My Intel i7 Mac mini has 64GB of OWC ram and the 2TB SSD soldered in by Apple. I have seen over 40GB of the ram being used as cache during massive file transfers. Perhaps the SSD could be smaller, but memory size does impact server activity.
Since I have a M1 Ultra MacStudio (128GB ram and 8TB SSD) now, when Apple drops support for Intel processors and I retire it for MacOS security reasons, I would put the M1 MacStudio into the job and get the most recent MacStudio. Since we can not change either memory or SSD after the build, we use what we have. Apple is offering $1,580 as a trade-in value for my MacStudio. The M2 Pro mini with 32GB os ram and a 2TB SSD lists for $2,599 before adding Apple care and sales tax. The M3 Pro mini would be that or higher.
Apple crippled the M3 Pro chip so the most memory of 32GB for the M2 Pro is now just 36MB for the M3 Pro. I had hoped for 64GB of ram as the memory had doubled going to the M2 Pro generation from the M1 Pro generation.
If the future M3 or higher Mac Studio failed, I could fall back to the M1 Mac Studio. Remember that a component failure results in a completely new motherboard and a quick replacement could be problematic and its would be a used board.
It seems all four M1 chips had nearly the same single core performance. We saw that all four of the M2 chips also had similar single core speeds and there was about a 15% increase in the average number. So, for me, there was no incentive to upgrade to the M2 series. The first three of four M3 chips repeat the average single core speed concept of the M1 and M2. But there was a larger percentage increase in the average single core speed over the M2.
I am intrigued that a M3 Max MacBook Pro can now be acquired that is nearly as powerful as my M1 MacStudio (128GB and 8TB SSD) with the same amount of memory and SSD in just two generations. I thought I had really future proofed my M1 MacStudio with those top go the line options. And the fully configured MacBook Prop price is fairly close to my MacStudio price.
I thought we had a good performance boost going from the 68030 in my IIci to the 68040 in my IIfx. The Intel years were snoozers in terms of performance increases over time.
But I am concerned how many more rabbits are in the hat for this M series chip to continue this pace of spec improvements. The crowd that needs and can afford the top models is far smaller than the more modest needs and prices group. With the lower operating temperatures of the M series computers, one could expect much longer service lives of the M devices.
Like the iPhone, the incremental changes are getting smaller and the crowd seems to be wanting to spread their acquisition cost over more years.
So could Apple turn off operating system support for the M1 series at some future time like they will Intel chips? And how many generations of M series will be supported?
The computing power necessary for the average non-pro users is already exceeded. One usually does not see a Ferrari V8 in a Karman Gaia.
The M1 architecture is the same as the M3. There’s no reason Apple can’t support the M1 as long as it supports the M3
So could Apple turn off operating system support for the M1 series at some future time like they will Intel chips? And how many generations of M series will be supported?
Apple control the operating system, they can turn off support for anything at any point. Why would they though, without good reason? It'd ruin their reputation.
Wonder if the price will be close to the prior generation like last time?
I may be tempted this time and take my M1 Ultra Mac Studio (128GB & 8TB SSD) and use it as my file server to replace the aging and soon to be unsupported Intel i7 powered Mac mini (64GB OWC Ram and 2 TB soldered SSD). The M1 Mac Studio trade in value is less than a possible M3 Pro mini with maximum memory and a 2 TB SD. and probably will be supported for quite a few more years.
I remember in the 70s that a 32kb memory card from Digital Equipment Corp (DEC) for my PDP-11 cost $2,000. Every generation that followed in the same form factor doubled the amount of memory for the same $2,000. The last one was 4MB. I used 1.5MB for programs and 2.5MB as a virtual swap disc for compiling Dibol code which was blazingly faster than any of the spinning platters of the day.
Just why? You don't need that much RAM or CPU for a file server!!!! I have base Mac mini as a file server and it's very cool (sipping only like 10 watts on average). A file server does not take much RAM at all. It's just transferring files.
Ultra Mac is built for media creation especially with videos and photography, not for file servers.
The energy savings and the performance of the Apple Silicon chips is too great for Apple to sit on the server sidelines forever…… Apple doesn’t need to make a server next week but it is inevitable.
It seems all four M1 chips had nearly the same single core performance. We saw that all four of the M2 chips also had similar single core speeds and there was about a 15% increase in the average number. So, for me, there was no incentive to upgrade to the M2 series. The first three of four M3 chips repeat the average single core speed concept of the M1 and M2. But there was a larger percentage increase in the average single core speed over the M2.
I am intrigued that a M3 Max MacBook Pro can now be acquired that is nearly as powerful as my M1 MacStudio (128GB and 8TB SSD) with the same amount of memory and SSD in just two generations. I thought I had really future proofed my M1 MacStudio with those top go the line options. And the fully configured MacBook Prop price is fairly close to my MacStudio price.
I thought we had a good performance boost going from the 68030 in my IIci to the 68040 in my IIfx. The Intel years were snoozers in terms of performance increases over time.
But I am concerned how many more rabbits are in the hat for this M series chip to continue this pace of spec improvements. The crowd that needs and can afford the top models is far smaller than the more modest needs and prices group. With the lower operating temperatures of the M series computers, one could expect much longer service lives of the M devices.
Like the iPhone, the incremental changes are getting smaller and the crowd seems to be wanting to spread their acquisition cost over more years.
So could Apple turn off operating system support for the M1 series at some future time like they will Intel chips? And how many generations of M series will be supported?
The computing power necessary for the average non-pro users is already exceeded. One usually does not see a Ferrari V8 in a Karman Gaia.
By adding CXL for PCIe and Fabric, Apple could do a joint venture SOC with a 3rd party GPU die provider. Say an Mx Azmith for a Max Die and an AMD Navi Die in one memory-shared package.
That sort of move would put a few cats in with the pigeons.
I thought we had a good performance boost going from the 68030 in my IIci to the 68040 in my IIfx. The Intel years were snoozers in terms of performance increases over time.
The iifx also had a 68030, just with a higher clock speed.
I wonder how much the hypothetical advantages of Apple’s integrated GPU approach have actually been realized in practice. With discrete CPU, GPU, and RAM there’s a lot more flexibility to optimize for different workloads. Are the performance advantages of a shared memory space large enough to justify the lack of flexibility?
I can believe the performance/watt advantages are real, especially in mobile. But for the Mac Pro I really wonder if a more modular approach would be better.
I wonder how much the hypothetical advantages of Apple’s integrated GPU approach have actually been realized in practice. With discrete CPU, GPU, and RAM there’s a lot more flexibility to optimize for different workloads. Are the performance advantages of a shared memory space large enough to justify the lack of flexibility?
I can believe the performance/watt advantages are real, especially in mobile. But for the Mac Pro I really wonder if a more modular approach would be better.
Apple spent many years with the modular approach, it didn’t work out so well with Motorola, IBM, Intel, AMD, and Nvidia, being tied to someone else’s schedule in the end just didn’t work. The only company left to be tossed out is Qualcomm and Apple is working patiently behind the scenes to get rid of them too. After all, how do you fit in the future all that is the Apple Vision Pro into a frame of eyeglasses, tied to that menagerie of companies?
I wonder how much the hypothetical advantages of Apple’s integrated GPU approach have actually been realized in practice. With discrete CPU, GPU, and RAM there’s a lot more flexibility to optimize for different workloads. Are the performance advantages of a shared memory space large enough to justify the lack of flexibility?
I can believe the performance/watt advantages are real, especially in mobile. But for the Mac Pro I really wonder if a more modular approach would be better.
It seems to be quite the advantage, considering that M series performs with the big boys in both cpu and GPU while using a fraction of the power of only one of those on the competitors side. It’s also why Qualcomm and others have straight up copied Apple’s approach.
That said, there is still the possibility of modular approaches.
Apple can design a system level “fabric” that connects multiple Ultra SOCs plug Hines into the system.
Another option is to add a GPU booster package - as theorized by another member - with more GPU cores in between the two Max chips making up the Ultra.
Or a combination of the approaches.
It’s all about what Apple wants to do.
Performance today is better than it has ever been - to the point where folks used to old standby computing paradigms wonder if we need anymore. And yet, new computing paradigms exist already that require even more power than we have now. Creatives are really pushing the envelope, ai is burgeoning, and complex medical/scientific applications always benefit from speed. What took years took months, then weeks, and in the future could perhaps be real-time.
It’s basically up to Apple to determine how much that matters to them.
Apple used to own Hollywood and they were using supercomputer tech.
Then, apple seemed to distance itself from the world of high performance computing circa 2016 Mac Pro. But in 2019, apple seemed to get serious again. Two years later, we had apple making their own silicon. Not only commodity SOCs, but high performance versions as well. Last years mac pro looked like apple was losing its hunger to be on top of the heap. But then we got the m3 max, which is a monster and the ultra isn’t even out yet.
So it seems that apple still wants to rule the roost, but they’re figuring out how to do that while keeping ahead in the profits game. Apple silicon is a nice foundation. But I doubt we’ve seen the extent of SOC configurations - or even system level configurations such as a modular Mac Pro. But one thing I seriously doubt we will see any time in the near future is apple adding third party GPUs or anything like that. Apple wants to BE the leader, not be the platform that feeds the leader. But again, they’re figuring out how to do that while being responsible with energy and maximizing profits - especially difficult immediately post-pandemic and in the current crazy inflation climate.
I wonder how much the hypothetical advantages of Apple’s integrated GPU approach have actually been realized in practice. With discrete CPU, GPU, and RAM there’s a lot more flexibility to optimize for different workloads. Are the performance advantages of a shared memory space large enough to justify the lack of flexibility?
I can believe the performance/watt advantages are real, especially in mobile. But for the Mac Pro I really wonder if a more modular approach would be better.
Apple spent many years with the modular approach, it didn’t work out so well with Motorola, IBM, Intel, AMD, and Nvidia, being tied to someone else’s schedule in the end just didn’t work. The only company left to be tossed out is Qualcomm and Apple is working patiently behind the scenes to get rid of them too. After all, how do you fit in the future all that is the Apple Vision Pro into a frame of eyeglasses, tied to that menagerie of companies?
Apple can go modular and still use their own cpu and GPU designs.
Comments
Since I have a M1 Ultra MacStudio (128GB ram and 8TB SSD) now, when Apple drops support for Intel processors and I retire it for MacOS security reasons, I would put the M1 MacStudio into the job and get the most recent MacStudio. Since we can not change either memory or SSD after the build, we use what we have. Apple is offering $1,580 as a trade-in value for my MacStudio. The M2 Pro mini with 32GB os ram and a 2TB SSD lists for $2,599 before adding Apple care and sales tax. The M3 Pro mini would be that or higher.
Apple crippled the M3 Pro chip so the most memory of 32GB for the M2 Pro is now just 36MB for the M3 Pro. I had hoped for 64GB of ram as the memory had doubled going to the M2 Pro generation from the M1 Pro generation.
If the future M3 or higher Mac Studio failed, I could fall back to the M1 Mac Studio. Remember that a component failure results in a completely new motherboard and a quick replacement could be problematic and its would be a used board.
The energy savings and the performance of the Apple Silicon chips is too great for Apple to sit on the server sidelines forever…… Apple doesn’t need to make a server next week but it is inevitable.
That sort of move would put a few cats in with the pigeons.
I can believe the performance/watt advantages are real, especially in mobile. But for the Mac Pro I really wonder if a more modular approach would be better.