Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

Joe The Dragon

macrumors 65816
Jul 26, 2006
1,031
524
the pro market may really need storage choices other then apple.
Unless apple can offer things like raid 1 or even raid 5 with there apple storage cards.
And lower pricing as apple price is way to high next to other pci-e storage hardware.
also the option for slower and bigger sata based disks in the case.
 

robco74

macrumors 6502a
Nov 22, 2020
509
944
You can get third-party storage that bolts right into the Mac Pro. There's a RAID 5 option that fits into one of the MPX slots, and another expansion cage that will fit standard SATA drives. These aren't from Apple, though Apple sells them.

It depends on your use case though. A lot of storage is moving to the network or the cloud. It depends on the industry, security and/or compliance concerns, how large an organization, etc.
 
  • Like
Reactions: Shibamigo

Joe The Dragon

macrumors 65816
Jul 26, 2006
1,031
524
You can get third-party storage that bolts right into the Mac Pro. There's a RAID 5 option that fits into one of the MPX slots, and another expansion cage that will fit standard SATA drives. These aren't from Apple, though Apple sells them.

It depends on your use case though. A lot of storage is moving to the network or the cloud. It depends on the industry, security and/or compliance concerns, how large an organization, etc.
Taking about network in some cases then may need pci-e slots or maybe SFP+ or better on board. IF 10G-E does not work for you.

some work loads may need lots of local raw storage or local network (not cloud unless you have a really good ISP)
 

Xiao_Xi

macrumors 68000
Oct 27, 2021
1,628
1,101
who should buy it to "pick up the slack" as it were.
The CEO of Qualcomm suggested that ARM main customers would buy small stakes if the Nvidia deal fails.

Apple has a perpetual architecture ARM license and Apple doesn't even use any stock ARM designs
What does its perpetual architecture license entitle Apple to? Can Apple use any set of ARM ISAs?
 

jdb8167

macrumors 601
Nov 17, 2008
4,859
4,599
What does its perpetual architecture license entitle Apple to? Can Apple use any set of ARM ISAs?
I don’t think any of the details of Apple’s agreement with Arm is public other than that Apple has an architecture license. Speculation says that because of Apple’s history of co-founding Arm that they might have a unique license agreement but it is all guesswork as far as I know.
 
  • Like
Reactions: Shibamigo

diamond.g

macrumors G4
Mar 20, 2007
11,438
2,665
OBX
So assuming Apple uses a hardware GPU Scheduler, I wonder how they it will work with the "off die" cores and how much latency that will introduce in the process.
 

Boil

macrumors 68040
Oct 23, 2018
3,478
3,173
Stargate Command
With the M1 Pro/Max Mac mini (Pro) & the M1 Pro/Max 27" iMac (Pro) rumored for this Spring, and the dual (possible quad) M1 Max ASi Mac Pro (Cube) rumored to either debut or actually launch at WWDC this year, Apple will have completed their 32 year transition to "Arm-based" Apple silicon...! ;^p
 

Brandon42

macrumors regular
Jul 24, 2019
207
588
I just hope that Apple reduces their margins on non-user upgradeable components like RAM. Their high margin is a relic of the days of user upgrades. Bureaucratic entities would make you order your computer from the supplier (Apple, Dell, etc) fully configured. You would not convince a purchasing manager to order a base desktop and then order some RAM from Crucial or Amazon. Price sensitive individuals and companies could always order the base model and then upgrade. Apple never expected those price sensitive people to pay the insane markups.
 
  • Like
Reactions: mi7chy

robco74

macrumors 6502a
Nov 22, 2020
509
944
Apple doesn't charge much more than other OEMs. Upgrades and add-ons are a profit well that many industries draw from. You also have to consider the time it takes for someone to receive the laptop and RAM and do the upgrade. Then deal with separate entities if something goes wrong. Upgrades from the vendor are usually covered by the support agreement, so if something goes awry, just send it back for repair or replacement.
 

mi7chy

macrumors G4
Oct 24, 2014
10,625
11,296
Why wouldn't better efficiency be desirable in a desktop machine to? Especially companies which run dozens or hundreds of Mac Pros surely wouldn't say no to powerful but yet efficient machines. Better efficiency = less power consumption = more money saved.

Time is money plus deadlines so efficiency is lower priority in a desktop. In reality, it's combination software availability/compatibility + performance + cost + efficiency otherwise more companies would use Raspberry Pi 4.
 

jdb8167

macrumors 601
Nov 17, 2008
4,859
4,599
So assuming Apple uses a hardware GPU Scheduler, I wonder how they it will work with the "off die" cores and how much latency that will introduce in the process.
Intel claims that they have something like that coming. Intel Deep Link Technology. If Apple has discrete GPUs coming for the Mac Pro they are probably planning something like this too.
 

leman

macrumors Core
Oct 14, 2008
19,521
19,678
So assuming Apple uses a hardware GPU Scheduler, I wonder how they it will work with the "off die" cores and how much latency that will introduce in the process.

If by "off die" you mean their rumoured multi-chip-technology, the latency is likely to be extremely low — the entire thing will work just like one big chip. I mean, you already have hierarchical arrangements within M1 chips themselves (e.g. 4 CPU cores per CPU P-cluster, separate GPU cores etc.). Maybe connecting multiple chips together will add more hops, maybe it won't — at any rate it's not something that will affect performance.

If by "off die" you mean a separate GPU (as in dGPU)... well, you know my thoughts on the matter :)

I just hope that Apple reduces their margins on non-user upgradeable components like RAM. Their high margin is a relic of the days of user upgrades. Bureaucratic entities would make you order your computer from the supplier (Apple, Dell, etc) fully configured. You would not convince a purchasing manager to order a base desktop and then order some RAM from Crucial or Amazon. Price sensitive individuals and companies could always order the base model and then upgrade. Apple never expected those price sensitive people to pay the insane markups.

Oh, upgrade pricing is one of the most successful elements of Apple's business. And they are 100% deliberate. They are choosing their upgrade tiers and pricing with great precision, exactly to lull in as many worried customers as possible.
 
  • Like
Reactions: JMacHack

diamond.g

macrumors G4
Mar 20, 2007
11,438
2,665
OBX
Intel claims that they have something like that coming. Intel Deep Link Technology. If Apple has discrete GPUs coming for the Mac Pro they are probably planning something like this too.
So it would be software based, which isn't bad. I just wonder how much overhead this will introduce. I also notice Intel isn't claiming their solution is for everything (they shy away from claiming it improves rasterization performance). It looks like it improves things Apple already have dedicated hardware for (well expect DP4a, maybe).
 

leman

macrumors Core
Oct 14, 2008
19,521
19,678
Intel claims that they have something like that coming. Intel Deep Link Technology. If Apple has discrete GPUs coming for the Mac Pro they are probably planning something like this too.

I just had a quick look... and it appears that "Deep Link" simply refers to the ability to use the iGPU and the dGPU simultaneously (e.g. you can run some sort of simulation on one and do graphics work on the other, or run compute kernels on both). So basically another marketing gimmick. You could do this kind of stuff on macOS since, well, forever.
 

diamond.g

macrumors G4
Mar 20, 2007
11,438
2,665
OBX
If by "off die" you mean their rumoured multi-chip-technology, the latency is likely to be extremely low — the entire thing will work just like one big chip. I mean, you already have hierarchical arrangements within M1 chips themselves (e.g. 4 CPU cores per CPU P-cluster, separate GPU cores etc.). Maybe connecting multiple chips together will add more hops, maybe it won't — at any rate it's not something that will affect performance.

If by "off die" you mean a separate GPU (as in dGPU)... well, you know my thoughts on the matter :)



Oh, upgrade pricing is one of the most successful elements of Apple's business. And they are 100% deliberate. They are choosing their upgrade tiers and pricing with great precision, exactly to lull in as many worried customers as possible.
I mean "off die" as in now you have 4 hardware schedulers for GPU workload that you have to talk to, how are you making them look like 1 hardware scheduler, or are you not bothering with that and just using software to "stitch" them together.

Or I could be completely wrong and Apple doesn't use hardware scheduling for their GPU's and runs it all via the CPU, in which case they should be able to scale and hide the "other GPU's" behind one GPU pretty easily. But then it also makes you wonder why they never bothered doing so the 11 years they have had graphics switching in their lineup.
 

jdb8167

macrumors 601
Nov 17, 2008
4,859
4,599
I just had a quick look... and it appears that "Deep Link" simply refers to the ability to use the iGPU and the dGPU simultaneously (e.g. you can run some sort of simulation on one and do graphics work on the other, or run compute kernels on both). So basically another marketing gimmick. You could do this kind of stuff on macOS since, well, forever.
I couldn't get enough information from their videos to draw any conclusions. But it looks like they are scheduling the iGPU and dGPU based on performance needs. It isn't clear if the scheduling is automatic or something programmed in by the application.

Edit: This is the company that claimed that your internet would be better with a new Intel CPU.
 

crazy dave

macrumors 65816
Sep 9, 2010
1,453
1,229
You’re not getting it.

Nvidia thinks ARM can do way better. Despite being such an important architecture, its shareholder value has not increased proportionally. ARM has barely made a dent in the server and supercomputer market for example despite many years of promise. AMD and Intel is still kicking ARM's ass there. For laptops, only Apple has successfully used ARM.

Nvidia thinks it can do better with ARM.

Why restrict the sale of ARM, which hurts the shareholders and employees of ARM, in order to increase the shareholder value of Qualcomm, for example?


Where is it in the law that ARM can only be bought via a consortium? And will this consortium pay as much as Nvidia is willing to pay?

Oh I am getting it. I understand the upside ... though the reason ARM designs haven't competed in server or PC or supercomputer until recently is that ARM only recently started designing their own cores with anything other than mobile in mind and partners who did custom cores for other sectors failed (until Apple and Amazon), often for reasons non-technical (see Centriq and Qualcomm). Qualcomm is getting back into the laptop business with their own custom cores in 2023 and we'll see if those are failures. They didn't need to buy ARM. Others are doing so using ARM's designs which are increasingly becoming competitive (X cores, N cores, and V cores, ARM v9) ... but yes this has also meant less profit in recent years so I get why an investor with deep pockets and an aggressive outlook would be beneficial.

However, Nvidia thinking it can do better with ARM isn't the point that regulators are having a problem with! Their concerns with the sale is what happens to the ARM ecosystem of players which is 500+ companies. This isn't about one ARM customer or helping it. Regulators have to look at the larger picture, deciding on the health of the market, not even just CPUs but automotive and AI as well. And Nvidia could substantially harm the larger ARM ecosystem by effectively locking out smaller players or at least gaining intrinsic advantages by gaining access to designs first. This is the most obvious problem but in some even worse then there's the trust issue. ARM holds a lot of customer data information. Thus other companies have to trust ARM for the system to work - it has to be a Switzerland. If it isn't anymore, even if Nvidia doesn't do anything wrong, the perception that they could is still very damaging. A good example is the current global supply chain crises. It relies on a lot of JIT shipments and payments which only works if companies trust each other to be able to do what they say they are going to when they say they are going to. In the current environment, nobody does and the system starts to screech to a halt as companies pull back which just reinforces the lack of trust which causes companies to pull back. Trust is a valuable commodity.


Where is it in the law that ARM can only be bought via a consortium? And will this consortium pay as much as Nvidia is willing to pay?

Many businesses that get acquired have a conflict of interest. It doesn't mean governments should stop them. Free market.

Then you don't support a free market then. You're supporting a conditional free market. In reality, if Nvidia wants to destroy ARM by refusing to give out future licenses, then other companies will respond by adopting something else. Nvidia will just screw ARM, which they will have spent $40 billion to acquire.


It isn't. But Nvidia acquiring ARM will have no impact on a company like Apple. If Nvidia signals that it wants to close ARM to outsiders, customers will have plenty of time to adopt something else.


And ARM/Nvidia are affected by not having this deal go through. You seem to be unfairly favoring one side. In my opinion, consumers will benefit from this deal. I think Nvidia will help accelerate ARM's adoption in laptops, desktops, servers, and supercomputers, giving everyone another option besides AMD, Intel, and Apple. In addition, I think Nvidia IP, money, and leadership will make stock phone ARM designs more competitive against Apple.

Here's the dirty little secret about our economy: pure unrestricted free markets aren't free. I don't mean to be insulting, but that they are is a naive viewpoint. Even the "father of capitalism" Adam Smith recognized that (some of what's attributed to him ain't him, a lot of his writings on the pitfalls of government intervention have been twisted, and he also wrote about the dangers and pitfalls of capitalism ... somehow those aren't as widely promulgated as his other works!). :) That's why we have government regulators - the main issue with which haven't been active enough and allowed too much consolidation in too many fields resulting in horrible inefficiencies and too big to fail companies (banking and the defense industry come immediately to mind). So yes, I support a conditional free market, because that's actually far more free. :)

Just taking this example, Nvidia driving off companies would have a massive ripple effect. It takes time and resources to build new CPU tech on a new ISA - everything from compilers to silicon design. So saying it would have no impact and those companies would have plenty of time is ... well ... rubbish.

It isn't two equal sides. As I wrote above, governments have to look at the whole picture. They don't always get it right or even when doing the right things, do them for the right reasons. No system is perfect - though I'd argue their imperfections have been in not being active enough.

I should also stress that so far only the US regulators have formally objected. The others have not yet. The UK will issue its report in May I think. While the tea leaves don't look good for the deal in those other regulators, they could decide that while they have concerns, they won't formally object and stop it. Nvidia could also take their suits to court and win. Governments lose regulatory battles all the time. The US just lost against Qualcomm. Personally I don't think they should've but they did. (If you’d like to know what laws the FTC is citing in their suit to block Nvidia’s ARM acquisition you can read the complaint and summaries/explanations online. I can try to find a good link if you’d like or you can search yourself).

TLDR: I don't disagree that deep pockets could help ARM maintain its current trajectory as it seeks to expand. I do object that Nvidia needs this deal to compete, they absolutely don't. I'm not sure why you're hung up on that, there are so many other ways for Nvidia to expand its CPU team and this deal is only partly about that in the first place. Market regulation is an important facet of a modern economy that properly applied keeps it far freer than a wild west.

If you want to continue this conversation, maybe DMs or a different thread? Tag me and others. This is pretty off topic for this thread and we shouldn't clog it up for those who want to discuss the OP's point.


It sounds like Softbank no longer wants to own ARM. So really the question becomes who should buy it to "pick up the slack" as it were. I mean the alternative is Softbank shutters ARM which seems worse, right?

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.
 
Last edited:
  • Like
Reactions: SBeardsl and ahurst

Boil

macrumors 68040
Oct 23, 2018
3,478
3,173
Stargate Command
With the ASi Mac Pro (Cube) coming along late in the whole M1 cycle, and rumors of the initial MCM only able to have two SoCs, what if...

Apple only releases a dual M1 Max SoC powered Mac Pro (Cube) this year, and WWDC 2023 sees the quad M2 Max-powered Mac Pro (Cube); with Apple giving a decent discount to those who purchased a dual M1 Max Mac Pro (Cube) in 2022...?!?
 

leman

macrumors Core
Oct 14, 2008
19,521
19,678
I mean "off die" as in now you have 4 hardware schedulers for GPU workload that you have to talk to, how are you making them look like 1 hardware scheduler, or are you not bothering with that and just using software to "stitch" them together.

Or I could be completely wrong and Apple doesn't use hardware scheduling for their GPU's and runs it all via the CPU, in which case they should be able to scale and hide the "other GPU's" behind one GPU pretty easily. But then it also makes you wonder why they never bothered doing so the 11 years they have had graphics switching in their lineup.

They definitely use hardware scheduling. As to how it might work, probably exactly how it works right now. They don't seem to have any troubles with handling scheduling for 8, 14, 16, 24 or 32 GPU cores, why would it be any different with cores spread between different chips? They have surely planned ahead so that their hardware can scale accordingly. Besides, as we don't have any idea how their GPU scheduling might work, it's difficult to speculate. Maybe they have a dedicated processor for handling these things or maybe the GPU cores do it by themselves by grabbing work from some sort of global work queue. Maybe Asahi people know more?
 

deconstruct60

macrumors G5
Mar 10, 2009
12,493
4,053
The CEO of Qualcomm suggested that ARM main customers would buy small stakes if the Nvidia deal fails.


They would need a mix of arch license, design license, and end users to do that buy out. If the top 3 Arch license holders buy up most of ARM that will probably get as much blow back as just Nvidia doing it would.

Nvidia and ARM have made some arguments that an IPO would be "bad". If let the bulk of the company get bought up by Wall St. "Greenmailers" then likely will cap the R&D spend under what ARM needs to be viable over a decade or so. ( "The Street" could give a frack. Pump and dump the company dry over 3-4 years and move onto the next one. )

What need is a majority company ownership that wants the company to right thing over the long term. A super narrow faction probably has skewed interest with respect to the more complete ecosystem.

(e.g., is likely interested in stuffing their GPU tech down folks throats to buy versus Mali or other options. The acquisition pricing pretty much leans that way. It is way, way ,way higher than what will get through the current revenue stream. )





What does its perpetual architecture license entitle Apple to? Can Apple use any set of ARM ISAs?

Highly likely there is not a single license that covers everything. That would grossly fiscally irresponsible of ARM to ask for a single fixed sum and then give away every IP invention the come up for the rest of the companies operation under that same agreement. How suppose to price something haven't invented yet?

Relatively safe to say that someone who gets an arch license doesn't get intermediate and distant future IP. Perhaps there is a X.0 and all the X.x updates (spread out in some fixed payment schedule). (e.g,. ARM 8 , 8.1 , 8.2 , 8.3 , 8.4 up until ARM goes to ARM 9. ) Likely not worth getting teams of lawyers looped in every time to a point ( x.1 , x.2 , x.3 ) update. Similarly, some folks may want to "drop out" at version v7 because it has everything that need for their narrow product range. That's the perpetual part. License holder can keep making the older stuff ( instruction and patent wise) for as long as you want.

Architecture license's major appeal in part is to get around paying on a per unit produced basis ( or pushing it down into the small fraction of $0.01 range. So SoC pricing round off error range. ). There in lies the rub for Arch licenses. Hand out too many to the biggest users and company may not be long term viable. ( if charge too little and license holders walk away with the bulk of the profits to keep the ecosystem moving forward. ) [ Also allows to pursue significantly different design tradeoffs than what ARM does with their more common standard reference designs. ]

The CPU is decoupled from the GPUs. The inter-core interconnect would be decoupled also.

If buy the more completed designs from ARM then have to pay royalties per unit in addition to the design cost. [ some ARM stuff they'll let you use to create initial R&D product and largely not start charging until shipping/using a product for revenue. bigger discounts up front and then take long term steady revenue stream. And if take a library and it doesn't work out for your product then don't have a huge sunk cost to recover; mainly time and resources. ]
 
Last edited:

deconstruct60

macrumors G5
Mar 10, 2009
12,493
4,053
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.

ECC tax? Have you looked at Apple Tax for memory. It is substantially higher than the open market increase for ECC ( using more memory bits should pay for them. )

Huge assumption there that going to get to buy "open market" RAM with the new "Mac Pro". Pretty good chance will get same kind of RAM that is attached to. M1 Max with any chiplet style scale up of a M1 Max.

As far as Xeon goes. The big hit for the W3200 class was the ' > 1TB RAM tax'. that doesn't exist on the W3300 series ( or on the AMD line up). At the buttom end of the Xeon line up, historically there was pricing overlap with the HEDT line up. [ Again. in 2019-2020 Intel got greedy groping for extra money to offset getting behind AMD and loosing share. Cover up of problems by planning to throw money at it. ] . The M-series "Mac Pro" probably won't get anywhere near 1TB of max RAM so never would have hit the ' > 1TB ' tax even if it was still around.

Can point to overall system wide cheaper but again like the M1 Max. Buying more CPU cores is highly likely going to come directly coupled to buy more GPU cores. The Price to upgrade the SoC on the BTO system at Apple isn't going to drop into the radically cheaper category. ( getting GPU additions too but Apple isn't going to make this "low cost". )
RAM also is likely going to coupled to core count. More cores more Apple priced RAM.


The shift to M-series hasn't lower CPU prices. The CPU upgrade costs in the BTO for the M-series systems offered so far haven't been cheaper than Intel update options. Apple slapped large mark up on Intel pricing. Same mark ups are still there. Apple just doesn't have to have some of that off the Intel. Apple didn't move to M-series to be become the low cost system pricing leader.

The lack of 3rd party GPU drivers on macOS on M-series means not going to get "open market" pricing on GPUs either. [ Unless Apple makes a 180 degree turn here in software support ... that won't happen. ]
 
  • Like
Reactions: whfsdude

Joniz

macrumors 6502a
Sep 21, 2017
676
1,646
Why even care?

Get what works for you; disregard the rest.

Others won’t find you more interesting because of the CPU you use.

Life is shorter than you believe.
 

senttoschool

macrumors 68030
Nov 2, 2017
2,626
5,482
Oh I am getting it. I understand the upside ... though the reason ARM designs haven't competed in server or PC or supercomputer until recently is that ARM only recently started designing their own cores with anything other than mobile in mind and partners who did custom cores for other sectors failed (until Apple and Amazon), often for reasons non-technical (see Centriq and Qualcomm)
I'll summarize my thoughts here. You don't need to reply.
  1. Nvidia needs its own world-class CPU team to compete with SoCs designed by Apple, Qualcomm, Mediatek, AMD, Intel, etc.
  2. Yes, Nvidia doesn't need to buy ARM to design ARM-based CPUs. But Nvidia thinks it can help accelerate ARM's adoption in devices other than low-powered devices, improve its designs by combining IPs, inject resources into the company. Remember that ARM doesn't actually make that much revenue or profit despite dominating mobile devices.
  3. A standalone ARM does not actually have enough money and resources to continuously compete with Apple, AMD, Intel designs. And it shows. Apple designs are far ahead of ARM designs on mobile. Intel and AMD designs are significantly more powerful in non-mobile devices (albeit at a lower efficiency). For context, ARM made around $2 billion in revenue in 2021. Intel $78 billion. AMD $15 billion. Apple $365 billion.
  4. ARM has been too slow to get into laptop, desktop, workstation, server, and supercomputer market. This is where I think Nvidia can help accelerate a lot and add a 3rd option to Intel/AMD duopoly. Only Apple has managed to successfully bring ARM to high power computers.
  5. Nvidia + ARM adds one more competitor to the laptop, desktop, server, supercomputer landscape. We've seen a duopoly by AMD/Intel for far too long which set innovation back a decade in the 2010s.
  6. Nvidia has been losing mega government contracts to AMD/Intel because they can't supply CPUs.
  7. Nvidia would like to control their entire hardware stack for server solutions. Right now, they have to use Epyc/Xeon.
  8. Even their upcoming Grace CPU simply uses stock Neoverse cores
  9. The future of mobile devices (maybe even desktops) is clearly a SoC/APU approach. This leaves Nvidia vulnerable because they don't make SoCs/APUs. This is a strategic acquisition for Nvidia.
  10. It's not fair to ARM employees and shareholders to block this deal. Blocking this deal benefits Qualcomm, AMD, Intel shareholders. You're just choosing one side over the other by blocking this deal.
  11. ARM has contracts with existing customers. Even if Nvidia buys them, they can't just deny existing customers with licenses.
  12. Nvidia is too smart to ruin ARM's business model. They're not going to spend $40b just to destroy all of ARM's revenue overnight.
  13. Qualcomm has clearly spent a lot of lobbying money to convince the US government to block this deal. It's actually weird that the US government would block an American company from acquiring important IP from a British/Japanese company. I'm guessing Qualcomm has bought off the right politicians.
  14. I think this deal benefits consumers, adds more competition. But unfortunately, there are too many myths and politics in this deal.
 

RedTheReader

macrumors 6502a
Original poster
Nov 18, 2019
532
1,312
Of course no one at apple ever thought of any of this, the Apple silicon transition has been a complete failure, if only they had listened to you they could be selling water-cooled scalder lake Intel laptops with 4 hours of battery life

P.S. @RedTheReader I hope you’re enjoying the conversation even if we’re a bit far off the OP topic ;)
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.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.