Concerns About an Apple Silicon Mac Pro

[Gamersoftdyzzie]

Having some issue after owning my Macbook Pro M1 for 4 months now and using it daily every day.
My purchase was an Apple Macbook M1 w/16GB ram. So far, I I am actually impressed with the performance of the Macbook in gaming, which i will say is really nice, but all 3 of the games that I play from the Mac version of steam crash often. I installed parallels w/Windows 10 pro so I could continue to use a few of the windows applications I can't get on mac and shockingly, playing those 3 games within the parallels windows VM don't have issues with crashing.

 

[robco74]

If ARM needs to raise licensing fees to be more profitable, most of the major players would likely pony up the cash. Nvidia can make their own designs, just like Apple does, they don't need to own ARM to be competitive. In fact, Apple never even considered buying ARM because they know the regulatory hurdles would be too big to overcome. Nvidia should have seen this coming.

If Nvidia were trustworthy, they wouldn't be attracting so much scrutiny, but they aren't trustworthy, so they are. They have clearly demonstrated that they would likely use the MS tactic of embrace, extend, extinguish.

In that vein, Qualcomm also engages in some questionable practices, mostly around patent licensing. They played a game of chicken with Apple, then blinked and they came to an agreement before it went to trial.

 

[iPadified]

Joint ownership of ARM by many companies would be an interesting model compared to vendor lock-ins like x86, CUDA. We have tried the latter and frankly it does not work very well at least not for keeping up a thriving market. Something so important as IT should not be dictated by a few companies. It should be seen as an infrastructure like water supply, roads, railways etc. Imagine how catastrophic that would work is there was a vendor lock-in... Reminds me of the different distances between rails before standardisation. "Sorry this train has too narrow of too wide length between the wheels so you have to move to another train".

PS. There is no Apple Tax. A tax usually has some hidden benefits such as roads, police etc. "Apple tax"=profit optimisation and thus only fills the pockets of the very few. DS

 

[Boil]

Why do they need to complete their transition by the end of 2022?

Because Tim Cook said they were undergoing a two-year transition at WWDC 2020, and the first M1-powered computers released in November 2020; so the end of 2022 would be just over two years from the first M1 release...

 

[mr_roboto]

Softbank will just go to a different buyer or ARM will have to do an IPO to raise cash itself. They want to sell because they screwed up in other investments (namely WeWork) and need to sell ARM to make up the difference. ARM doesn't get shuttered if this deal fails.

If ARM needs to raise licensing fees to be more profitable, most of the major players would likely pony up the cash.

Like @crazy dave said, this is all happening because Softbank's Vision Fund made a lot of mistakes. It's run by a guy who got high on his own supply after guessing right on Alibaba, and was then given control of enormous amounts of money to invest in things, but has only managed to prove that he's a sucker for tech startup grifters.

WeWork was the biggest disaster, but SoftBank also poured tons of money into Uber when Uber was at its most irrational highs. And there are many, many more.

Arm Holdings is up for sale not because Arm needs to raise licensing fees or whatever, but because there's real value there, unlike so many of Softbank's other bad investments. I say Arm was a bad investment only because Softbank massively overpaid for no particular reason, so they're probably going to take a big hit on selling Arm on to someone else. But it's one of the few assets they actually can get something out of, so in Softbank's current circumstances it has to go, and they're doing their best to raise the selling price by trying to involve companies like NVidia and Apple.

 

[crazy dave]

Nvidia should have seen this coming.

Oh they absolutely did. They prepaid the breakup $1.25B fee to SoftBank and paid $750 million upfront to ARM itself for IP licenses, possibly perpetual licenses similar to Apple that again they get to keep even if the deal doesn’t goes through. And if the deal does go through they effectively paid the $750 million to themselves.

So they still walk away with a lot even this fails. And Softbank may have insisted on the money upfront too. Pretty shrewd on both sides. Despite Nvidia's claims that this is all so shocking and must be politics or Qualcomm corrupting the regulators from 5 different countries (none of whom particularly like Qualcomm), truthfully neither Nvidia nor Softbank were caught unawares that regulators were exceedingly likely to object to this sale.

The first thing I thought when I heard this was "wow! is that really going to fly past regulators?" The only thing surprising to me was that it was the US FTC to object first (I was thinking it would be the UK and/or EU to challenge first) ... and that even the conservative free-marketers in the FTC (who were previously up in arms over the more liberal faction voting for new rules to go after mergers more aggressively in general) didn't vote against suing to stop this. In other words, even in an otherwise divided organization, the whole FTC basically went: "uhhhh ... no!, strike that ****, no" (or at least didn't object to others saying it, there a couple of abstentions, but no nays - that's unusual, especially in the current context)

https://twitter.com/i/web/status/1310574517937344514

Arm Holdings is up for sale not because Arm needs to raise licensing fees or whatever, but because there's real value there, unlike so many of Softbank's other bad investments. I say Arm was a bad investment only because Softbank massively overpaid for no particular reason, so they're probably going to take a big hit on selling Arm on to someone else. But it's one of the few assets they actually can get something out of, so in Softbank's current circumstances it has to go, and they're doing their best to raise the selling price by trying to involve companies like NVidia and Apple.

To be fair ARM’s yearly profits have gone down recently (still profitable but lower than they were) probably to support the development costs with the new ISA and higher power chips. That’s why I don’t fundamentally disagree than in the short term ARM getting deep pockets behind it while it grows would be a bad thing. Unfortunately SoftBank screwed up and here we are.

I am! I knew there'd be talk about things that I never considered here; that's why I chose to voice my thoughts in the first place.

Thanks for being understanding (I try to be polite when hijacking someone else's thread. I still do it ... but I try to be polite about it!)

 

[senttoschool]

Joint ownership of ARM by many companies would be an interesting model compared to vendor lock-ins like x86, CUDA. We have tried the latter and frankly it does not work very well at least not for keeping up a thriving market. Something so important as IT should not be dictated by a few companies.

Joint ownership of ARM is frankly stupid, sorry. It's been suggested a few times in this thread already, including @crazy dave

Why is it stupid?

Let's suppose that most current ARM license customers put up $40 billion to buy ARM together. This set of customers includes Apple, Intel, AMD, Samsung, Qualcomm, Mediatek, Nvidia, Ampere, Amazon, Microsoft.

• Apple does not want stock ARM designs to be good because that would mean faster Android devices
• Qualcomm and Mediatek wants stock ARM designs to be good because it would mean more competitive with iPhones
• AMD/Intel does not want ARM designs to be good because it would compete with their laptop and server CPUs
• Amazon/Microsoft would want ARM to prioritize server designs over mobile designs
• Qualcomm and Mediatek would want to prioritize mobile designs over server/laptop/desktop designs
• Nvidia would want ARM to prioritize high bandwidth CPU designs instead of general-purpose CPUs

Almost every customer would have different goals for ARM. The conflict of interest would make this a silly exercise.

 

[crazy dave]

Joint ownership of ARM is frankly stupid, sorry. It's been suggested a few times in this thread already, including @crazy dave

Why is it stupid?

Let's suppose that most current ARM license customers put up $40 billion to buy ARM together. This set of customers includes Apple, Intel, AMD, Samsung, Qualcomm, Mediatek, Nvidia, Ampere, Amazon, Microsoft.

• Apple does not want stock ARM designs to be good because that would mean faster Android devices
• Qualcomm and Mediatek wants stock ARM designs to be good because it would mean more competitive with iPhones
• AMD/Intel does not want ARM designs to be good because it would compete with their laptop and server CPUs
• Amazon/Microsoft would want ARM to prioritize server designs over mobile designs
• Qualcomm and Mediatek would want to prioritize mobile designs over server/laptop/desktop designs
• Nvidia would want ARM to prioritize high bandwidth CPU designs instead of general-purpose CPUs

Almost every customer would have different goals for ARM. The conflict of interest would make this a silly exercise.

You also just backed into one of the reasons why Nvidia owning them is also bad : Nvidia's interests would conflict with the ecosystem's.

BTW I didn't say that a consortium was the best solution. I said it's the best solution which involves ARM customers owning ARM (and as an aside Apple actually did own a substantial amount of ARM stock until recently. Probably a holdover from when they helped found ARM as a joint venture with Acorn ... which if Nvidia really wanted ARM IP and employees to help with CPU and AI and automotive, they could also do with ARM without incurring any regulatory wrath).

 

[Xiao_Xi]

Joint ownership of ARM is frankly stupid

Almost every customer would have different goals for ARM.

Every customer wants a different product from ARM. Investing in ARM would let them make sure ARM would focus on what they want.

 

[iPadified]

Joint ownership of ARM is frankly stupid, sorry. It's been suggested a few times in this thread already, including @crazy dave

Why is it stupid?

Let's suppose that most current ARM license customers put up $40 billion to buy ARM together. This set of customers includes Apple, Intel, AMD, Samsung, Qualcomm, Mediatek, Nvidia, Ampere, Amazon, Microsoft.

• Apple does not want stock ARM designs to be good because that would mean faster Android devices
• Qualcomm and Mediatek wants stock ARM designs to be good because it would mean more competitive with iPhones
• AMD/Intel does not want ARM designs to be good because it would compete with their laptop and server CPUs
• Amazon/Microsoft would want ARM to prioritize server designs over mobile designs
• Qualcomm and Mediatek would want to prioritize mobile designs over server/laptop/desktop designs
• Nvidia would want ARM to prioritize high bandwidth CPU designs instead of general-purpose CPUs

Almost every customer would have different goals for ARM. The conflict of interest would make this a silly exercise.

Is different goals incompatible with joint ownership? The opposite I would say. Take what you need from the IP pool and pay accordingly to the consortium. NIVDIA=CUDA=Vendor Lock-in and therefore unsuitable for sole ownership as is everybody else in you list.

Yes it is stupid for the sole owner because then the owner cannot misuse the situation.

 

[Bodhitree]

Well Apple does have a stake in this. They use a lot of ARM technology in the ISA and the compiler stack and the software around it, and if an NVidia purchase were to alter the direction of the technology then that might not be a good thing.

 

[Bug-Creator]

Well Apple does have a stake in this.

Only to a certain point, if ARM want to go one way and Apple the other we will just end up with 2 different implementations that aren't 100% compatible.

Result would be similar to 68k vs ColdFire (both made by the same company...) were most code would still run unaltered (with some performance hits) and compilers would only differ so little that both backends would still ship as one.

 

[diamond.g]

Well Apple does have a stake in this. They use a lot of ARM technology in the ISA and the compiler stack and the software around it, and if an NVidia purchase were to alter the direction of the technology then that might not be a good thing.

They have a perpetual forever license and can do whatever they want with ARMs ISA, right? So why would any changes by Nvidia matter to Apple?

 

[diamond.g]

Every customer wants a different product from ARM. Investing in ARM would let them make sure ARM would focus on what they want.

Or it could turn out like Vulkan/OpenGL/OpenCL....

 

[tdar]

I fully believe that the AS Mac Pro is going to be the most exciting computing thing this entire year. But it’s also going to be controversial. Because it’s going to be unlike what many people expect. First, I really believe that it’s not going to have expansion slots. Or places for external drives. Or discrete GPUs. It’s going to be a very fast, Mac mini with AS. Sure I expect lots of Thunderbolt four ports. But it’s going to redefine what a tower PC is.

 

[JMacHack]

I fully believe that the AS Mac Pro is going to be the most exciting computing thing this entire year. But it’s also going to be controversial. Because it’s going to be unlike what many people expect. First, I really believe that it’s not going to have expansion slots. Or places for external drives. Or discrete GPUs. It’s going to be a very fast, Mac mini with AS. Sure I expect lots of Thunderbolt four ports. But it’s going to redefine what a tower PC is.

They tried to redefine what a tower pc was in 2013, it didn’t work.

 

[Joe The Dragon]

I fully believe that the AS Mac Pro is going to be the most exciting computing thing this entire year. But it’s also going to be controversial. Because it’s going to be unlike what many people expect. First, I really believe that it’s not going to have expansion slots. Or places for external drives. Or discrete GPUs. It’s going to be a very fast, Mac mini with AS. Sure I expect lots of Thunderbolt four ports. But it’s going to redefine what a tower PC is.

redefine as in the ash tray 2.0 now with extreme markup on storage and ram.

Need drives in an workstion or at the very least raid 1 with an local only way (no need to hook up an 2th system) to manage it.

Maybe have an SFP+ port, E-sata, ect.

And TOWER should not have lot's of TB ports.

 

[diamond.g]

redefine as in the ash tray 2.0 now with extreme markup on storage and ram.

Need drives in an workstion or at the very least raid 1 with an local only way (no need to hook up an 2th system) to manage it.

Maybe have an SFP+ port, E-sata, ect.

And TOWER should not have lot's of TB ports.

I don't see any reason for Apple to keep the Mac Pro as is with the move to their own Silicon.

 

[arvinsim]

I considered adding a 3rd paragraph about price when writing my post but chose not to because in the past everyone's insisted that the Mac Pro is for customers who don't care about cost. If cost did matter, though, I think it would be a tough sell for Apple if they stuck with their current pricing for the Mac Pro. Consider what senttoschool said bellow:



That's awesome… if Apple prices the system better than they do right now! If it's a $6000 system, I can imagine a 4080 owner just throwing a second 4080 into their system if they need the power or pocketing 35% of the price if they don't. I do think they're going to price it better, though. They won't have to pay the Xeon + ECC tax anymore, after all.

Going with the prices of the new Macbooks, I am not too confident they would charge less. I wouldn't even be surprised if they charge MORE

 

[senttoschool]

You also just backed into one of the reasons why Nvidia owning them is also bad : Nvidia's interests would conflict with the ecosystem's.

But there is no law that says you can't buy a company that would have a conflict of interest.

 

[robco74]

But there is no law that says you can't buy a company that would have a conflict of interest.

Um, there are literally laws in almost every jurisdiction that say this.

 

[senttoschool]

Um, there are literally laws in almost every jurisdiction that say this.

Maybe you can show me a US law that says Nvidia can't buy ARM?

 

[Unregistered 4U]

I speculate a Mac Pro Cube, with a single edge connector (PCIe Gen5 x16) to allow mating with an optional expansion chassis for those who need PCIe cards (excepting GPUs, of course)...

I’d be surprised if they have any non-Thunderbolt/USBC expansion options. The main thing they were used for in the current Mac Pro was graphics and Afterburner and both of those are no longer a thing.

 

[robco74]

Maybe you can show me a US law that says Nvidia can't buy ARM?

The acquisition could theoretically happen, but the various regulatory agencies have to agree, and there are usually restrictions. See AT&T's attempt to acquire T-Mobile, NASDAQ and the NYSE. A few airline mergers have also been blocked over the years, or at least there were several conditions that had to be met before the merger was allowed. If any or all of the regulatory agencies determine that Nvidia acquiring ARM would stifle competition or give Nvidia an unfair advantage, they can nix it. Or they can but some pretty onerous requirements on Nvidia.

 

[Boil]

Looking at 2021 ASi MBP pricing, upgrading from...

• M1 Pro SoC / 8-core CPU (6P/2E) / 14-core GPU / 16GB RAM

...to...

• M1 Max SoC / 10-core CPU (8P/2E) / 32-core GPU / 64GB RAM

...is a $1500 upcharge...

Now, the initial SoC & RAM have a cost, so lowball it at $2k for a full-die M1 Max SoC & 64GB RAM...

This would put a dual SoC ASi Mac Pro (Cube) at $4k and a quad SoC model at $8k, just for the SoCs & RAM; mobo, chassis, PSU, active cooling, storage, etc., all extra on top...

So a dual could run you $6k, like the current 2019 Mac Pro entry pricing; a quad will go for $10k...?

More for larger storage options, of course...!

If Apple goes with high density (64GB chips) LPDDR5X RAM at some point, add an addition $12k or more to max out a quad SoC model with 1TB of RAM and 2TB/s UMA bandwidth...?

But in light of an ASi Mac Pro (Cube) most likely starting at the same $6k of the current Intel model; please Apple, please...! Give us a M1 Max-powered Mac mini (Pro) with an internal PSU, proper active cooling, & plenty of ports...! ;^p