleneck rather be CPU speed or Kontakt's code?

In general, I am curious to know if I will see a significant improvement in this already when upgrading my computer because of the better processor and faster ports, or when upgrading SSD (from 500MB/s with USB 3 SSD to either 1000 MB/s with USB 3.2 Gen2 SSD or to 2500MB/s with Thunderbolt m.2 SSD) or if it won't matter because Kontakt doesn't permit faster loading anyway.

 

[davidson]

There are currently several threads about SSDs, but as we are on topic here.. would a m.2 drive in a fitting enclosure over thunderbolt load Kontakt patches considerably faster than a simple, external USB 3.2 Gen2 nvme SSD?

Not sure if we are all on the same page, but with "Kontakt loading time" I mean the time it takes to complete this process:



Will a 2500MB/s read speed complete this process 2.5x faster than a 1000MB/s read speed?

Or will the bottleneck rather be CPU speed or Kontakt's code?

In general, I am curious to know if I will see a significant improvement in this already when upgrading my computer because of the better processor and faster ports, or when upgrading SSD (from 500MB/s with USB 3 SSD to either 1000 MB/s with USB 3.2 Gen2 SSD or to 2500MB/s with Thunderbolt m.2 SSD) or if it won't matter because Kontakt doesn't permit faster loading anyway.

I compared my internal mac studio drive to some of my external usb sata3 drives, and kontakt instruments loaded approx 10% faster in general. Other plugs seemed to do better such as omnisphere. The bottleneck seems to be kontakts codebase.

So the answers no!

 

[Vik]

Dealing with 5-600 megabytes/sec may be useful for a lot of stuff, but that won't take me back to rely on SATA drives. With a lot of layers, round robins, mic positions etc within one single instrument – plus sometimes wanting to layer multiple instruments, combined with the fact that orchestral libraries will keep developing, offer several types of legato transitions etc, I want to invest in storage that is ahead of what we need right now.

I believe that DAW workflows (in terms of dealing with large sample libraries) will change over time. This, combined with the high RAM prices (at least for Macs) for those who want 96/128 gigabytes and upward, suggests that various ways of bypassing the need for huge RAM amounts probably will pop up soon.


Many of those who work with complex orchestral mockups, don't really need to have all their 200 tracks being real time streamed from an internal or external drive. With some intelligent solutions (artificial intelligence or not), a good DAW should be able to figure out which tracks that need to be streamed from a drive, and which tracks that simply can be frozen/unfrozen, either manually, automatically or both. That's the main reason I want something which is much faster than eg. SATA. I already want something faster than 2800 megabytes second, since very high read/write speeds will make any kind of (background freezing/unfreezing) much faster. In Logic, we can freeze or unfreeze, say, 50 tracks with one click (select 50 tracks and click on the freeze buttons), and this process will potentially be much faster with very high internal/external read/write speeds.

 

[gsilbers]

 

[gsilbers]

I’m not sure this guys audio assessment is the most apt. Seems he only renders a project and therefore the ultra is best for a few seconds.

I’m leaning towards the max studio.

 

[Vik]

I’m not sure this guys audio assessment is the most apt.

Agree – I'm actually sure that his approach is not the most apt.


IMHO YouTube clips like this are relatively meaningless, along the lines of clips from some of the other youtubers who talk fast, compare apples and oranges, fail to inform about important aspects related to what they discuss and in general, are rather superficial.

Comparisons like these should IMO show CPU meters, since even some important apps (including Logic) fail to use all the cores for tasks that would benefit from doing so. Until they do that, these YT-guys should at least inform that the reason some tasks are slower than expected isn't that the computer is slow, but that the app sometimes only one core when they shouldn't.

When comparing five different models, like he does, he should also mention the size of the internal drive, since it's known that some of the basic model (in terms of storage) some times are slower than the others.

Instead of that somehow boring unboxing stuff, it would also be better, IMO, to mention that 14" M2 MBPs are slower than the 16" models.

When he is combining M1 and M2 based Macs, and also mixing Ultra, Max and Pro processors, and Studio, Mini and Macbook Pro models, he should go slower. The Logic part was less than a minute long – why don't they make more videos instead, and videos that are a little more in-depth than this one?

What he says about 'audio' can also be misleading for another reason: we (VI users) also work with audio, and often need a lot of power/cores/memory, while playing back audio files is a very different process. It's not only important to have a look at how many cores that are being used in this process, but these youtubers should also pay attention to, when doing stuff in Logic or other DAWs, which buffer size that is being used.

Another youtuber compared Logic performance between several Macs using a very high buffer size which ended up with very misleading results. I wish there was Ignore buttons on YouTube.

 

[gsilbers]

Agree – I'm actually sure that his approach is not the most apt.


IMHO YouTube clips like this are relatively meaningless, along the lines of clips from some of the other youtubers who talk fast, compare apples and oranges, fail to inform about important aspects related to what they discuss and in general, are rather superficial.

Comparisons like these should IMO show CPU meters, since even some important apps (including Logic) fail to use all the cores for tasks that would benefit from doing so. Until they do that, these YT-guys should at least inform that the reason some tasks are slower than expected isn't that the computer is slow, but that the app sometimes only one core when they shouldn't.

When comparing five different models, like he does, he should also mention the size of the internal drive, since it's known that some of the basic model (in terms of storage) some times are slower than the others.

Instead of that somehow boring unboxing stuff, it would also be better, IMO, to mention that 14" M2 MBPs are slower than the 16" models.

When he is combining M1 and M2 based Macs, and also mixing Ultra, Max and Pro processors, and Studio, Mini and Macbook Pro models, he should go slower. The Logic part was less than a minute long – why don't they make more videos instead, and videos that are a little more in-depth than this one?

What he says about 'audio' can also be misleading for another reason: we (VI users) also work with audio, and often need a lot of power/cores/memory, while playing back audio files is a very different process. It's not only important to have a look at how many cores that are being used in this process, but these youtubers should also pay attention to, when doing stuff in Logic or other DAWs, which buffer size that is being used.

Another youtuber compared Logic performance between several Macs using a very high buffer size which ended up with very misleading results. I wish there was Ignore buttons on YouTube.

I think most of these reviewers are video centric guys so it makes sense to them to do a render test. Aka bounce and the amount it takes. In their world thats a huge pain in the ass to have to sit their waiting. And logic seems to take into account more of the cores when bouncing.

Still, looking at all the other tests focusing on the cpu, makes me think the Studio m2 max will be good enough for all my audio needs. at 96gb of ram, i think its about what i need, for me at least. Ill be using turnoff memory track and im coming from an 5,1 mac pro. For about half the price of the ultra, i could at least try for a little while and used prices are still good enough.

Hopefully more poeple will make videos with the m2 max just to make sure theres anything wierd.