System76 Comments On Their Open-Source Hardware Plans & US Manufacturing

[freespirit]

Not gonna happen. Not gonna happen either, also where is ARM getting powerful? It's still struggling in anything beyond low-end laptops.

https://www.bloomberg.com/news/articles/2018-04-02/apple-is-said-to-plan-move-from-intel-to-own-mac-chips-from-2020

let's see, they already changed from power to x86, the world change everytime, nothing is forever

[c117152]

Ultimately, creating an open source gpu that can compete with AMD/Intel/Nvidia would require a company of comparable size. Like IBM with Power9. There might also be a chance for HP to try something I guess.

There half a dozen Chinese and Europeans startups doing GPU designs as we speak. Depending on politics and market conditions (especially in the AR/VR, AI and vehicular UIs), one of them might end up releasing open source.

But it's 2-3 years away at least.

[GruenSein]

Heh, what I read was that its singlecore performance was somewhat close to the single-core performance of a Xeon 8176 https://www.anandtech.com/show/13392...icon-secrets/4
which is a Xeon with like 28 cores, 56 threads and can be mounted on boards with 4 CPUs (i.e. it is designed for parallel loads, not single-core).

And they claimed "desktop performance", no that's still low-end. A desktop CPU from Intel has much more IPC than a server CPU with 28 cores.

The "pure feasibility perspective" should also deal with the annoyance of transitioning all their current x86 ecosystem to ARM, which is not really a fun thing.

Apple has pulled off a transition like that twice already with only minor problems. On my first Intel Mac, all my software kept running as expected and was updated to a native binary relatively quickly. Additionally, macOS is basically the same as iOS. Even the GUI frameworks are going to be compatible pretty soon as can be seen with the current audio memo or stock apps, which are literally the same on both OSs. What that means is that the vast majority of Apple‘s software stack already runs on ARM in a production ready state. It is tried and tested by 100s of millions of devices already. Also, AFAIK all software submitted to the App Store must be submitted in LLVM IR. Turning most software into ARM native binaries would require little more than flicking a switch.

A quick search also turned up these single core geekbench scores:

Apple A12 in the iPhone XS - 4795@2,5ghz
Intel i7 8700K - 6497@3,7ghz (this is the base clock. In the single core test, it probably boosts well beyond 4ghz)

Even if we consider only the base clock, the A12 already provides 9,2% higher IPC in this test.

[rabcor]

There half a dozen Chinese and Europeans startups doing GPU designs as we speak. Depending on politics and market conditions (especially in the AR/VR, AI and vehicular UIs), one of them might end up releasing open source.

But it's 2-3 years away at least.

That is wishful thinking. While china is a communist nation, seems like it would be more friendly towards open source... The people who run the companies, their capitalists, are even greedier than the ones in western countries.

It's a very natural way of thinking that you won't be able to compete with other companies if they can read every detail of your designs freely while you can't do the same for theirs.

[microcode]

I respect the ambition, and I can respect the desire to move away from just reselling Clevos.

[tuxd3v]

Not gonna happen. Not gonna happen either, also where is ARM getting powerful? It's still struggling in anything beyond low-end laptops.

Hi1610, this one could be a option for Desktop

[wizard69]

This Is abit excessive! As long as they are honest and up front about the open parts it is an old machine in my book.

The bigger problem in my mind is the focus on X86 and in that regard I agree with you a bit more. ARM would have been the better choice for distinctive and innovative hardware. ARM does suck though for open hardware and especially driver support.

Beyond that I really think they are nuts to be going after the laptop market. Instead the world needs a Linux based tablet that an really give Apples iPad a challenge. No I don’t mean Android!!! The biggest problem here would be finding a SoC equal to the 10 billion transistor A12X.

The problem with laptops is the very crowded market. Very few retailers will carry the machine. On top of that commodity laptops are revved almost monthly. At l ast with a tablet you are competing against one manufacture that at best has updated models ever year.

So it's not open, it's just another pc, like Dell, Apple and a self-made pc, all that enphasis for nothing
I think they shoudn't say open hardware if is not completly open, i can't say i have a Ferrari just because my car is red

I really hope the laptop they are talking about will be for real a complete open hardware, maybe using power, risc-v or arm, in case of arm not the one with blobs inside or required in firmware or driver

I understand that is not easy to accomplish that kind of purpose, but call open something is not open, it's a bad move, if they wanna push intel/amd to be more open they should try maybe with purism, or just let x86 behind and looking for alternative.
Apple will leave x86 and i think in the next 5 years x86 will lose a lot of power on market, because arm is getting powerfull enough, and if the big heads on Intel do not understand it, they will just make the 2nd big fail after say no to Apple for a mobile cpu.
Things change now they are leader tomorrow i doubt it, to completly open the architecture could be a great move for them, because people and company who care about open hardware will not leave them, and manufacturer and developers will have a lot of info to make it better.
2019 is close, and the close software and hardware are lost causes, because everything open will have more effort for everyone, and everyone will put in there their 2 cent because they can use it when the techology is mature, big companies put money on linux, not in windows, smartphone manufacturer, communities use android not windows mobile or iOs, and if you take a look at market android is the big player, not windows mobile.
Once upon a time Nokia and symbian, Blackberry, and other BIG company, now they are dead under open source project

[wizard69]

Not gonna happen.

I’m not too sure about that. Obviously they didn’t with today’s Macs, on the other hand with the A12X in the iPad the certainly could have! You really can’t laugh at 10 billion transistors and all that specialized hardware. You really can’t say never here.

Not gonna happen either, also where is ARM getting powerful? It's still struggling in anything beyond low-end laptops.

A12X is certainly powerful enough to do low or middle end machines. A flocked boost and enhanced version could easily handle most laptops. Of course we will likely ever see Apples A12 series available to builders.

That is OK because there are ARM based chips with as many cores as you could reasonably want available. The idea that ARM can’t cut the mustard is really based on past memories not modern realities. You just need a chip engineered for the platform.

[wizard69]

Yep one look at A12X and it’s 10 billion transistors should highlight ARM potential to anyone. These are extreme complex chips too with a good deal of special function logic.

Apple's SOCs are immensely powerful considering that they run at TDPs around or below 5W. I am sure, you can find benchmark scores on par with mid-tier i-series Core CPUs. And that is accomplished at relatively low frequencies. I certainly cannot predict Apple's plans but from a pure feasibility perspective, they could 100% switch at least their entire laptop line to ARM within a few years.

Apples processors indicate two things. One is that Apple could go ARM at anytime. The Second is that ARM is very capable of high performance in general.

[wizard69]

I might add that most of the Linux world also runs on ARM, so a “HomeBrew” for an ARM based Mac would be easy. On that topic; we have never enjoyed a time in history where C and C++ compilers where so compatible across platforms. If you build code on Linux x86, with GCC, it is very likely to compile on a CLang compiler on ARM based Macs. I’m not claiming absolute perfection here but there was a time when code written on one C++ compiler simply wouldn’t compile on another. So while there may issues beyond compiler compatibility, the reality is most open source code now runs on ARM and thus will quickly make its way to an ARM based Mac.

Apple has pulled off a transition like that twice already with only minor problems. On my first Intel Mac, all my software kept running as expected and was updated to a native binary relatively quickly. Additionally, macOS is basically the same as iOS. Even the GUI frameworks are going to be compatible pretty soon as can be seen with the current audio memo or stock apps, which are literally the same on both OSs. What that means is that the vast majority of Apple‘s software stack already runs on ARM in a production ready state. It is tried and tested by 100s of millions of devices already. Also, AFAIK all software submitted to the App Store must be submitted in LLVM IR. Turning most software into ARM native binaries would require little more than flicking a switch.

A quick search also turned up these single core geekbench scores:

Apple A12 in the iPhone XS - 4795@2,5ghz
Intel i7 8700K - 6497@3,7ghz (this is the base clock. In the single core test, it probably boosts well beyond 4ghz)

Even if we consider only the base clock, the A12 already provides 9,2% higher IPC in this test.