Arm Predicts Stagnation If Nvidia Deal Fails | Hacker News Reader

Arm Predicts Stagnation If Nvidia Deal Fails

eetimes.com

45 pointsaluminussoma29 comments

Of course they predict that. They want the deal to go through so they can cash in on Nvidia's sweet, sweet buyout offer.

Yes, fire them and get a board and executives who can do better

I predict stagnation if I do not get a juicy salary increase. (Totally unbiased opinion on my part)

There are also many (very smart) people who predict stagnation if the Nvidia deal succeeds.

Given how many devices nowadays depend on arm IP i think they don't have cash issue. If so than maybe rise prices. Also i don't think RISC-V is a real competitor yet. Arm has Apple as its customer and old founding company. I wouldn't trust Apple with Arm but I don't think Apple would allow for "stagnation".

This is all prediciton and Apple invested a lot into switch to Arm which paid off so going back to x86 isn't an option. Especially because of the i86 instruction decoder situation. The fixed size instructions on Arm is a really good thing imho. And the much better thermals which are key in mobile devices. I know thermals are not an attribute of architecture but more of a super low EUV in TSMC process nodes but it adds up.

Genuine question: why would Apple be worse than Nvidia for ARM?

Aren't apply taking chip design in house? They're a customer now, but will they still be in 5 years?

I believe that you should have a license for ARM ISA to implement it and sell chips.

I also know that AMBA, with AXI and other popular bus designs from ARM are free to use without ARM CPU in the system and should be separately licensed when you use ARM ISA in your system. I can assume with high probability that most of peripherial devices like DRAM controller and others are using one of the AMBA buses.

Basically, if you use ARM CPU commercially, you are on hook to pay some not insignificant amount of money in perpetuity.

So, if Apple will continue to use ARM, they will be an ARM's customer, a significant one.

What I heard is that Apple has a perpetual license having founded the company in the first place.

Can someone confirm this?

What does Apple switching to ARM *instruction set* have to do with ARM chip design stagnating.

Apple doesn't use ARM designs and RISC instruction sets shouldn't technically "stagnate"

Wasn't Apple looking for some RISC V developers or hardware people recently? Perhaps Apple will add some small RISC V components to their devices now but maybe envision moving to RISC V in the future. Or perhaps it's because of worries around the Nvidia deal perhaps ...

they have a core design, instruction decoder is like no more than 1% of the silicon and they're running a risk of depending on nvidia, which really likes how it prints money now and really wants to continue doing that. also, they've changed the architecture three times now?

> In dismissing the IPO option, Arm said a stock offering would suffocate its ability to invest, expand and innovate, noting that capital markets demand a focus on short-term revenue growth and profitability.

Yea, sure would be bad to be part of a (checks notes) publicly traded company like Nvidia, subject to the whims of the capital markets.

translation: 'We won't get that much money anywhere else'

Bullshit. Ask Jeff Bezos how to do it.

Did he have majority interest?

> while noting that Arm faces stiff competition from emerging RISC-V competitors

Is the competition from RISC-V really that stiff?

Not now, but the ISA is so superficial for a modern superscaler out-of-order design I don't see why existing designs couldn't easily be "reskinned", say if Nvidia-ARM jacked up fees.

I think most of the effort is getting GCC, LLVM, Linux, etc. support, and some working design to put RISC-V on top of. In other words, the difficult is in the number of cats to be herded.

In a way, I hope the deal does go through just to accelerate the transition away from ARM.

All of the RISC-V software ecosystem exists, and you can buy high performance out-of-order RISC-V implementations now from SiFive, as well as low power implementations that are very competitive with low power ARM.

The only problem right now is that high performance in the RISC-V world is about 4-5 generations behind ARM, so the fastest SoCs you can buy have cores that are roughly equivalent to a Cortex A73. SiFive is catching up a little each generation but they are a way off yet.

Yeah I think that "4-5 generations" is not about RISC-V intrinsically needing to catch up, but about the current RISC-V manufacturers needing to catch up.

I see 0 reason an apple silicon or amazon graviton couldn't be reskinned in a few months time, loosing no speed.

In some areas, mainly microcontrollers especially with additional custom features. But not really in the high end at this point.

Yes, if it eats them from the low end up, it can potentially do what Intel did to minicomputers and what ARM is trying to do to Intel.

There is a low barrier to hardware design for microcontrollers, and there is almost no innovation in the ISA these days it is all in the digital logic and software around it, so charging for that is approaching rent-seeking. Not saying the ARM64 ISA is not nice and clean, but it's a warmed over RISC. There's nothing revolutionary about it and no new ideas. It just shouldn't be something that locks you in to a vendor, or requires you to pay a tax on every device. It's an "API", it should be open.

I suspect the Nvidia wants to get their hands on the Arm patent portfolio.

> “While Arm’s licensees such as Apple, Qualcomm and Amazon have enjoyed skyrocketing revenue growth and profits, as well as soaring market valuations, Arm has lately endured comparably flat revenues, rising costs and lower profits that would likely present challenges for a 30-year-old public company.

Why though?

> The filing also [notes] that Arm faces stiff competition from emerging RISC-V competitors.

It seems like only yesterday that Arm was telling the world that RISC-V was a non-starter: https://github.com/arm-facts/arm-basics.com/blob/master/inde...

Can someone explain to me why softbank is selling arm:

1. So quickly 2. For such small relative profit?

my guess, they need money because wework was a giant clusterf*#,k

softbank is so much more than wework and wework might actually start making sense now. softbank has a lot of china exposure, which got hit really bad in the past year.

ARM China has gone rogue: https://news.ycombinator.com/item?id=28329731

That was a big chunk of business to give away for free. Now you have a Western ARM, and an Eastern ARM competing with each other, in a market which Western ARM used to have all to its self. That can't be good for your valuation.

Reply on news.ycombinator.com