$pple is just a cynical company that is so greedy and selfish and never wants to give to others unless they get something in return. 200B in the bank and act like a troll.
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.
Apple Possibly Exploring Open-Source Alternative to Arm Architecture
- Thread starter MacRumors
- Start date
- Sort by reaction score
Do you by any chance know how ARM licensing tends to work? I’ve seen claims that Apple doesn’t pay any royalties to ARM at all, since those are only charged for hard IP. It’s possible that Apple simply paid 50-100 millions a couple of years ago for the use of the Aarch64 ISA and that was the end of it. They have already changed a lot of low-level aspects of the architecture anyway, for example their interrupt system and intra-CPU communication are non standard, abs of course you have the fully custom matrix coprocessor instructions as well as architectural state…
That could very well be, but then again, I was never on the Arm is better bandwagon either, backwards compatibility is too important to me and my job.
That's cool, I can understand that, and at least we share one concern, even though it isn't to the same degree. I'm hoping Apple will backtrack on that eventually and I can happily buy whatever Mac comes after that -- I'm too much of an OS geek to resist it for my home PCs and Apple's OS is definitely in the good category.
And I'm a software geek, not a hardware geek. I got through all my hardware courses in college by thinking like I was making software.
It might be kind of hard to get me to buy another iPhone though, I actually like the tech of my new Flip 3, and Android has come a very long way in usability.
When the Macintosh Quadra was introduced in late '91, the base level unit cost US$5,700 (equivalent to $10,830 in 2020). If you want to talk about prices 'going up'.
People are taking this totally wrong. It's probably for the baseband in the modem chip.
Agreed. . . many people here are too young to remember the "good old days"
I can’t comment. But consider - Apple provided the funding for Arm Ltd. You think they didn’t negotiate themselves a good deal?
Could be. Of course, they could just use any old apple watch CPU core they have lying around for that, too.
Apple will always have a plan B … but this definitely isn't plan A. It just doesn't provide enough benefits to justify the effort. The cost of transitioning platforms will be way more than what they could safe.
And it's foolish to assume that Apple didn't ensure they have all the rights to ARM they'll ever need before they made the switch.
And it's foolish to assume that Apple didn't ensure they have all the rights to ARM they'll ever need before they made the switch.
Last edited:
That is (X). The amount of upstream going into FreeBSD is miniscule.
Note: (x) = Insert anything you deem appreciate, because the word I have in my mind would be against the rules on this forum.
I mean, Macrumors could have at least posted the most relevant information.
This is interesting because without this piece of information I would have thought Apple is trying to save pennies in those embedded ARM cores ( not the the ones used in CPU ) such as U1, W1 etc. But this is being specific to VaNG. And RISC-V is certainly a great fit for this use case with their Vector extension instructions. ( If you want to understand this I recommend searching for SIMD and Vector ) It will be interesting to see how all these fits together.
This is interesting because without this piece of information I would have thought Apple is trying to save pennies in those embedded ARM cores ( not the the ones used in CPU ) such as U1, W1 etc. But this is being specific to VaNG. And RISC-V is certainly a great fit for this use case with their Vector extension instructions. ( If you want to understand this I recommend searching for SIMD and Vector ) It will be interesting to see how all these fits together.
RISC-V can achieve 5ghz at 1 Watt.
And Chinese government steal their technology for their own...ARM China recently went rogue and declared independence from the parent company/joint venture.
[Updated, Correction] Arm China Goes Rogue In Hostile Takeover And Starts-Up As An Independent Company
Allen Wu, who serves as the CEO of Arm China, has quite literally gone rogue.hothardware.com
It's about having options. Pretty sure Apple is fine making their own silicon. But, just like with PowerPC. They had a backup plan with intel. The roadmap was better on intel over the next 10 to 15 years. So they had to make the switch. I'm sure they would have been very happy to stay on PowerPC had IBM been better able to deliver CPU's more suited to Apples needs. But, IBM's future plans was more well, POWER then Apple could use in a portable or even a mobile device.
Workstations and servers could have used it and not many would have complained about how hot these things get. So long as they don't over heat and break. But, a portable not so much. Which is where we find ourselves with intel today. Too damn hot and not small enough for Apples future and current plans.
I don't think RISC-V is a manufacturing thing. More of just give me the instruction set so I can be compatible with it. And Apple would still make processors that are on TSMC's production lines. It would then be ISA compatible with RISC-V and most likely even ARM64 at least for a transition period. Again, only if they had to switch.
This is what RISC-V can do.
1. 5ghz at 1 watt.
2. More cores but less power consumption than ARM
3. Open Source.
4. Easy to increase cores. 1000~4000 cores already developed.
5. RISC-V can still be used with ARM chip such as controllers.
1. 5ghz at 1 watt.
2. More cores but less power consumption than ARM
3. Open Source.
4. Easy to increase cores. 1000~4000 cores already developed.
5. RISC-V can still be used with ARM chip such as controllers.
Pragmatically the first really isn't true because the second ( "low end devices") are often used in the first.
So if 'measly' , 'low end' Apple SSD controller inside of an Apple SoC is technically used in a 'high-performance' personal device. If they are entangled on the same die they are definitely 'used in' the same locations where the die is provisioned.
Apple is making Celluar modems. Is the generic compute engine in that going to run userland level apps? No. Might it run some low-level , optimized library fragment? Possibly. And Ethernet offload engine? same thing.
It already is viable. If it is viable as a coprocessor then it is viable. Take a look at the recent Hot Chips presentation by Esperanto Technologies
https://www.anandtech.com/show/1690...g-machine-learning-esperanto-enflame-qualcomm
They have 1000's of 64 bit cores implementation of RISC-V assisting in doing ML training workloads.
similarly EdgeQ is using RISC-V as the baseline of their signal processors.
Hot Chips 2021 Live Blog: New Tech (Infineon, EdgeQ, Samsung)
Note that Apple is looking for someone to do low level library calls. Like how Apple libraries solely call the Afterburner card , no userland level apps.
Intel has a large FPGA product line ( Altera ) that is using Arm at he moment for the management "brains" for their products. Intel has already lined up a custom to fab RISC-V for them. It would be pretty startforward for Intel to start to flip some of their internal Arm usage to non Arm side.
As for Apple a large chunk of the computational growth of Apple Slicon ( A-series and not both A and M series) has come through co-processor; not the application code generic CPU cores. Neural cores (not scheduled by the OS) , AMX is a effectively a co-processor that the Arm cores "supervise" but is not scheduled by the OS.
Viable is different than breadth of deployments and also different than userland software inertia.
Hauwei/HiSilocon had an ARM license until it got "blow up" in a trade-war. It isn't a likely case for Apple at the moment, but they are not completely immune. If go back 10 years ago bet Hauwei thought they were completely immune from that problem too.
Apple doesn't want to move. But Apple doesn't completely control the universe.
P.S. When Apple finally decommissions Intel x86_64 in more than several they'll need a "Plan B" alternative to do contingency builds on. Gong back to x86_64 wouldn't be an architecture license option.
You are right, I went a bit overboard with that bit. If I remember correctly, FreeBSD benefited from Apples LLVM work but it was a while ago that I followed those developments. Didnt mean any insult towards the FreeBSD community, sorry for making a silly statement!
I was commenting on the idea of Apple replacing their ARM-based products with RISC-V in the near future. As to use of RISC-V in various controllers, I fully agree.
Personally I am very happy for Apple's success. It's good for shareholders, the industry and society generally. It also regularly gives to charities -- no strings attached by definition. Those are the facts. Maybe it should just "give" money to Intel to support x86.