Comment by mschuster91 | Hacker News Reader

Comment by mschuster91 | Hacker News Reader

Parent Full thread

The problem that Qualcomm and all the other ARM manufacturers have: they're impossible for tinkerers to get ahold of, outside of Raspberry Pi and a truckload of shitty, barely supported clones. And even then, the RPi still doesn't have the basics of PCIe working - in 2023 [1]. What. The. Fuck. Yes, Apple theoretically offers ARM devices, but they are not cheap, not extendable at all beyond USB-C and a ton of stuff doesn't work on Linux.

In contrast, say I want to develop something on Intel? No problem, I head to Amazon, buy a CPU, a motherboard and try if that old ATX power supply is still working. I plug in whatever card I need and it Just Works.

Steve Ballmer was right on track with "developers, developers, developers" - because if the ecosystem is crap or impossible to use for creative people on a low budget, guess what, they won't and go for the alternative. Linux started out on x86 for a reason, and Android blasted Windows Mobile (and everyone else but Apple) to pieces despite it being a solidly established player. A large part was due to neglected developers: outdated APIs, expensive and half-broken dev tools (developing for WinCE was a real pain in the proverbial arse), and a complete inability to even try to match Apple's innovation. Apple had a monopoly on capacitive touchscreens for years!

If the ARM ecosystem players actually want to throw punches towards the unholy duopoly of Intel/AMD, they need to standardize on a common core of UEFI and get modular, working components out on the market.

[1] https://www.jeffgeerling.com/blog/2022/external-graphics-car...

> The problem that Qualcomm and all the other ARM manufacturers have: they're impossible for tinkerers to get ahold of

This sounds more like a problem tinkerers have rather than a problem that Qualcomm or ARM manufacturers have.

Qualcomm doesn't care about the tinkerers so it's not a problem to them.

Embedded ARM systems like my router or smart thermostat seem fine.

Google controls Android and it also seems fine for 99% of their users. There are thousands of Android developers.

There are ARM boot standards like

https://developer.arm.com/documentation/den0029/latest/

https://en.wikipedia.org/wiki/Server_Base_System_Architectur...

But in general embedded device makers don't care about having a common boot system. They want to make their device as cheaply as possible. If the common standard helps them do that then they would adopt it. Otherwise it is unnecessary overhead.

So what do you actually want? A cheap desktop level ARM system? I doubt that is going to happen. The number of people who want such a thing is extremely low even though places like Hacker News make it seem like it would be popular.

> Qualcomm doesn't care about the tinkerers so it's not a problem to them.

Qualcomm is selling performance. Performance is a combination of the hardware and the tubing that developers do to make their products perform well on the hardware.

On x86 or M1/M2, developers develop on those chips. Graviton and Qualcomm trail significantly in this regard.

Listening to the Qualcomm management they seemed more interested in selling licenses to technology than actual physical products.

But Qualcomm is selling more than just performance. Cost, quality, schedule, delivery of products. Qualcomm has shown they can make an SoC good enough satisfy a lot of phone manufacturers and there are thousands of Android developers and applications.

Would the Qualcomm / Google / Android ecosystem be better if they opened up more of the hardware specs to tinkerers? Maybe but I don't think they care or they don't see the business case. Apple's ecosystem is even more closed and their performance seems to be better in general.

I sat in a lot of all hands meetings when I was there listening to revenue and profit breakdown. The QTL licensing division was far more profitable than the QCT hardware division. These numbers are publicly available.

Anyone can easily develop on Graviton.

The solution is to do as we did back when everyone on the company shared UNIX development servers. There wasn't an Aix tower under my desk.

What I want is anything ARM that's actually general purpose (i.e. no SoC whose BSP only is usable on Android or whose IO peripherals barely meet standards), with full Linux upstream support for all components, and at least some performance compared to your average Intel i5.

The status quo is that the most powerful general-purpose ARM system you can get is an RPi or its clones (which both fail the performance test, and outside of the RPi a lot fail the Linux test), as the only ones actually able to buy high-performance CPUs are phone manufacturers and large hardware ODMs such as QNAP (who're using Annapurna CPUs IIRC).

Apple is the "compromise" - extremely high performance, to a tune no one other matches, but sparsely documented/supported by anything not macOS (as Asahi Linux shows), and a price tag to match.

Nothing stops you from buying this machine https://www.ipi.wiki/products/ampere-altra-developer-platfor... (except the poor performance Vs cheaper x86 processors).

Yeah... 2.666 $ and a lead time of 3 weeks. That's exactly the kind of bullshit I'm talking about - the target class for this is embedded developers, but not tinkerers and ordinary developers, the latter of whom you need as a platform the most.

> AADP is a prototyping system targeting general embedded applications based on the COM-HPC Ampere Altra module and a reference carrier board, the COM-HPC Server Base.

They do use the term embedded on that site which I thought was strange.

Ampere is an ARM server startup. Their big customer so far is Oracle who sells Ampere cloud instances.

https://www.oracle.com/cloud/compute/arm/

I've worked at multiple companies with ARM architectural licenses including ARM itself. No one there cares about making a non embedded, non server, chip in a system at a price equivalent to a desktop or laptop x86 system that you can buy from Newegg / Best Buy.

You may think it is cool but no one is going to make it for you.

The trend for ARM is to continue down the embedded way and at the hyperscale data center area like Amazon Graviton and their competitors with their own custom ARM server chips. Since they are custom you won't be able to buy one to put in your house, only rent it in their cloud.

Ampere is a startup so their goal is to get bought by a big company. I don't know how much longer you will even be able to buy that system.

What is the actual price and performance point that you would pay as in the exact amount of dollars? Go talk to the marketing departments of these companies. They have already looked at this. They think the market is tiny and know the investment would be hundreds of millions and is not worth it.

Why do you want an ARM system so bad? What is wrong with x86 for the non embedded / non server / middle desktop range? Literally billions of people are fine with x86 in this range.

> Why do you want an ARM system so bad?

I want more actually viable options in the ARM space for consumer general-purpose computing. Why? Because Windows laptop battery life sucks ass because both Intel and AMD can't be arsed to do what it takes to even get to half the runtime Apple gets out of their systems. I'm happy with my 2022 MBA, but Apple's pricing surcharge on memory and storage is only possible because there is no competition.

At the moment, the only option consumers have is Apple, and that only runs macOS, but no Windows (and Linux, but with some serious drawbacks). People didn't buy Microsoft's first dabbles in ARM that used Qualcomm SoCs because the SoCs were underpowered garbage (no wonder, they were designed for phones) and there were no apps, app developers didn't develop for ARM because there were no actually usable development machines that didn't cost an arm and a leg/had ridiculous lead times, and there was no Rosetta equivalent because the CPUs were garbage and underpowered... a self-reinforcing circle.

Volterra devices are quite affordable.

Windows will always have this issue, because Microsoft isn't Apple, their hardware is only to inspire OEMs, they don't have the means to drag everyone screaming into new hardware no matter what.

Windows NT supported several CPU architectures from day 1, they died as consumers couldn't care less that PowerPC, Alpha, Itanium... were supported.

Highly doubtful that you will see any in the near future.

Qualcomm's previous custom ARM core team and server team got let go. Most went to Microsoft. Microsoft was trying to copy Apple and integrate hardware and software more tightly together. Microsoft laid off their custom ARM core team about 4 months ago. Most of them have gone to ARM. Anything Microsoft does in the future with ARM would be a licensed design from ARM.

Qualcomm bought Nuvia a couple of years ago. Nuvia was a startup that had a custom ARM core design. Most people think Qualcomm will use the team for either the mobile or server markets again. This article does mention desktop but with ARM's lawsuits against them who knows when or if it would happen.

https://www.datacenterdynamics.com/en/news/qualcomm-pushes-f...

The other companies like Amazon will want to push you to their Graviton ARM cloud. Ampere wants to get bought. The people there that I know joined because it was a startup. They want to get rich more than they want to push some kind of ARM general purpose computing concept.

To be fair, the PCI Express bus on the current Pi was kind of an afterthought, only meant to work with a very limited set of devices, so I'm pretty sure nobody at Broadcom, and few in the design stages at Raspberry Pi, had ever tested more complex devices with it.

It works fine in _most_ cases for simple devices (USB controllers, SATA, NVMe, WiFi, and the like), but really falls apart for more advanced devices (hardware RAID, GPU, TPU, etc.).

And all Arm processors have to deal with cache coherence issues (which aren't a problem on X86), meaning some drivers (notably, AMD still) need to program for the different architecture (some patches exist but they're not perfect yet, and not in mainline Linux).

> To be fair, the PCI Express bus on the current Pi was kind of an afterthought, only meant to work with a very limited set of devices, so I'm pretty sure nobody at Broadcom, and few in the design stages at Raspberry Pi, had ever tested more complex devices with it.

The first Raspberry Pi was sold over a decade ago and I 'member people actually using them as embedded boards for whatever stuff they were working with eight years ago (especially once the GPU performance became powerful enough to run digital signage). Sorry but that a company like Broadcom can't be arsed to develop a standards-compliant PCIe interface is a joke, and with this kind of attitude the ARM world complains that no one buys their chips?!

(Edit: Oh, just noticed whom I replied to - the person who wrote the article I referred to. HN is a small world indeed, and I guess we share at least some of our frustrations)

> meaning some drivers (notably, AMD still) need to program for the different architecture (some patches exist but they're not perfect yet, and not in mainline Linux).

Drivers... oh don't get me started on that front. Everyone in the x86 space seems to have learned over the last two decades that it is a good idea to submit drivers to the Linux kernel early. Intel and AMD both do that for CPUs and also for a lot of their other stuff. In contrast, the entire embedded world still locks away drivers behind years-old kernel forks, ridiculous NDAs, absurdly expensive dev boards, completely whack u-boot forks and even more whack BSPs.

> And all Arm processors have to deal with cache coherence issues (which aren't a problem on X86), meaning some drivers

Why are they a problem on Arm and not x86?

x86 is two well organized behemoths designing chips (including the base designs for most motherboards).

On ARM, some SoC designers slap together IP blocks without fully understanding the implications. Not all SoC designers. But there are a lot out there, and not all have as high of standards as Intel and AMD.

IIRC, the PCIe implementation on the SolidRun Honeycomb is quite good and can be used with GPUs.

> ...they need to standardize on a common core of UEFI and get modular, working components out on the market.

Except that this DOES already exists. Right now.

The standard is called: Arm SystemReady SR [1]. I'll be the first to admit: It's far from ideal. There are only a couple of high-end options on the market and lead times are measured in weeks.

But this development is new as of this year. And I think it's terribly exciting movement in the right direction (finally).

[1]: https://www.arm.com/architecture/system-architectures/system...

I admit it's been two years since I last dabbled in embedded, never heard of that one. I hope this initiative finds some success!

Qualcomm doesn't have a storefront, that's true, but they have a partner search that will show you all the partner-made development kits you can buy. They aren't going to be nearly as inexpensive as an rPI but they don't make finding products you can get too difficult and it isn't just nabbing some random board from AliExpress.

https://www.qualcomm.com/support/qan/member-directory?active...

I’ve been wondering if ANYONE has done something with the more recent AppleTV4K hardware and ported say ashai or other Linux versions being worked on for Apple Silicon. Or perhaps getting containers running on the hardware.

Reply on news.ycombinator.com