Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Nvidia to Acquire Arm for $40B

nvidianews.nvidia.com

Post reply on HN
Print view View original

331–340 of 784 posts

Re: Nvidia to Acquire Arm for $40B

#331

Post by KMag »

dragontamer wrote: post #305
darksaints wrote: post #302

Earlier quoted context omitted.

Traditional compiler techniques may have struggled with maintaining code for different architectures, but a lot has changed in the last 15 years. The rise of widely used IR languages has led to compilers that support dozens of architectures and hundreds of instruction sets. And they are getting better all the time. The compiler has nearly all of the information that the CPU has, and it has orders of magnitude more. A…

> At best, your CPU can think a couple dozen cycles ahead of what it is currently executing. The 200-sized reorder buffer says otherwise. Loads/stores can be reordered for 200+ different concurrent objects on modern Intel skylake (2015 through 2020) CPUs. And its about to get a bump to 300+ sized reorder buffers in Icelake. Modern CPUs are designed to "think ahead" almost the entirety of DDR4 RAM Latency, allowing re…

With proper profiling (say, reservoir sampling of instructions causing pipeline stalls), and dynamic recompilation/reoptimization like IBM's project DAISY / HP's Dynamo, you may get performance near a modern out-of-order desktop processor at the power budget of a modern in-order low-power chip.

You get instructions scheduled based on actual dynamically measured usage patterns, but you don't pay for dedicated circuits to do it, and you don't re-do those calculations in hardware for every single instruction executed.

It's not a guaranteed win, but I think it's worth exploring.

Re: Nvidia to Acquire Arm for $40B

#332

Post by hu3 »

scarface74 wrote: post #234
hu3 wrote: post #170

Earlier quoted context omitted.

Even perpetual license to future IP?

Why would Apple need future ARM IP? They have more than enough in house talent to take their license in any direction they wish.

Such deal could certainly benefit both players. Hence my question.

Re: Nvidia to Acquire Arm for $40B

#333

Post by siberianbear »

pathseeker wrote: post #296
Blammar wrote: post #243

No one has seemed to notice the following two things: "To pave the way for the deal, SoftBank reversed an earlier decision to strip out an internet-of-things business from Arm and transfer it to a new company under its control. That would have stripped Arm of what was meant to be the high-growth engine that would power it into a 5G-connected future. One person said that SoftBank made the decision because it would hav…

>Also, to first order, when a company issues stock to purchase another corporation, that cost is essentially "free" since the value of the corporation increases. This isn't correct. If investors thing Nvidia overpaid, its share price will decline. There are many examples of acquiring companies losing significant value on announcements to buy other companies even in pure stock deals.

In addition, it's not even a valid argument if the cost was entirely in cash.

One would be making the argument, "the cost is essentially free because although we spent $40B, we acquired a company worth $40B". Obviously, that's not any more correct than the case of paying in stock.

Re: Nvidia to Acquire Arm for $40B

#334

Post by shaklee3 »

sillysaurusx wrote: post #330
shaklee3 wrote: post #323

Earlier quoted context omitted.

Let's just leave this one alone then. I can't argue with someone who claims anything is possible, yet absolutely nobody seems to be doing what you're referring to (except you). A100 now tops all MLPerf benchmarks, and the unavailable TPUv4 may not even keep up. Trust me, I would love if TPUs could do what you're saying, but they simply can't. There's no direct DMA from the NIC to where I can do a streaming applicatio…

FWIW, you can stream at 10Gbps to TPUs. (I've done it.) 10Gbps isn't quite 40Gbps, but I think you can get there by streaming to a few different TPUs on different VPC networks. Or to the same TPU from different VMs, possibly. The point is that there's a realistic alternative to nVidia's monopoly.

When I can run a TPU in my own data center, there is. Until then it precludes a lot of applications.

Re: Nvidia to Acquire Arm for $40B

#335

Post by leptons »

paulmd wrote: post #145
RubberShoes wrote: post #3

This is not good

It wouldn't have been good if any of the people who could actually afford to buy ARM did so. Would you rather have TSMC in control of ARM? Maybe have access to new architectures bundled with a mandate that you have to build them on TSMC's processes? How about Samsung? All of the fab ownership concerns of TSMC plus they also make basically any tech product you care to name, so all the integration concerns of NVIDIA. h…

Microsoft can't afford to buy ARM? lol... well that's not true at all.

Re: Nvidia to Acquire Arm for $40B

#336

Post by ttflee »

incognition wrote: post #293
yangcheng wrote: post #289

I am surprised that no one has mentioned China will very likely block this deal.

Are you thinking retribution for Huawei?

China de-facto blocked Qualcomm-NXP merger during a trade talk by not providing a decision before the deadline to the deal.

Re: Nvidia to Acquire Arm for $40B

#337

Post by patfla »

How does this get past the FTC? Oh right, it's been a dead letter since the Reagan administration. Monopoly 'R Us.

Never mind the FTC - the rest of the semiconductor industry has to be [very] strongly opposed.

Re: Nvidia to Acquire Arm for $40B

#339

Post by dragontamer »

KMag wrote: post #331
dragontamer wrote: post #305

Earlier quoted context omitted.

> At best, your CPU can think a couple dozen cycles ahead of what it is currently executing. The 200-sized reorder buffer says otherwise. Loads/stores can be reordered for 200+ different concurrent objects on modern Intel skylake (2015 through 2020) CPUs. And its about to get a bump to 300+ sized reorder buffers in Icelake. Modern CPUs are designed to "think ahead" almost the entirety of DDR4 RAM Latency, allowing re…

With proper profiling (say, reservoir sampling of instructions causing pipeline stalls), and dynamic recompilation/reoptimization like IBM's project DAISY / HP's Dynamo, you may get performance near a modern out-of-order desktop processor at the power budget of a modern in-order low-power chip. You get instructions scheduled based on actual dynamically measured usage patterns, but you don't pay for dedicated circuits…

But once you do that, then you hardware optimize the interpreter, and then its no longer called a "dynamic recompiler", but instead a "frontend to the microcode". :-)

Re: Nvidia to Acquire Arm for $40B

#340

Post by dannyw »

mr_toad wrote: post #297
paulmd wrote: post #116

Earlier quoted context omitted.

This is NVIDIA's "xbox one/PS4" moment. AMD has deals with console manufacturers, their clients pay for a huge amount of R&D that gets ported back into AMD's desktop graphics architecture. Even if AMD doesn't make basically anything on the consoles themselves, it's a huge win for their R&D. Now, every reference-implementation ARM processor manufactured will fund GeForce desktop products, datacenter/enterprise, etc as…

If I was an nVidia shareholder I wouldn’t want ARM profits to subsidise GPU development. GPU profits should be able to cover their own R&D, especially given the obscene prices on the high end cards.

Why not? The winning strategy since the 2000s have been to reinvest all profits into growing the business; not trying to chase juicy margins.
Post reply on HN