Hi, I don't do video editing. The scientific and engineering programs that I use still require Rosetta 2 even M1 has been around for a year. There is no telling when they will be running on M1 chips natively. In this case, any advantage in getting the M1 Max over M1 Pro when running applications that require Rosetta 2? I read that the M1 Max has double the memory speed but it seems that there is no gain in performance over the M1 Pro. I don't do video editing and programs that I use don't take advantage of M1's GPU at least for the time being and nobody knows if they will in the future.
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Any advantage in using M1 Max over M1 Pro when running applications that still require Rosetta 2?
- Thread starter hajime
- Start date
- Sort by reaction score
Yes max RAM can help because IMHO Rosetta is hog of memory! Intel only apps take up so much rAM that eats up all it can especially on big applications! That's why I only try to get Universal or better software for my Mac Book Air!
Rosetta doesn't matter here, M1 Max vs Pro generally comes down to these questions:
1. Do you need 64GB RAM?
2. Do you need higher GPU performance?
3. Do you need more than two external displays?
4. If you edit video, would you get something out of the additional hardware video codecs?
If your answer to all of these questions is "no", just buy M1 Pro.
Max does have double the memory bandwidth, but it doesn't really help the CPUs. It's there to feed the double-size GPU.
Rosetta apps do use slightly more memory than they would on an Intel Mac, because there's some overhead involved in translating Intel code to Arm code. Despite what satcomer claims, this is not a reason to assume you need to max RAM out. In typical apps memory use is mostly a function of data, not code.
1. Do you need 64GB RAM?
2. Do you need higher GPU performance?
3. Do you need more than two external displays?
4. If you edit video, would you get something out of the additional hardware video codecs?
If your answer to all of these questions is "no", just buy M1 Pro.
Max does have double the memory bandwidth, but it doesn't really help the CPUs. It's there to feed the double-size GPU.
Rosetta apps do use slightly more memory than they would on an Intel Mac, because there's some overhead involved in translating Intel code to Arm code. Despite what satcomer claims, this is not a reason to assume you need to max RAM out. In typical apps memory use is mostly a function of data, not code.
1. If applications I use can run M1 natively, I might need 64GB. Many years ago I even used over 100GB. Since they don’t run natively, 16 or 32GB is fine for the time being as I have a workstation PC.
2. If applications I use can take advantage of GPU, sure.
3. Currently I have a plan to buy a 5K2K and a 4K or two 5K displays. Not sure if there is an advantage in M1 Max here.
4. Haven’t edited videos for 20 years, don‘t know.
The purchase decision is difficult since software I use still cannot take advantage of the M1. I am thinking about M1 Pro 16GB and trade in when software catch up or just get M1 Max 32GB or 64GB and trade in when the next big redesign happen.
2. If applications I use can take advantage of GPU, sure.
3. Currently I have a plan to buy a 5K2K and a 4K or two 5K displays. Not sure if there is an advantage in M1 Max here.
4. Haven’t edited videos for 20 years, don‘t know.
The purchase decision is difficult since software I use still cannot take advantage of the M1. I am thinking about M1 Pro 16GB and trade in when software catch up or just get M1 Max 32GB or 64GB and trade in when the next big redesign happen.
Last edited:
To a certain degree, it is like iPad Pro having very powerful hardware but software can’t take advantage of it. I guess for M1 Mac, people who do video editing can enjoy the hardware but people like me can’t take advantage of it.
On the Max, the extra two memory interfaces are a constant extra power drain, so it generally has slightly worse battery life. If you're actively using the extra GPU resources, Max can have substantially worse battery life.
(There are some narrow cases where Max may have better battery life than Pro. Think fixed demand for GPU throughput, and that demand happens to be nearly 100% of the Pro's GPU. A Max running the same load will only need to run its GPU at half frequency, and that can result in less power consumption since the power/frequency curve of CPU and GPU cores isn't linear - the last few MHz cost a lot more power than the first few MHz.)
Strictly speaking, no. There was a performance increase that I could feel with some of my Java apps going from M1 to M1 Pro, but M1 Pro and M1 Max perform the exact same way. It's not a surprise to me considering the CPU configuration is about the same (2e + either 6p or 8p). In practice, even M1 to M1 Pro may not provide a performance increase depending on the workload. The only exception was with... engineering programs. Specifically, Autodesk Fusion 360 on M1 Pro is 2-3x faster to me compared to M1. On M1 Max, more complex models ran up to 1.5x faster than M1 Pro. This makes sense considering GPU scaling. If you need more GPU, regardless if an app is Rosetta 2 or not, M1 Pro and Max will provide a benefit.
This is kind of true and not at the same time. Consumer-level apps and even most pro apps tend to stress the CPU for processing a lot more, and so you don't ever see memory bandwidth being an issue. The only exception to this is if you are coding up algorithms that need to access large chunks of data at a time. In which case, more memory capacity and more memory bandwidth help tremendously. Then again, the M1 Max has more than just that: it also has more GPU cores. For machine learning, the M1 Max is the way to go for maximum performance.
Here's food for thought:
Apple’s M1 Pro and M1 Max Outperform Google Colab by up to 54%
2021 Apple M1 Pro and M1 Max Machine Learning speed test comparison.
towardsdatascience.com
Honestly, it's not theoretical. I went with my M1 Pro MacBook after extensive testing and honestly, it is the best machine learning laptop that I have used thus far. This may not be your use case but there are use cases where M1 Pro and Max do show a massive difference in performance, corresponding exactly to their supposed specs multipliers. The only problem is that... as noted, M1 Max does draw more power than M1 Pro. In practice, M1 Pro is the better compromise for performance and efficiency, I think. M1 Max is the absolute limit for what the 14" chassis can handle and if you're planning to go that route, you should go with the 16" MacBook.
Strictly speaking, no. There was a performance increase that I could feel with some of my Java apps going from M1 to M1 Pro, but M1 Pro and M1 Max perform the exact same way. It's not a surprise to me considering the CPU configuration is about the same (2e + either 6p or 8p). In practice, even M1 to M1 Pro may not provide a performance increase depending on the workload. The only exception was with... engineering programs. Specifically, Autodesk Fusion 360 on M1 Pro is 2-3x faster to me compared to M1. On M1 Max, more complex models ran up to 1.5x faster than M1 Pro. This makes sense considering GPU scaling. If you need more GPU, regardless if an app is Rosetta 2 or not, M1 Pro and Max will provide a benefit.
This is kind of true and not at the same time. Consumer-level apps and even most pro apps tend to stress the CPU for processing a lot more, and so you don't ever see memory bandwidth being an issue. The only exception to this is if you are coding up algorithms that need to access large chunks of data at a time. In which case, more memory capacity and more memory bandwidth help tremendously. Then again, the M1 Max has more than just that: it also has more GPU cores. For machine learning, the M1 Max is the way to go for maximum performance.
Here's food for thought:
Apple’s M1 Pro and M1 Max Outperform Google Colab by up to 54%
2021 Apple M1 Pro and M1 Max Machine Learning speed test comparison.
Honestly, it's not theoretical. I went with my M1 Pro MacBook after extensive testing and honestly, it is the best machine learning laptop that I have used thus far. This may not be your use case but there are use cases where M1 Pro and Max do show a massive difference in performance, corresponding exactly to their supposed specs multipliers. The only problem is that... as noted, M1 Max does draw more power than M1 Pro. In practice, M1 Pro is the better compromise for performance and efficiency, I think. M1 Max is the absolute limit for what the 14" chassis can handle and if you're planning to go that route, you should go with the 16" MacBook.
I think most software for machine learning cannot take advantage of the M1 GPU and not all of them are M1 optimized. For other applications, a lot of software that have both PC and Windows versions still have the M1 version under development. They take forever to get it done.
Due to the long wait time and lack of support for M1 and GPU in software, I try to decide between MacBook Pro 16" M1 Pro 16GB and M1 Max 32GB. If the MacBook Pro M1 were able to run Windows natively, that would be a very easy purchase decision.
When I tried the MacBook Pro 16" with M1 Pro last year, when idling, it took about 14GB out of 16GB memory for no reason. It could be due to a memory leak bug in the OS. Has it been fixed? Since memory is shared between CPU and GPU, if I connect a MacBook Pro to two or three 4K or 5K displays, does that mean I will have less memory to run applications?
14 out of 16GB used while idle doesn't mean much. macOS uses otherwise free RAM to cache data that has been read from disk, and will purge cache as needed to free up memory when applications try to allocate some.
The amount of memory required to support another display is mostly the frame buffer. A 5K display needs 5120*2880*4 = 58,982,400 bytes - round up to 64MB. They might do double buffering, so figure on about 128MB total. So yes, less memory for applications, but it shouldn't be much, even with very high pixel count displays.