Hacker News Reader

Comment by Sakos | Hacker News Reader

It's a chicken or the egg dilemma. Non-Apple ARM is too slow, so nobody wants Windows ARM, and nobody wants Windows ARM, so nobody is seriously working on consumer-targeted ARM. It will probably fail because Microsoft is generally bad at managing and marketing their hardware projects and coordinating with manufacturers and retail partners. Also, yeah, the pricing and performance is abysmal. But there is some sense to trying to lay the groundwork for future ARM SoCs running Windows.

Right now, the biggest driver for Windows ARM adoption is, ironically, Apple's M1 and onwards because of people running it as a guest OS through Parallels.

Also, frankly, I'm okay with Microsoft failing at projects like this. We don't need another Apple-like presence on the market.

View on HN

The solution to chicken-or-egg problems is known: you spend money to just overpower the problem. But MS and Qualcomm aren't doing that; they're half-assing their ARM hardware to save a little money.

Yup.

Compare this to Apple's $499 ARM Dev test kit that you could return for a $200 credit.

If Microsoft is serious about ARM, they need a very low barrier of entry for those willing to port their software.

Apple's ace card is that they bought PA Semi and have a competent team to pull off what Qualcomm apparently can't.

Well, the Nuvia acquisition seems to indicate that at least Qualcomm wants to. might be some generations before we see their designs.

Lol, Qualcomm is Oracle of hardware world. Buying a company is a no way an indicator that they want to do anything. They haven't delivered a good SoC since Apple released A7.

The Snapdragon 870 just to name one.

To name what? Subpar SoC that looses in benchmarks compared to older A14? If it weren't for patents Qualcomm holding, there would be so many good SoCs on a market.

Except for the fact ARM is suing them to prevent using any of the Nuvia tech in their chips, since the license terms with ARM didn't convey with the purchase.

There is a good chance they get on the verge of shipping and find they can't actually sell any of their new chips...

https://www.axios.com/2022/09/06/arm-qualcomm-nuvia-chip-gia...

...and for some reason I conspire Apple is behind this

Does Qualcomm actually need to whole-ass a solution though? They're making gobs of money on mobile SoCs, any additional market for these chips is just gravy. Microsoft is the party that stands to really gain from a successful x86 alternative here but they don't seem like they're willing to pony up Apple or Google money to design their own chips yet.

Qualcomm is fully capable of building an SoC with eight X2 or X3 cores, for example, if MS is willing to pay for it. I think it's on MS that they didn't set higher performance goals for Qualcomm. Nvidia can also design good ARM chips (see Orin) but MS went and got married to Qualcomm (never do this!) so they can't use them.

Qualcomm's value proposition isn't the chip itself. It's the package, that includes the baseband.

>The solution to chicken-or-egg problems is known: you spend money to just overpower the problem.

Ah yes, "just spend more money" has solved so many problems over the years. Can anyone even name a single time a lot of money was spent and the product failed? Victory is practically guaranteed!

>I'm okay with Microsoft failing at projects like this. We don't need another Apple-like presence on the market.

Personally I’m not ok with being locked out of potential computing power by platform so I can’t get excited as MS struggles to keep up.

It’s either MS keeps up or the future ahead is dark.

> It’s either MS keeps up or the future ahead is dark.

You say that as if Apple will become some sort of PC-market hegemon, driving consumers and professionals to use its proprietary ARM hardware and OS despite having traditional Windows- or Linux-based workflows, simply because Apple's ARM is just so much faster.

Here's the third option: nobody cares what Apple is doing over in its corner of the PC market; the PC market remains an x86 market; and it continues to be driven by the needs of corporate buyers buying 1000+-part orders of PCs to outfit entire (non-IT!) businesses with; where those businesses don't care about having the fastest computer, but simply need "a" computer, with support and parts their internal IT department can swap out when needed; where the biggest factor driving purchases is TCO; and where TCO is driven down by commoditization and competition, not by vertical integration.

This still misses a large risk for Microsoft. Business software is moving to web apps. The orders of 1000+ PCs you've described will be soon replaced by BYO personal devices (who's owners prefer vertical integration like Apple for ease of consumer use) and Chromebook esque devices which are even cheaper than traditional PCs for a business.

Microsoft cannot keep doing what they are doing.

I wish Apple's proprietary computing platform keep niche but continue building good products. However iPhone become top share on some market and perhaps is it much higher share on enterprise?

PCs and "sealed consumer information-consumption appliances" (phones, tablets) are different market verticals — they will never converge, because the buyers of them have different needs that result in mutually-exclusive design constraints.

Even in Japan, where "nobody" owns a home PC, offices are still full of PCs; nobody is being handed a Macbook to do their work from, let alone a phone or tablet.

In my opinion, it's perfectly fine for Apple to win the "sealed appliance" space — since ~everybody else (your Samsungs, your Xiaomis, even Google) is just trying to do the same proprietary vertical-integration play that Apple is doing in that space; they're just worse at it than Apple is.

In contrast, Apple will never win the PC space, since their whole market strategy — "give people something different that forces them into our ecosystem" — is (and always has been) anathema to how boring, non-IT businesses want to use computers. Even companies whose workflows depend entirely on macOS "killer apps" (like desktop publishes back in the 1980s), who begrudgingly buy Macs to fulfill those needs, constantly lobby their app-developer ISVs to go multiplatform, so that they can toss out the Macs and revert to doing the same boring PC-centered IT that every other company around them is doing.

(And do note that Apple isn't trying to win the PC space. They were trying, historically — maybe up until the year 2001 or so. But ever since the iPod, and then the iPhone, Apple's strategy has moved to treating the Mac as a halo product category — a nice-to-have for those already in the Apple ecosystem for other reasons — rather than as something that's going to usurp the PC in its place one day. It's why, around that time, they killed XServe and Airport routers: they saw no further benefit in attempting to achieve corporate ubiquity, rather than complementarity.)

>> Also, frankly, I'm okay with Microsoft failing at projects like this. We don't need another Apple-like presence on the market.

Don't we?

I’ve gotta agree with you there. I love my Apple gear, but I hear and understand the criticisms against them. The above argument sounds to me like “you think one monopolist is bad? Wait until there are 2 of them!”

It will be an oligopoly then.

But wouldn't that be trending the right direction? Seems like a case of letting the perfect be the enemy of the good.

No, in the non-Apple world, Microsoft doesn't drive the computing industry. It's market-driven coordination between a multitude of manufacturers and retailers where no one instance has control over the others. Apple is a vertically integrated monopoly that effectively rules all its partners (from apps to hardware) with an iron fist and can push its decisions on users (and partners) unilaterally.

So no, I don't want Microsoft to become another Apple.

From my point of view, Microsoft failing to do what Apple does is a good thing because it means there's some semblance of a working, competitive (if imperfect) market. They can't force other companies to dance to their tune.

> so nobody is seriously working on consumer-targeted ARM.

Consumer targeted arm chips power most phones and tablets. Many of them are more than a handful for i5 and i3 class Intel chips and draw a lot less power. When you are talking about a $599 price point, you aren't talking about top-shelf Intel, anyway.

> But there is some sense to trying to lay the groundwork for future ARM SoCs running Windows

Now that ARM has grown up to be a viable alternative to Intel, this makes a ton of sense.

i3/i5 12 and 13 gen?

Outside m1/m2 there isn't any chip that can be compared to i5 intel 13 gen especially for PC space [that is in the same price bracket]. (maybe some arm chips can get close to i3 12100f but you can get it for 90 usd and get good single core performance so im not sure if in this price point is even any arm alternative)

what's the incentive for a chip manufacturer to put out a better ARM chip (eg comparable to the Apple M1)? I don't know that world well, so my best guess is that the margins go to the OS manufacturers, so unless you have guaranteed commitment from MSFT for a given volume, and some sharing of the margin, it's too risky to invest in the R&D to make a better ARM chip. The market for x86 chips is large and known, so for someone like Intel, it makes more sense to invest in the i3/i5 than in a new line of ARM. But that's all speculation - would love to hear the perspective of someone who understands the industry better.

Apple has done pretty amazing things with the m1. IMO the most unique part is scaling memory bandwidth.

The vast majority of PCs are running 128 bit wide memory, with workstation CPUs like the threadripper (and pro) being the exception, but a VERY small fraction of the market.

The M1 has 128 bit wide 67GB/sec peak (that you'll never see) bandwidth, like most PCs. Upgrade to the Pro and you get 200GB/sec. Max will take you to 400GB/sec, and Ultra takes you to 800GB/sec.

On the Intel (i3, i5, i7, i9) or AMD (ryzen r3, r5, r7, r9) you get ... the same memory bandwidth. Check the 8 core vs 16 core scaling numbers and for most benchmarks you'll see poor scaling. Sure you can increase GPU performance by adding GPUs, which reduces (but not removes) the need for extra memory bandwidth. Sadly iGPUs (outside the XboxX and PS5) largely stink and are only good enough for non-GPU intensive workloads. Apple on the other hand does scale GPU performance, granted not to the levels that AMD and Nvidia do.

So why can't anyone in the PC space do more memory bandwidth and a decent iGPU, especially when for years the GPUs were in short supply and had exorbitant prices. I think it does come down to OS support, volume (which could be problematic if current GPU customers avoid you), and potentially reducing profits for AMD (who would have sold an expensive external GPU). Not to mention that fast/wide ram requires soldering chips on board or increasing size/cost with large banks of ram. Even servers with 8 memory channels (minimum 8 dimms) only get you to the M1 pro level (1/2 of the m1 max and 1/4th of the m1 ultra).

Apple can say we have X% of the market today, and all new customers will be on our new platform with 2 years, so the driver, OS, iGPU, memory bandwidth, etc will be amortized over substantial volumes. Additionally Apple gets a larger fraction of the revenue, since they aren't paying Nvidia or AMD for a GPU. Who is going to push a MBP or Apple studio competitor that could ship the same volumes?

Long term there is a huge market for server-oriented CPUs that can compete with Intel Xeon, which currently has a near monopoly in the server market. Note that a lot of companies are already working on this: Ampere (ARM), Amazon (Graviton3, etc.), and likely Nvidia, Rivos (RISC-V), etc.

The best of market ARM designs don't really compete head-to-head with Xeon right now, but there are still a ton of server applications where they make already make sense. As a simple example companies like Google and Facebook have hundreds of thousands of servers that are doing things like running memcached or running some application like D/GFS where the server is mostly just doing a lot of I/O and doesn't necessarily need really beefy single-threaded CPU performance.

Longer term obviously if there are ARM or RISC-V CPUs that can compete head-to-head with Xeon in terms of features and single threaded performance then that opens up pretty much the entire enterprise/server market.

Performance is irrelevant depending on metrics.

If my new zCPU chip is 60% as fast as a Xeon at your task, that’s a problem.

If it can do it at 40% of the Xeon’s power, things get interesting.

I could use twice as many zCPUs, be 20% faster, and use 20% less power. That also means less cooling capacity in my DC.

Some tasks will always need the absolute best single threaded performance. But a lot don’t. And the Xeon’s power requirements leave a large opening we’re starting to see other companies poke at with things like Graviton.

> what's the incentive for a chip manufacturer to put out a better ARM chip (eg comparable to the Apple M1)?

Intel & AMD’s consumer and/or server market share.

I can’t imagine going back to an Intel after an M1. The battery life is better. It’s dead silent. It doesn’t get warm. It’s like a totally different kind of object.

I know PC people (those who want Windows) aren’t interested in an M1 Mac. That’s fine.

But I see PC laptop reviews with 4 or 6 or maybe 8 hours of battery life. They get hot but the fan isn’t “too loud”. And I know the performance isn’t the same.

And I just wish reviewers would call it out. They’re not on the same level. I’m sure fanboys would complain about the comparison in every review, but why shouldn’t Windows users have something much better? It’s been proven possible. Hold AMD/Intel/Qualcomm to the fire more.

If Qualcomm could get a chip with reasonable performance at a reasonable price that just doesn’t get hot and waste all its battery playing space heater, I bet they could really get a hold on the laptop market.

Proving your point, Amazon designed graviton, an ARM processor for their data centers.

Data centers know power/heat is everything. I’m not surprised they’re leading the way.

But as a consumer you’re stuck. You should have a machine with the performance of any normal/good laptop with way better thermals and battery life.

I’m not sure what the Uber-high end laptop would look like but surely it would do better than today.

The scale needs adjusting. What counts as “too hot”, “too short battery life”, “minimum performance.”

It all should have changed. But it didn’t. The industry acts like Macs are magic and therefor incomparable. “Of course that Boeing goes faster, it’s a jet engine plane. You can’t compare that to our cars.”

They’re both computers. It can be done. So why are Intel/AMD/Qualcomm getting off the hook to such a degree?

I just don’t understand it. It’s almost like Stockholm Syndrome or something. “Intel is nice to us, who are we to complain?”

>The industry acts like Macs are magic and therefor incomparable.

The cheapest M1 is $1300 in Canada. MacBook Pros are well over $2000. These prices are far beyond the budget of the majority of computer users. They are also beyond the requirements.

>Of course that Boeing goes faster, it’s a jet engine plane. You can’t compare that to our cars.

No, it's like comparing a BMW to a Honda. You spend a lot less money and then are surprised it doesn't perform as well?

I’m only trying to talk about just the chips.

Yes, the MBA is more expensive than most people would want to pay.

But it’s also not comparable to those laptops. It has a very high quality Retina display, the all-metal unibody construction, and Apple’s infamous profit margins.

There is no reason to think putting a similarly efficient processor in an otherwise standard PC laptop would raise the price much. Heck, Intel loves there margins too.

I don’t think the chip is too expensive. It’s all the other choices Apple makes that push the price up so much.

> . But there is some sense to trying to lay the groundwork for future ARM SoCs running Windows.

They've been doing exactly that since Windows 8, if not earlier. Perhaps the Year of Windows on ARM is somewhere around the corner from the Year of the Linux Desktop

I think we are on the ... third? ... attention cycle for this? Because they were trying pretty hard when W10 dropped too

The year of Linux-on-Windows-on-ARM Desktop.

> Microsoft is generally bad at managing and marketing their hardware projects and coordinating with manufacturers and retail partners

X-Box would beg to differ since 2002.

Surface products as well.

XBox is a great example of horrible mismanagement that cost Microsoft many billions of dollars, was almost abandoned, and went through a very rocky path to get to where it is today which is still frankly not that great of a position (it lags behind Switch and Playstation):

https://www.shacknews.com/article/121384/last-one-at-the-tab...

Don Mattrick almost tanked that entire product and it survives today because Phil Spencer miraculously managed to turn it around after all of the previous leadership was forced out of the company.

>XBox is a great example of horrible mismanagement that cost Microsoft many billions of dollars

Guess what, so was the first NES, and the first Play Station, the first Game Boy, etc. for their respective companies.

Breaking into a new market, with a new product, in uncharted waters, with no prior experience, with no support from clueless executives who don't believe in the new product looking for any reason to stop you from burning cash, endless turf wars such large and expensive projects create, makes it is hard, brutally hard, for any company to succed on the first try.

>and went through a very rocky path

The Xbox 360 sold 85 million units, one of the best selling consoles of all time.

The first NES and Playstation were MASSIVE successes for their respective companies, as was the Gameboy. In fact, the NES is credited with single handedly putting an end to the video game crash of 1983. The Gameboy sold out in a matter of weeks and Nintendo managed to sell every single Gameboy that it produced for the course of its first two years.

Here's an article about how massive of a success the original NES and Gameboy were and how it revived Nintendo as a company:

https://www.polygon.com/2019/4/19/18295061/game-boy-history-...

I have no idea where you got the idea that any of those products were mismanaged or cost those companies enormous amounts of cash and in fact those specific examples are among the most successful product launches in video game history.

>The Xbox 360 sold 85 million units, one of the best selling consoles of all time.

The XBox 360 is the best selling console from Microsoft and ranks 9th among all consoles behind the Playstation, Playstation 2, Playstation 3, Playstation 4, Switch, Gameboy, and Wii.

Read into that what you will.

>The first NES and Playstation were MASSIVE successes for their respective companies, as was the Gameboy.

I never said they weren't successful, I said they were also mismanaged during development like you said about the xbox, because management at Nintendo did not believe in the product.

>I have no idea where you got the idea that any of those products were mismanaged

Documentaries and war stories on youtube rabit holes.

>in fact those specific examples are among the most successful product launches in video game history

Today it's easy to say that with hindsight, but before they were launched, during their development, many in the company did not believe in those projects would succeed at all, leading to many internal fights and turf wars.

Also, Nintendo has a number of fuckups that bombed as well. Anyone remember the Virtualboy? Or the Wii-U? Gamecube also didn't sell too good.

The only thing that moves Nintendo merch is their exclusive IP (Zelda, Mario, Pokemon, etc), as their HW products are mediocre at best both in technical capabilities and in quality.

Can you please link to a Youtube video or article indicating that Nintendo management thought the Gameboy or NES would be a failure or that Sony thought that the Playstation would be a failure.

I am looking at some quick sources that I can find, and it looks like the complete opposite, that the management at Nintendo was very eager to develop a home video game console based on the success of their arcade games. They believed in the NES so much that when Atari bailed on its partnership agreement with them (due in no small part to the video game crash of 83), they went ahead and decided to do it alone.

Here is an article that was posted to HN awhile back that does a very deep dive into the development of the NES. It's an excerpt from the book "Console Wars" and it does not paint a picture at all like the one you're suggesting:

http://grantland.com/features/the-rise-of-nintendo-video-gam...

A relevant quote is:

"Yamauchi wanted Nintendo to aggressively get into the videogame business, which was really two separate businesses: home consoles and coin-operated arcade games. He saw the potential in these industries and took the necessary steps for Nintendo to enter both."

As an FYI, Yamauchi was the President of Nintendo.

I'm not sure what point you're making here. As the OP said, you took some anecdata about the Xbox's issues and speak about it like it was a disaster. They had problems, as did Nintendo and Sega (as talked about in Console Wars). It certainly wasn't all roses. Sega stopped building systems altogether!

The Xbox 360/One outsold the NES/SNES. That's amazing.

I'm not the one making a point in this thread, I'm refuting the laughably false point that the other person made that the NES, Gameboy, and Playstation were mismanaged and ended up costing their respective companies significant sums of money.

>As the OP said, you took some anecdata about the Xbox's issues and speak about it like it was a disaster.

What anecdata are you talking about? I linked to a source on the subject and not just any source, it's an interview with the head of XBox, Phil Spencer. I have also followed up every single one of my posts with a reputable source.

OP claims that they also have sources that suggest that Sony and Nintendo did mismanage the launch of their respective consoles in a manner that cost them significant amounts of money, so I'd like to review those sources.

The fact that he was actively replying to me until I requested some evidence to back his claims is a strong indication that those sources never existed.

>The fact that he was actively replying to me until I requested some evidence to back his claims is a strong indication that those sources never existed.

I stopped replying to you because I saw you have a bone to pick and I have a life to live rather than trying to win an online argument with you.

The sources are on youtube, feel free to use the search function and watch them at your own convenience and make your own opinion. Good day sir.

What a sad excuse of a reply.

Just have the decency to admit you have no source, maybe you misheard something, instead of doubling down on your ignorance. It's unbelievably pathetic otherwise.

> The Xbox 360 sold 85 million units, one of the best selling consoles of all time.

Define the best? It is 10th in a list of consoles that sold at least one million units. It sold less than console release later and went through even rockier path - PlayStation 3.

In a way, it's not much different than when they were porting Windows NT to Dec ALPHA. (But I suspect ARM has legs, where Alpha didn't)

Didn’t NT actually start on the Alpha and end up with x86 becoming a co-platform (and later main platform) during development?

Close. The original NT targeted an experimental Intel cpu called i860 (RISC based) nicknamed “N10”. NT was, according to Wikipedia, a play on words for the target cpu architecture.

From Wikipedia: “One of the original NT developers, Mark Lucovsky, states that the name was taken from the original target processor—the Intel i860, code-named N10 ("N-Ten")”.

NT was always supposed to be multi architecture (MIPS, Alpha, IA-32). I’m not sure what happened to i860.

Link: https://en.wikipedia.org/wiki/Windows_NT

That’s right. I forgot a third (failed) processor. Thanks.

Reply on news.ycombinator.com