New "Mac Studio" from Apple

[Phil O'Keefe]

Originally Posted by huzzam

Just to keep perspective, one mono 24-bit 48kHz audio channel has a bitrate of 144,000 bytes/sec, or ~0.137MB/s. (Double that if you use 96 kHz.) So even a thousand mono 24/48 tracks would use only 137MB/s. That's only about 6% of your "slow" internal SSD's bandwidth. (And 22% of a decent SATA3 SSD's bandwidth.)



The moral: SSDs are not a bottleneck anymore for audio work. We can just forget it completely at this point. The only place where it comes into play is when we're moving entire sessions around, but for tracking/mixing, it's a non-issue.

Agreed. At least for a lot of people / workflows.

And if your sessions are small enough, you can just load all the audio files into RAM and take the storage completely out of the equation. Or use it just for streaming samples for your VIs.

I don’t think the Thunderbolt ports and their bandwidth are going to be a big bottleneck on the Mac Studios.

I don’t think the internal SSD speed will be, either. I was happy to see that those are socketed, so even if we work them hard and one fails, Apple should be able to fix them. Heck, we may even be able to upgrade the onboard storage eventually.

I can see where some users will need tons of RAM and several of the fastest SSDs they can find, but that doesn’t apply to everyone.

[Phil O'Keefe]

Originally Posted by off the wall

The M2 from all reports will be a replacement for the lower end M1, not an evolution for the higher end M1|Pro, Max and Ultra.

I’ve seen speculative videos suggesting it will be a replacement for, and an improvement over the M1 - possibly with a couple of extra CPU and graphics cores. Other videos and articles suggest the single core performance will be somewhat improved, too.

Others speculate that the M2 will be scalable, just like the M1, and that we’ll eventually see M2 Pro, Max, Ultra, and possibly even “extreme” versions that use four interconnected Max SoCs.

I don’t see Avid getting an M1 native build out anytime soon. The performance in Rosetta 2 is good enough for them to take their time, as is frustratingly their usual course of action.

I’m happy they got the plugin bundle updated. Sure, I’d love to see a solid, stable, M1 native version that didn’t require Rosetta released next week, but Ultimate 2021.12 has been really stable for me, and I’d rather they take their time and get it right on the next version. But there are going to be a lot of people who will be using PT on new M1 Macs, so there’s going to be a lot of demand for them to release a fully-compatible version.

My guess? Sometime this calendar year. We’ll see.

[off the wall]

Originally Posted by Phil O'Keefe

I’ve seen speculative videos suggesting it will be a replacement for, and an improvement over the M1 - possibly with a couple of extra CPU and graphics cores. Other videos and articles suggest the single core performance will be somewhat improved, too.

Others speculate that the M2 will be scalable, just like the M1, and that we’ll eventually see M2 Pro, Max, Ultra, and possibly even “extreme” versions that use four interconnected Max SoCs.



I’m happy they got the plugin bundle updated. Sure, I’d love to see a solid, stable, M1 native version that didn’t require Rosetta released next week, but Ultimate 2021.12 has been really stable for me, and I’d rather they take their time and get it right on the next version. But there are going to be a lot of people who will be using PT on new M1 Macs, so there’s going to be a lot of demand for them to release a fully-compatible version.

My guess? Sometime this calendar year. We’ll see.

No arguments on any of that.

I predicted PT would be cleared to run in Rosetta on Monterey in March 2022 back in Oct.

On that dumb luck, shot in the dark guess, I’m sticking with my end of 2022 prediction as the earliest we will see a native M1 PT build.

[ramonetl]

Originally Posted by James67

I don't think this is off track at all, because I was interested in swapping my Mac Pro for a Mac Studio due to the huge difference in single core performance. .

Depends. If I had the 2019 or later Mac Pro absolutely not. You are not in the sweet spot for the ROI and its still way more upgradeable and for longevity you have a lot of life. Besides you already sold your soul to the devil to get it. If its a trash can Pro like I have then yes hurry up. Im expecting mine tomorrow. As always YMMV...

[Eric Lambert]

Originally Posted by James67

I don't think this is off track at all, because I was interested in swapping my Mac Pro for a Mac Studio due to the huge difference in single core performance.

Good point.

[Frank Kruse]

Some ProTools benchmarks I collected from users running the "SOL LEVANTE" open source Netflix PT-Session (https://opencontent.netflix.com/). Make of it what you want.

(based on the PT system usage info)

Studio Ultra 128GB RAM: 90-100% (red bars from time to time but will play fine without stopping)

7,1 16core 20-30%
7,1 8-core 48GB RAM around 60%

6,1 6-core won't play the session without CPU errors. Started playing after de-activating a couple of multiband processors plugins on the MX stem)
6,1 12-core 80-90% CPU but no stopping.

5,1 48GB RAM 12-core 55-75%, no errors. no red bars. (this is my own machine)

i9 PC-Based Hackintosh (OpenCore) 50-60%

Mini M1: won't play the session without CPU errors (test done pre plugin update release with de-activated plugins).

M1 Laptops: So far co-workers couldn't get the session to play without stopping (some won't even start playing without a "too slow"-error) without deactivating plugins

All tests done with PT 2021.12 and the AAX plugins update from 2022. Buffer at 1024, dyn. plugin processing ON, ignore errors: OFF. disk cache: normal. Picture at full quality.

Studio Ultra will play down to buffer of 32 with red bars, but won't stop.

On all the systems the PT CPU meters corresponded to the mac's system activity info. Not sure about the Mac Studio though. What's also interesting is that the Studio Ultra doesn't seem to care about red bars and 100% CPU load. It will keep playing even with dramatically more CPU load than the 7,1 and more than a 5,1 12-core.

What was interesting to see is that the M1-Mini and all the M1 laptops are no real improvement over the 5,1 12-core. At least not when running PT.
Of course this not taking into account speed achieved outside PT (bus speeds, transfer speeds etc.)

[Carl Lie]

Originally Posted by Frank Kruse

Some ProTools benchmarks I collected from users running the "SOL LEVANTE" open source Netflix PT-Session (https://opencontent.netflix.com/). Make of it what you want.

(based on the PT system usage info)

Studio Ultra 128GB RAM: 90-100% (red bars from time to time but will play fine without stopping)

7,1 16core 20-30%
7,1 8-core 48GB RAM around 60%

6,1 6-core won't play the session without CPU errors. Started playing after de-activating a couple of multiband processors plugins on the MX stem)
6,1 12-core test pending.

5,1 48GB RAM 12-core 55-75%, no errors. no red bars. (this is my own machine)

i9 PC-Based Hackintosh (OpenCore) 50-60%

Mini M1: won't play the session without CPU errors (test done pre plugin update release with de-activated plugins).

All tests done with PT 2021.12. Buffer at 1024, dyn. plugin processing ON, ignore errors: OFF. disk cache: normal. Picture at full quality.

Studio Ultra will play down to buffer of 32 with red bars, but won't stop.

On all the systems the PT CPU meters corresponded to the mac's system activity info. Not sure about the Mac Studio though. What's also interesting is that the Studio Ultra doesn't seem to care about red bars and 100% CPU load. It will keep playing even with dramatically more CPU load than the 7,1 and more than a 5,1 12-core.

What was interesting to see is that the M1-Mini and all the M1 laptops are no real improvement over the 5,1 12-core. At least not when running PT.
Of course this not taking into account speed achieved outside PT (bus speeds, transfer speeds etc.)

Thanks for posting. Will be interesting to see what the next mac pro is like. May not be so great for PT no matter how fast the benchmarks are. Hopefully when Avid codes for Apple silicon we'll see much better performance.

C

[Eric Lambert]

It seems that ProTools' CPU meters on a Mac Studio are even less indicative of true CPU performance than on other machines.

[audiolex1]

Originally Posted by Frank Kruse

Some ProTools benchmarks I collected from users running the "SOL LEVANTE" open source Netflix PT-Session (https://opencontent.netflix.com/). Make of it what you want.

(based on the PT system usage info)

Studio Ultra 128GB RAM: 90-100% (red bars from time to time but will play fine without stopping)

7,1 16core 20-30%
7,1 8-core 48GB RAM around 60%

6,1 6-core won't play the session without CPU errors. Started playing after de-activating a couple of multiband processors plugins on the MX stem)
6,1 12-core test pending.

5,1 48GB RAM 12-core 55-75%, no errors. no red bars. (this is my own machine)

i9 PC-Based Hackintosh (OpenCore) 50-60%

Mini M1: won't play the session without CPU errors (test done pre plugin update release with de-activated plugins).

All tests done with PT 2021.12. Buffer at 1024, dyn. plugin processing ON, ignore errors: OFF. disk cache: normal. Picture at full quality.

Studio Ultra will play down to buffer of 32 with red bars, but won't stop.

On all the systems the PT CPU meters corresponded to the mac's system activity info. Not sure about the Mac Studio though. What's also interesting is that the Studio Ultra doesn't seem to care about red bars and 100% CPU load. It will keep playing even with dramatically more CPU load than the 7,1 and more than a 5,1 12-core.

What was interesting to see is that the M1-Mini and all the M1 laptops are no real improvement over the 5,1 12-core. At least not when running PT.
Of course this not taking into account speed achieved outside PT (bus speeds, transfer speeds etc.)

Did you down load the PT Final Mix or Mastered zip?

[kings79]

Originally Posted by Frank Kruse

Some ProTools benchmarks I collected from users running the "SOL LEVANTE" open source Netflix PT-Session (https://opencontent.netflix.com/). Make of it what you want.

(based on the PT system usage info)

Studio Ultra 128GB RAM: 90-100% (red bars from time to time but will play fine without stopping)

7,1 16core 20-30%
7,1 8-core 48GB RAM around 60%

6,1 6-core won't play the session without CPU errors. Started playing after de-activating a couple of multiband processors plugins on the MX stem)
6,1 12-core test pending.

5,1 48GB RAM 12-core 55-75%, no errors. no red bars. (this is my own machine)

i9 PC-Based Hackintosh (OpenCore) 50-60%

Mini M1: won't play the session without CPU errors (test done pre plugin update release with de-activated plugins).

All tests done with PT 2021.12. Buffer at 1024, dyn. plugin processing ON, ignore errors: OFF. disk cache: normal. Picture at full quality.

Studio Ultra will play down to buffer of 32 with red bars, but won't stop.

On all the systems the PT CPU meters corresponded to the mac's system activity info. Not sure about the Mac Studio though. What's also interesting is that the Studio Ultra doesn't seem to care about red bars and 100% CPU load. It will keep playing even with dramatically more CPU load than the 7,1 and more than a 5,1 12-core.

What was interesting to see is that the M1-Mini and all the M1 laptops are no real improvement over the 5,1 12-core. At least not when running PT.
Of course this not taking into account speed achieved outside PT (bus speeds, transfer speeds etc.)

No M1 Max results?

Yeah the PT CPU Monitor is useless and should be ignored.