Poll: When will there be an arm mac that exceeds a 2019 mac pro in performance?

[burgerrecords]

Specifically let's say an ARM mac that exceeds the blender performance of the fully loaded 2019 mac pro?

 

[0128672]

I'm throwing a dart and it landed on 2 years, 17 days, 6 hours and 43 minutes.

 

[ian87w]

Apple wouldn't introduce a new machine that is slower than the previous generation. It would defeat the purpose. Based on the PPC to intel transition, the PowerMac was the last Mac to be transitioned to intel/Mac Pro. I would say the same thing here.

The question is, how soon would apps and devs take full advantage of the new architecture to maximize its potential?

 

[Flint Ironstag]

I’d wager the prototype already exists.

 

[Nermal]

Apple wouldn't introduce a new machine that is slower than the previous generation.

*Looks at 2014 Mac Mini*

 

[Boil]

Mac Pro Cube - starting at US$3,999.00

48 P cores / 4 E cores / 96 GPU cores - CPU / GPU Chiplets & RAM on interposer / package design
HBM3 Unified Memory Architecture - 128GB / 256GB / 512GB
NVMe RAID 0 (dual NAND blades) 4TB / 8TB / 16TB
Four USB4 (TB3) ports
Four TB4 ports
Two 10Gb Ethernet ports
One HDMI 2.1 port
One MPX-C slot (for use with GPGPU expansion card)

;^p

 

[burgerrecords]

Apple wouldn't introduce a new machine that is slower than the previous generation. It would defeat the purpose. Based on the PPC to intel transition, the PowerMac was the last Mac to be transitioned to intel/Mac Pro. I would say the same thing here.

The question is, how soon would apps and devs take full advantage of the new architecture to maximize its potential?

I don’t think anyone would expect a new Mac Pro to be slower, but I don’t think they’re going to release a new Mac Pro right away.

 

[Stene]

Mac Pro Cube - starting at US$3,999.00

48 P cores / 4 E cores / 96 GPU cores - CPU / GPU Chiplets & RAM on interposer / package design
HBM3 Unified Memory Architecture - 128GB / 256GB / 512GB
NVMe RAID 0 (dual NAND blades) 4TB / 8TB / 16TB
Four USB4 (TB3) ports
Four TB4 ports
Two 10Gb Ethernet ports
One HDMI 2.1 port
One MPX-C slot (for use with GPGPU expansion card)

;^p

MacPro 2013 version 2?

 

[Bandaman]

I don’t think anyone would expect a new Mac Pro to be slower, but I don’t think they’re going to release a new Mac Pro right away.

Their usual timeframe of a decade should do the trick.

 

[ian87w]

I don’t think anyone would expect a new Mac Pro to be slower, but I don’t think they’re going to release a new Mac Pro right away.

The transition was said to be complete in 2 years. The new Mac pro would be released in two years as the last Mac being transitioned from Intel.

 

[Boil]

MacPro 2013 version 2?

Yes, but now it has USB4 / TB3 / TB4, and there is an established TB3 presence in devices (audio i/o, video i/o, storage , egpu, etc.). Thunderbolt 4 doubles the throughput, so A/V folks will have sufficient data rates for their needs...?

 

[burgerrecords]

The transition was said to be complete in 2 years. The new Mac pro would be released in two years as the last Mac being transitioned from Intel.

I still believe it’s possible that doesn’t mean a Mac Pro replacement in 2 years; it’s weird they released this halo piece for such a small window. Or maybe Mac Pro just stops being sold in two years and isn’t replaced for a period of time (it’s low enough volume and it’s not like Apple’s semantics are not run through PR filters)

Im as excited as anyone to see how thorough up and down the pure technical dominance is going to be (including their gpu) Usually there are trade offs somewhere? I’m glad Apple exist though.

 

[satchmo]

I’d wager the prototype already exists.

Yup. Many factors including performance, cost, control, supply helped Apple determine whether or not to stay with Intel.

I’m sure performance was high on that list and knowing they have something that’s comparable.

 

[deconstruct60]

I’d wager the prototype already exists.

Probably doesn't. If it already existed there would be no reason for the 2 year transition timeline.
Most likely Apple is going to put Mac Pro "work" in last. ( Just like last after doing iMac Pro work first in 2016-2018 timeframe ). And that would likely mean next year. ( to roll something out the year after).

Perhaps something faster than the 8-12 core models with standard configuration RAM capacity. But triple digit RAM installed and 28+ cores .... probably not now. Maybe better chance by end of this year. Something running early in 2021? yes.

Apple shifted to Apple Silicon in part to chase after the enclosures they wanted to chase the most. Thiner laptops. Desktop performance in a Mac mini case. Perhaps thinner iMacs. The Mac Pro enclosure doesn't have much constraints on it. The power supply feeding it is already limit of normal household plug. macOS can't handle more than 64 threads anyway ( more than 64 threads benefits iOS devices how???? It doesn't ... so probbaly not high up on the list. ) . The Mac Pro represent the place where there was the least problem with the x86-64 solutions. Put AMD on the table and not really much of problem at all in 2020-2022 time frame CPU wise.

So since the least 'pressed to flip ' CPU , then probably the last one Apple is going to work on. It is far more strategically critical for the Mac ecosystem they Apple gets the others ( laptops -> basic iMac) 100% correct than try to expand too quickly into the iMac Pro - Mac Pro zone and suffer a miss on one the other systems.

Pretty good chance system won't show up until second half of 2022 ( if not Q4 2022) if go through the broad spectrum 3rd party and integration validations the workstation-server class hardware typically goes through. Apple's track record over the last 10 years in the Mac Pro space says nothing about super duper speed delivery at all.

 

[deconstruct60]

Yup. Many factors including performance, cost, control, supply helped Apple determine whether or not to stay with Intel.

That doesn't mean Apple has finished the design of all the solutions they are switching over too. There are probably more than a few i's not dotted and t's not crossed off that Apple hasn't completed that Intel does for them across the complete Mac line up.

they have probably started on the design and iterating bugs down closer to tape out, but completed past tape out and mostly working silicon ? Probably not.

I’m sure performance was high on that list and knowing they have something that’s comparable.

Performance isn't high(est) on the list. The graphic from their presentation said so. Apple is looking for better performance per watt... Not flat out performance. They'll get good performance. But faster laptops is a bigger focus than the Mac Pro workstation space.

 

[Realityck]

Performance isn't high(est) on the list. The graphic from their presentation said so. Apple is looking for better performance per watt... Not flat out performance. They'll get good performance. But faster laptops is a bigger focus than the Mac Pro workstation space.

If people look at it in that light, lower power, long battery, good performance, with a built in GPU, that is likely the direction this is headed. I just don't understand the ARM pundits comparing a Mac Pro with multiple state of the art graphics cards being approached by just a SoC?

 

[burgerrecords]

If people look at it in that light, lower power, long battery, good performance, with a built in GPU, that is likely the direction this is headed. I just don't understand the ARM pundits comparing a Mac Pro with multiple state of the art graphics cards being approached by just a SoC?

Even though I didn’t write that I do agree with you. It’s possible they will dominate at computing up and down but it seems far from inevitable.

 

[Realityck]

^ Sorry had a quote attached, poor editing, fixed now.

 

[konqerror]

I'm going to guess that Apple will not make Mac Pro silicon. They will buy a server-level ARM chip from somebody like Fujitsu or Marvell/Cavium.

There is a huge difference in architecture between Xeon SP and the desktop chips in order to allow scalability. There's no ring bus, it's mesh. There's distributed L3 cache. There's a whole directory-based architecture to handle the cache. The memory controllers are different. Because of the mesh, the system agent got split into multiple pieces.

I think it is too much work for a niche product, making new pieces which are unique. High-performance ARM chips already exist, they just need to add the software pieces to make it work with macOS.

 

[NT1440]

I'm going to guess that Apple will not make Mac Pro silicon. They will buy a server-level ARM chip from somebody like Fujitsu or Marvell/Cavium.

There is a huge difference in architecture between Xeon SP and the desktop chips in order to allow scalability. There's no ring bus, it's mesh. There's distributed L3 cache. There's a whole directory-based architecture to handle the cache. The memory controllers are different. Because of the mesh, the system agent got split into multiple pieces.

I think it is too much work for a niche product, making new pieces which are unique. High-performance ARM chips already exist, they just need to add the software pieces to make it work with macOS.

I wouldn’t advise you to follow through on that bit. Their lead chip engineer made it clear they have chips in development all across the Mac family.

 

[jetjaguar]

Mac Pro Cube - starting at US$3,999.00

48 P cores / 4 E cores / 96 GPU cores - CPU / GPU Chiplets & RAM on interposer / package design
HBM3 Unified Memory Architecture - 128GB / 256GB / 512GB
NVMe RAID 0 (dual NAND blades) 4TB / 8TB / 16TB
Four USB4 (TB3) ports
Four TB4 ports
Two 10Gb Ethernet ports
One HDMI 2.1 port
One MPX-C slot (for use with GPGPU expansion card)

;^p

I doubt they are changing the case. They went overboard with the engineering and design of the case as well as creating the mpx modules that I think this case design will be around for awhile.

 

[konqerror]

I wouldn’t advise you to follow through on that bit. Their lead chip engineer made it clear they have chips in development all across the Mac family.

I'd carefully read between the lines of advertising.

They call Marvell or Fujitsu up and have them customize core counts and speeds, delete some RAS features not needed in a workstation (external cache coherent interface), and print an Apple logo on it. Now it's a chip that Apple has "developed".

They've done that with Intel multiple times.

 

[tranceking26]

I said 4 years, I think it could be sooner but after the first gens we'll all find out just how good they may be.

 

[Realityck]

I wouldn’t advise you to follow through on that bit. Their lead chip engineer made it clear they have chips in development all across the Mac family.

Oh you mean Johny Srouji portion. All the keynote transcript text is below:

Johny Srouji: (01:29:06)
Another opportunity for the team was the iPad. While iPhone chips could drive our mainstream iPads, we wanted to push the iPad even further. It began with the iPad’s retina display, which demanded a custom chip. So the team scaled our architecture and design the most optimized and the highest performance chip possible, for the iPad. Starting with A5X, we built a line of SOCs specifically designed for the iPad. We doubled the iPhone’s graphic performance through a larger GPU and a wider memory subsystem. This put the iPad in a class by itself. Compared to the very first iPad, the latest iPad Pro, delivers over 1000 times faster graphics performance in just 10 years. This is part of the reason why the iPod Pro is faster than the vast majority of PC laptops. And this foreshadows how well our architecture will scale into the Mac

Johny Srouji: (01:29:57)
Another place where we applied our focus was the watch. We scaled our SOC architecture to optimize performance for the device’s unique low power requirements. And we built a chip perfectly suited for Apple Watch. Our SOCs enable each of these products with unique features and industry leading performance per watt. And it makes each of them best in class. And we do this at an enormous scale. In fact, adding all of the processors across these three products, we’ve shipped over 2 billion in just 10 years. And we’ve designed and shipped billions of additional chips, that work together with our SOCs to enable our amazing products. And now we’re bringing all of that expertise and that same focus and discipline approach, to the Mac. The first thing this will do is give the Mac a whole new level of performance. Now, when we talk about performance, we have to talk about power, because all systems built today are constrained by power consumption, thermals, or both. Among today’s consumer systems, desktops deliver the highest performance, but consume the most power. Notebooks, trade off performance for lower power, making them portable.

Johny Srouji: (01:31:07)
As you can see, normally to get more performance, you will have to consume more power. When you take a closer look at this chart, you’ll realize you want to operate in the upper left corner. You want to deliver the highest performance at the lowest power consumption. And that’s exactly where we want to take the Mac. Building upon our years of experience, designing the world’s most energy efficient chips, our plan is to give the Mac a much higher level of performance, while at the same time, consuming less power. So, much better performance is a reason enough to transition the Mac to Apple SOCs. But that’s just part of the story. Our scalable architecture includes many custom technologies, that when integrated with our software, will bring even more innovation to the Mac. With our advanced power management, we will maximize performance and battery life better than ever before.

Johny Srouji: (01:31:59)
Our Secure Enclave will bring best in class security. And our high performance GPU is going to bring a whole new level of graphics performance to every Mac, making them even better for pro applications, and really great for games. And combined with our Neural Engines, our chips will make the Mac an amazing platform for machine learning. And we’re bringing many other custom technologies such as our video display and image processing engines, that will help make the Mac better than ever before. So, what does all of this mean for the Mac? First, we’re designing a family of SOCs specifically for the Mac product line. Second, just like we did with the iPhone, iPad, and watch, we’re going to bring great technologies to the Mac. This will give the Mac a unique set of features and incredible performance. And third, we’ll have a common architecture across all of our product lines, making it far easier for developers to write on optimized software for the entire app and ecosystem.

Johny Srouji: (01:32:55)
Ultimately, we know that bringing our SOCs to the Mac will allow us to build much better products. And the Mac will take another huge leap forward. Now, a key advantage we have is the tight integration of our Silicon with our software. To tell you more about how macOS will run apple SOCs, here’s my colleague, Craig.

At this time they attempted to compare the A12Z as more powerful than the vast majority of laptops (ignoring any GPU comparisons) only a demo of Shadow of the Tomb raider running in Rosetta 2 emulation at 1080P.

So the question is SoC's specially for the Mac product line performance needs to be answered. Does low power high performance really manifest itself then? Will the GPU be more capable with higher scores? Consider me a Apple pessimeist that literally has seen everything Apple has produced. Rather the say you can dance, show me you can dance and really good, because its a greats sales pitch but we all use these computers doing real world tasks. Thats all.

 

[NT1440]

I'd carefully read between the lines of advertising.

They call Marvell or Fujitsu up and have them customize core counts and speeds, delete some RAS features not needed in a workstation (external cache coherent interface), and print an Apple logo on it. Now it's a chip that Apple has "developed".

They've done that with Intel multiple times.

The *entire point* of Apple investing in their own silicon designs is that they control the entire stack top to bottom. They’re simply not going to outsource a higher end design.