Report: ARM is for sale and Nvidia’s interested, Apple isn’t • Hacker News Board

Re: Report: ARM is for sale and Nvidia’s interested, Apple isn’t

#161

Post by Qwertious » Fri, Jul 24, 2020, 7:08 AM UTC

stefan_ wrote: post #125

saxonww wrote: post #61

Earlier quoted context omitted.

I think they also totally own the console space, don't they? The next-gen consoles from both Sony and Microsoft are based on custom AMD cpu/gpu combos.

The lowest bidder owns the console space. It's not a good business, it's a bit of cash flow.

The real benefit of owning the console space isn't money, it's forcing all the console developers to develop specifically for your hardware. If you want to eke out all the performance on Xbone or PS4, you must take full advantage of AMD's CPU/GPU architecture.

Re: Report: ARM is for sale and Nvidia’s interested, Apple isn’t

#162

Post by thu2111 » Fri, Jul 24, 2020, 7:15 AM UTC

jasonwatkinspdx wrote: post #143

klelatti wrote: post #136

Earlier quoted context omitted.

Isn't the point though that a lot of the enthusiasm around RISC-V is that it's 'free' and that's no longer the case if you're using a proprietary extension. Plus you're looking at much greater possible fragmentation than can happen with ARM?

The big difference is RISC-V has been designed to handle that situation from the start. That's why the ISA is modular.

RISC-V is modular in the same way Windows or x86 is. Yeah, you can break up opcodes into groups and say there's some kind of probe to find out what groups are present. If your app needs that module then you're out of luck - it needs a different chip. This is nothing special and I don't see why it'd change anything fundamentally, competition wise.

Re: Report: ARM is for sale and Nvidia’s interested, Apple isn’t

#163

Post by imtringued » Fri, Jul 24, 2020, 7:25 AM UTC

jacquesm wrote: post #127

ithkuil wrote: post #124

Earlier quoted context omitted.

Could you summarize the main architectural (not implementation) differences between arm and risc-v that apply to being a good/poor match for mobile in particular?

I actually don't think there are any at the lowest level, it's mostly a matter of ARM being able to reduce power consumption by aggressively throttling down the clock rates and various advanced sleep modes. In principle you should be able to do this with any CPU, given a roughly equal transistor count power consumption should be close. Obviously ARM is much further along in the practical implementation of these featu…

ARM has big little on every chip except embedded ones. The mix of super scalar and in order CPUs significantly improves power consumption during idling.

Re: Report: ARM is for sale and Nvidia’s interested, Apple isn’t

#164

Post by imtringued » Fri, Jul 24, 2020, 7:29 AM UTC

bsder wrote: post #115

jacquesm wrote: post #96

Earlier quoted context omitted.

It is work in progress on a very impressive trajectory.

It's effectively MIPS warmed over. And MIPS never went anywhere because it never cracked the cellphone market (which is where the real volume is). When someone puts out a RISC-V phone, you've got a battle on your hands. Until then, you're fighting over table scraps.

Isn't MIPS owned by a single company? How is that ever going to get traction if MIPS doesn't deliver exactly the types of chips that its customers want?

Re: Report: ARM is for sale and Nvidia’s interested, Apple isn’t

#165

Post by imtringued » Fri, Jul 24, 2020, 7:31 AM UTC

stefan_ wrote: post #125

saxonww wrote: post #61

Earlier quoted context omitted.

I think they also totally own the console space, don't they? The next-gen consoles from both Sony and Microsoft are based on custom AMD cpu/gpu combos.

The lowest bidder owns the console space. It's not a good business, it's a bit of cash flow.

The complaint was that AMD was insignificant, not that it's low margin which is a perfectly acceptable business strategy.

As dralley mentioned. AMD is actually one of the few businesses with both strong CPU and GPU expertise. Building a console with any other company would require buying the CPU and GPU from two different companies. Intel would refuse and ARM CPUs including the ones Nvidia designed are not powerful enough, so really AMD CPU+Nvidia GPU is the only alternative to AMD CPU+AMD GPU but there is absolutely no reason for AMD to collaborate with Nvidia. The truth is that AMD is the only one that can serve this market.

Re: Report: ARM is for sale and Nvidia’s interested, Apple isn’t

#166

Post by klelatti » Fri, Jul 24, 2020, 7:52 AM UTC

Qwertious wrote: post #160

klelatti wrote: post #129

Earlier quoted context omitted.

The reason why I'm supportive of the ARM business model is that the fees are really not that significant (eg compared to x86) and actually I think it's good to support a firm with a primary focus of getting the most performance per watt from a mobile chip. A few dollars in licensing fees is really a small price to pay.

I think there's "X should be free", and then there's "there should be a free implementation of X you have the option of using". The former is up for debate, but I doubt many people would disagree with the latter.

If 'should be a free implementation' means that it undermines the business model of the firm that I rely on to develop X then I'd disagree on the latter. ARM has a business model that I think broadly works for the industry and consumers. I'd rather not throw that away.

Re: Report: ARM is for sale and Nvidia’s interested, Apple isn’t

#167

Post by klelatti » Fri, Jul 24, 2020, 7:57 AM UTC

bfuclusion wrote: post #147

klelatti wrote: post #142

Earlier quoted context omitted.

But that's not an ARM issue is it? PI has strange boot by GPU but issue is SoC implementation I think not CPU architecture.

Right, it's not CPU arch, but there's a much higher chance of having the entire thing open arch if the CPU is. I expect the probability of reaching a de-facto "standard" SoC to be much higher in RISC-V and I'll take the pain of being the early adopter.

I have a lot of sympathy but fear that RISC-V will lead to even more fragmentation than we see with ARM - eg proprietary extensions. PC standardisation is largely due to the level of control that a couple of dominant vendors have not due to the status of the IP in the core.

Re: Report: ARM is for sale and Nvidia’s interested, Apple isn’t

#168

Post by imtringued » Fri, Jul 24, 2020, 7:59 AM UTC

BbzzbB wrote: post #152

starfallg wrote: post #126

Earlier quoted context omitted.

Softbank and Vision Fund aren't exactly looking for Nvidia shares though. They are looking for cold hard cash to plug the massive hole in the failing investments.

I think Son is looking for the next big thing, probably something about shared workspaces.

Didn't they already have such a business? Wasn't it called WorkForMe (Me = Adam Neumann)?

Re: Report: ARM is for sale and Nvidia’s interested, Apple isn’t

#169

Post by klelatti » Fri, Jul 24, 2020, 8:01 AM UTC

pankajdoharey wrote: post #158

klelatti wrote: post #139

Earlier quoted context omitted.

If reports are correct they did hire some of the Imagination GPU design team. So they probably could but whilst the relationship is good the best of all worlds is a strong ARM team plus a strong team in-house.

Actually Apple benefits if the ARM Core team isn't very good, the rest of the industry would suffer if Mali GPU's dont get upgrades. Already they are behind iPhones, a poor team at ARM Ensures competition doesn't catchup. But all this might change should NVIDIA buys it and adds their own GPUS to the SOC.

Lots of existing GPU alternatives already - including Nvidia (Tegra) so can't see this as a game changer.

Re: Report: ARM is for sale and Nvidia’s interested, Apple isn’t

#170

Post by imtringued » Fri, Jul 24, 2020, 8:05 AM UTC

solarkraft wrote: post #46

How can anyone not be interested in owning ARM? Whoever controls that company can decide who's allowed to make processors that run consumer facing software (sure, RISC V might present a threat in the long term , but it's a good decade out). Of course that would be a nightmare for every other licensee (which is why I'm hoping for the IPO), but the potential gain is immense, isn't it? Or are the licenses so nice they'r…

There is no value in owning ARM for its licensees other than engaging in non competitive practices against its other licensees. That means you need to have 32 billion dollars to burn with the expectation of never getting them back in a way that can be attributed to the ARM acquisition. The potential gain is pretty slim because the only way a licensee can extract value from the purchase is by destroying ARM which negates the entire purpose of an acquisition.