One of the key elements I’ve always found frustrating with basic software development is that it can often be quite difficult to actually get the hardware in hand you want to optimize for, and get a physical interaction that isn’t delayed by networking or logging in or anything else. Having a development platform on the desk guarantees that direct access, and for the non-x86 vendors, I’ve been asking for these for some time. Thankfully we’re now starting to see some appear, and Avantek, one of the Arm server retailers, have built an Ampere eMag workstation out of a server board, with some interesting trickery to get it to fit. They sent us one to have a look at.
This is only the unboxing and short first impressions, but I am unreasonably excited about what are effectively bog-standard PCs, but with an ARM processor. I can’t wait for these machines to come down in price, because this is the first time in a long, long time that we’ve seen what could become a serious challenge to x86 in its traditional space: desktops and laptops.
Once AnandTech publishes its actual review, I’ll be on top of that, too.
i wonder what kind of bootup process is implemented. is there some kind of in-firmware bootloader?
afaik this is a bit of a wild-west territory in arm space.
It isn’t wild-west. It’s UEFI and ACPI on arm(64), pretty well-known.
In my mind ARM’s momentum isn’t something to be excited about, it’s dangerous.
In no way do I want to enable the architecture that finally dethrones x86 to be another license-locked, closed mess.
If that happens we will be stuck with it for decades to come.
I’d rather oppose ARM at all avenues and throw support behind RISC-V
license-locked, closed mess.
It is not the mess. It is a well-defined architecture with wide software support. ARM license is relatively cheap. Modern SoCs development cost anything from 300 millions to billions. It just doesn’t matter if arch is free or licensed at that point.
viton,
I think he’s partly referring to the closed nature of many ARM SBCs for the purposes of end users. What is often a problem with ARM in practice is being dependent upon the manufacturer to provide a fully compatible kernel and then not being able to update it, just as we’ve been suffering with android for over a decade. This is (and could continue to be) a huge problem with ARM based devices.
Ironically x86 CPUs, while proprietary themselves, have a long & reliable history of separation between software developers and hardware manufacturers, this is key!!! It removes the conflict of interest that many hardware vendors have in terms of ceasing support and planned obsolescence forcing us to buy new hardware in order to use new software. While this has absolutely nothing to do with the underlying technological differences between x86 and ARM, it’s still a biproduct of how the respective industries have evolved for better or worse.
I see merit in ARM. It feels like christmas when brand new ARM SBC kit arrives at the door, but frankly it needs to be supported by mainline linux 100% without exception so that I can support the hardware indefinitely on my own terms. I have enough experience to know that I will ultimately always come to regret being stuck with manufacturer kernel builds. And it’s not just a problem that grows over time, even brand new ARM SBCs often come bundled with a couple year old legacy kernels *on day one*! This isn’t acceptable! I’ve expressed a desire to get some good & affordable ARM server products for a long time now, but I will not compromise on mainline linux requirement, otherwise it’s more pain than gain.
This.
Idealistically, ARM looks like a great idea. In real life, it’s filled with predatory companies.
I always look at ARM boards to get some low power stuff, but I always end up with x86 stuff. Stuff just works with the x86 boards.
I’m hopeful POWER will pick up some momentum.
Define “predatory” in this context, give some examples, show your work.
While I can think of companies that don’t openly document their chips — they require NDAs or well established business relationships — I can’t think of a single ARM manufacturer I would call predatory.
Example predatory ARM vendor: Qualcomm – https://duckduckgo.com/?t=ffab&q=qualcomm+lawsuit&atb=v147-1&ia=web
Does Apple count too? They’re not really a chip vendor…
@chrish I will point out the lawsuit you point to doesn’t really have anything to do with ARM chips per se, but with wireless chipsets. So I am not quite sure how that applies to @Flatland_spider’s point.
While Qualcomm requires establishing a business relationship, signing NDAs, and a high MOQ, what I know about their pricing isn’t predatory, or that they actively discriminate in who can purchase. But those conditions I described are not unusual in the semiconductor industry. Qualcomm from everything I can tell supports the Linux kernel quite well
Finally Qualcomm is also a member of the RISC-V Foundation, so by that logic the RISC-V community is filled with predatory companies?
RISC-V is like ARM in that it is a specification and reference implementations, but — like ARM — it doesn’t have a standard set of peripherals or interfaces. So RISC-V vendors are free to limit who they sell or share detailed register level information to… just like ARM.
I don’t know how many architectural do-overs we get as a society before something is so embedded that it is impossible to change.
At this stage if something is going to dethrone the long-standing defacto, I want it to be open, not closed.
Note that this refers to the specification not implementation.
If some company wants to implement a closed-design product, fine, but the spec should be open and free to implement.
Why would we limit ourselves innovation wise to a closed, controlled standard.
Why can’t we have an architecture that defines a way to inter-operate, and let the merit of what companies can come up with speak for itself?
What would things have turned out like if anyone could make an x86 chip?
(This is a reply to the thread in general, not specifically you Alf)
jmorgannz,
I completely agree.
I actually think it’s easier (meaning less expense and fewer skills necessary) to create an ISA spec than it is to build the physical chips. A single grad student could realistically write a workable ISA spec as part of their thesis (and probably one that’s better than x86 too). However outside of billion dollar companies, few could actually fabricate the chip.
So I’m not sure how much difference it would make. Let’s assume apple & google could manufacture their own x86 chips…it’s hard to say if that materially changes things. Maybe the market becomes more competitive, which would be good. Maybe we’d finally get x86 CPUs with unlocked management firmware (which is a security risk today), this would be good. More vendors could mean more vendor-specific extensions, which could be bad for fragmenting the x86 ISA. Maybe they could do a better job than intel and AMD at speeding up x86, which would be nice, but a lot of skepticism is warranted.
I personally feel there could be value in incorporating FPGAs tech into standard CPUs, something that intel hasn’t been doing, maybe if there were another vendor they’d be able to get the ball rolling on that.
That’s right.
Most of the time SBC is either some embedded stuff, toy or intentionally closed system.
And it creates a lot of pain to the user. There are projects like https://rpi4-uefi.dev/ to ease the pain.
But the serious machines are supporting the necessary standards . Boring, like x86.
Avantek has been selling ARM based desktops for years.
ThunderX desktop was even affordable. But ThunderX2 Station is still very expensive – $13k down from ~16k.
Interesting, ARM is going back to where it started, to the desktop.
Kinda? I mean you could always buy non embedded ARM motherboards if you wanted. And that machine is really a server class CPU in a tower case. Still a very cool piece of kit
I meant like this.
https://en.wikipedia.org/wiki/Acorn_Archimedes
Yes so did I. RiscOS machines continued to be sold by other manufacturers, as well as Linux/AROS ARM motherboards from companies like Genesi Efika. Like I said, ARM never left the desktop, it was just much much more popular in other places
I guess we agree. 🙂
“ARM is going back to where it started, to the desktop”.
They have been back for years …. RPi is desktop.
Iapx432,
It’s impressive that raspberry pi and others like it are capable of running desktop software, but it’s a very low end desktop. I suspect a lot of people are like me and consider it more as a flexible IOT platform 🙂
125 watts? Yikes!
I know, I know, 32 cores, 2.8 GHz makes it a screamer. That’s a lot of Folding@Home instances on one machine.