initially all RPi boards were booted from an sdcard (that's why it has an sdcard slot). Now the RPi4 can boot from an usb3 device, which is much faster. You can use a disk or a flash drive or whatever you wish.
Also you can buy an "extension" board if you wish to have a sata port.

But anyway you are looking for an arm based rival of an x86 pc. This obviously depends on the usage. RPi4 is a very good competitor, but there are cases when we would need more...

I've run my RPi4B & RPi400, both 4GB, with micro SDHC cards (from 8GB to 64GB), USB2 & USB3 pendrives, external HDD, external SSD.
The speed of the pendrive & mSDHC cards are close, but using external drives, especially USB3 SSD, are certainly fast enough for a basic desktop computer.
(I'm currently running mine from a USB3 250GB HDD.)

Re Sinclair, I remember him well, especially his C5....

Agreed. RazPi 4/400 are good enough, but not very future proof.

It appears there are Arm servers. Most are more cores/less speed which ties down individual threads. On ebay, I can buy a ThunderX 48core Arm server, 2Ghz for $2000 = £1673 (pre-Brexit but watch that space). They do 2 Motherboards

• 48 core M/B @ 2.0Ghz - $1200
• 32 core M/B @ 1.8Ghz - $900

Frankly, I am definitely not at that place. I also noticed the M1 can do turbo to 4.5Ghz. So Arm cores can be made to go fast. I'm disappointed to see so many iterations of them aimed at the low power/low performance space (which they own, incidentally). The big endian design is definitely the way to go.

The SD card is required for the OS to boot since the Pi has no internal storage. If used only for the OS with a USB drive for data it would work well.

USB drives are a norm today so why would SATA be required? Not to mention that SATA would require a beefier power supply than the 5V 3A that is standard for the Pi4 and a separate power connection for the SATA device. The Pi4 is only rated at 5V3A maximum so extra power demands would quickly overload the capacity of the board and likely cause trace failure or under voltage conditions. AFAICT none of the PIs have ever hinted at SATA support except with add-ons.

Some choices up to servers I'd think but no real massive home user hardware like x86/64.

https://www.gigabyte.com/us/ARM-Serv...T92-rev-100#ov


hahaha doesn't have to cost an ARM and a leg. https://www.toptal.com/back-end/arm-...or-datacentres

Yeah, there are servers out there.

Unfortunately, most of them are under-speed, ≤2Ghz and throw cores rather than speed at the problem. That may work for server loads, but when you're looking for a Future-proof Desktop, that's far from ideal.

The Mount Snow & mount Jade from Ampere Computing are offering 3.3Ghz core speeds at about 5W per core which means the fab is quite small.

Frankly I see Moore's Law as outdated, I know there will have to be a massive internal shakeup in ASICs because the doping layers at a microscopic level aren't simply wide enough to provide insulation, and you are running out of compromises very fast in there. So FET gates shoot through to transistors, and the whole house of cards collapses. I reckon that if you own a PC company, now is the time to sell it, because the 'upgrade every 2 years' cycle can't continue.

Mobiles isn't much better. But you probably trash one every 2 years, so that's ok. I bought a Samsung S7 Edge in 2016 or 2017 - last year's model then. It has 7nm fab, a snapgragon 820 = 4 core 2.2 Ghz. That beats the RazPi 4 hands down. Just like the RazPi (@1.5Ghz) has been successfully overclocked to 2.2Ghz, I'm pretty sure my phone could be overclocked to near 3 Ghz, which would be fine.

I just didn't realize it was going to be such a big deal to get an Arm desktop. Servers are not inclined to be priced for the faint of heart!

Just a random thought, what does the re-launched Amiga use? If memory serves that thing's SoC comes from Freescale, and I know it isn't x86, is it ARM? (and be forewarned: that thing is a little pricy)

I know if you're looking for 'RISC' more generally there's those Raptor Systems towers with POWER9, which are supposed to be much more competitive with Xeon/Opteron, if you can get anything to run on them.

I had a look. They obviously don't expect many sales, because they're using an FPGA, the Altera V4. That's the option when you don't want to lay out serious money to get an ASIC ready for fabrication. It has to load it's configuration each boot, then it's good for some hundreds of Mhz

Previous iterations used a 68K core asd a games machine. I think the thing is a souped up pin compatible board for your old Amiga to give it a new lease of life. They're calling it 'Vampire' V4, but 'Zombie' might be more appropriate. Is that what you meant?

Fab is 28nm, but it's the chip design that slows you. But back in 2013-4 I was looking for an FPGA chip to do 300Mhz, and there was no difficulty with Altera.

I thought they had a fully functioning 'modern' desktop based around a non-x86 CPU, not just the FPGA drop-in. I found it on Wikipedia, but they appear to be PowerPC machines (although if you follow that rabbit hole down, the newest generation use an NXP chip which has some ARM counter-parts in its product line.

Links:
https://en.wikipedia.org/wiki/AmigaOne
https://en.wikipedia.org/wiki/QorIQ (note the 'L series')

From doing some more digging, this may be of interest to you:
https://www.techradar.com/news/qualc...-experts-think

And queue the obligate "anything Apple does is always 2-3 years behind Microsoft, yet they get all the credit for being 'innovators'" - no idea if you can boot linux on any of those devices, but they should be certainly faster than the raspi 4...

Ok, I think we're agreed that the resurrected Amiga is not at the races power wise. It has a power pc.
The Qualcomm Snapdragon however, is a different beast. These are Arm based, so compatibility with PPC is gone. Snapdragons are multicore Arm 'best-in-the-west' jobs but profiled at mobile operations. I may have mentioned my 5 year old phone has the Snapdragon 820 (4 core, ~2Ghz, 12nm fab). With a mobile platform, they have a low base clock and turbo boost for power, which is fine, considering there is battery, and no cooling. On a desktop, however, you primary focus is power rather than economy. The Snapdragon 835 is more of the same with 8 cores, and various optimizations for mobile loads. They even do one for windows loads. All are Arm based with lithography (Fab size) as low as 7nm.

In their market, Snapdragon have hit the plateau, as X86 has in theirs. Personally, my first impression is that the M1 from Apple has more to give - 20 watts more to be precise. What I'd like is something like Ampere Computing's 3.3Ghz cores (before turbo IIRC) in a desktop, not a server.

Right.
Hey - its the best I could find by stumbling upon it randomly...

Still might be more useful/interesting than a Raspi 4 since, at least in the form of those laptops, it will be a more 'finished' product as well - like I'd assume they will have RTC, soft power support, etc as opposed to the kind of 'hacky' nature of the rpi.

Yeah, the Snapdragons are good. and we're not mentioning the dark horse in the stable.

Nvidia bought Arm for $40,000,000,000. That's an awful lot of zeroes behind the '4.' That offers the potential of a match made in heaven, multicore Arm CPUs and peerless Nvidia GPUs in a tiny fab sized APU. Arm also have experience with integrating peripheral chips. The only snag is, I don't see the market. Sure, it would be nirvana for us ≤5% of linux geeks. Kernel compile time might get down to seconds(!) but who (except us) cares? I don't see M$ stepping up to the plate, or every software company recompiling for Aarch64.

In fact, somebody probably had done all the thinking before Arm was ever approached. You could buy an awful lot of companies with $4^10. I'm just wondering what was the plan?

Although even for the geeks, not sure how many of us would want this. If Nvidia released the GPU for these w/ open source drivers, sure, I'm down. But as long as it requires proprietary drivers, I'm out. I'll stick to anything with open source drivers before ever touching Nvidia again as long as I live.

Agreed, Timothy, a lot of the geeks (perhaps including myself) would be out.

I have just marked this thread solved, because I think I found the Answer - Sadly, it's along the lines of "No … Get lost."

Nvidia are BIG into AI. Something I wasn't aware of is that Arm has a Neoverse core which is unlike the consumer offerings we have of cheap cores. It's profiled for paralleling, for the server and the AI types who are building ridiculous numbers of Arm cores with ridiculous numbers of Nvidia DSP cores to manufacture for AI, where people will pay silly money. After Nvidia bought Arm, all of that silly money will be going to Nvidia. X86_64 can't compete, because it's too power hungry, and consequently you can't integrate as tightly, or be as competitive. Apple, who think they are clever with their M1 in their laptops and their fancy graphics are actually still going after peanuts - a 5% share in the extremely price sensitive pc market. As for linux geeks … who are they?

So Arm will keep their mobile business, and develop their AI business, which is IP Cores. Sure, if they have a few engineers sitting around and doing crosswords, they might get them to stick an 8/16 core Arm CPU with a decent Nvidia card,and stick it out there as a games machine for exhibitions or displays. I can't see them selling it.

My SWAG as to why nVidia did this is for a few reasons:
1) They're 'over-valued' right now and probably trying to get something more tangible out of that 'equity' before it corrects down.
2) If I remember right, they actually integrate ARM chips into their GPUs as supervisor controllers, I read somewhere they had been looking at RISC-V to replace it (because the 'current' solution is supposedly starting to be inadequate vs the complexity of their newer GPUs), but probably it made more sense to someone to keep everything more 'locked down.' If I remember right the lack of documentation/support for that supervisor IC is also why noveau performance takes such a hit vs their proprietary driver.
3) They've been shipping the exact kind of ARM + GeForce APU you describe for a while, with Jetson and Tegra (as well as some mobile chipsets for Nintendo), and they push this hardware for machine vision, IoT, machine learning, etc. If I remember right the 'big market' for them is self-driving/autonomous vehicles and they're probably trying to better cement their position as that segment starts to expand.
4) They probably also have some fantasy of re-capturing the more mainstream gaming market (consoles/arcades/etc which is a multi-billion/yr industry) from AMD by being able to offer a complete APU package (and/or getting more into 'mobile' against Qualcomm and Samsung), but A) they've already been trying that (since they were/are an ARM-licensee) and it hasn't taken off and B) it would require aarch64 to be more supported by game publishers (Microsoft actually does support this 'in house' - most of the Snapdragon laptops I linked ship with Windows 10 out of the box for example, but good luck throwing games on them), which hasn't materialized (again apart from Nintendo's closed platform).