Interesting while I suppose the increased Rpi4 microSD card performances should be enough for any o.s. considering they ran their Debian verion of linux with much slower bandwidth on any microSD before this version; the idea of the USB3 to external SSD might be faster but not that compact. I don't have the Rpi4 but sure is a interesting while not entirely convincing solution imho having tried every previous versions. There're many SBC too and also some x64 ones that might be much more interesting for retrogaming for sure.

386SX wrote on 2021-07-02, 09:35:

Interesting while I suppose the increased Rpi4 microSD card performances should be enough for any o.s. considering they ran their Debian verion of linux with much slower bandwidth on any microSD before this version; the idea of the USB3 to external SSD might be faster but not that compact. I don't have the Rpi4 but sure is a interesting while not entirely convincing solution imho having tried every previous versions. There're many SBC too and also some x64 ones that might be much more interesting for retrogaming for sure.

Thank you for your reply! ^^ I can't disagree with what you said. I do have a Pi 4 and a the predecessors.
From what I can tell the Pi 3 had the best compatibility and the most modern CPU core so far.
Alas, it wasn't as quick as the Pi 4 and had less memory, also.

The Windows 11 is interesting in so far, I think, because it introduces a change in policy.
Microsoft actually spends some efforts in making both x86/x64 emulation functional and takes the ARM platform seriously.

This kind of reminds me of good old OS/2, which considered itself a "migration platform" at some point if memory serves.
I don't know how things will continue to evolve, but it might be an interesting time. *"Wind of Change" plays*

Simple applications that rely on API calls mainly, but are otherwise not computing-intensive, say IrfanView or Winamp, may run almost as quick as native applications.
Or maybe even slightly faster, if some sort of clever dynamic-recompiling happens along the way.
Anyway, it might be sufficient for Win95 games in terms of performance. *plays Hover* 😉

Also, depending on how Microsofr writes its code, this Windows might be portable once and again.
A RISC-V release of Windows would be interesting, provided that RISC-V manages to compete against ARM.

Anyhow, I'm no fan boy and could be wrong when it comes to Windows.
As an observer, it's just nice to feel a breeze of fresh air in the computing scence! ^^

PS: Here are two videos of running Unigine benchmarks on an M1 Mac (ARM64 based).
The benchmark's exeutable are non-native, ie they used x86 instructions.

Rosetta 2 does "convert" intel binaries before they are executed.
Some could say, Rosetta 2 is like a binary transpiler.

https://www.youtube.com/watch?v=yQga3XLd3rc
https://www.youtube.com/watch?v=i1wxylW0b9Q

Here's also a Windows version (x86) that runs on Windows 10/ARM in Parallels on an M1 Mac.
https://www.youtube.com/watch?v=ZlM96xHGLEs

Jo22 wrote on 2021-07-02, 10:32:

Thank you for your reply! ^^ I can't disagree with what you said. I do have a Pi 4 and a the predecessors. From what I can tell […]
Show full quote

386SX wrote on 2021-07-02, 09:35:

Interesting while I suppose the increased Rpi4 microSD card performances should be enough for any o.s. considering they ran their Debian verion of linux with much slower bandwidth on any microSD before this version; the idea of the USB3 to external SSD might be faster but not that compact. I don't have the Rpi4 but sure is a interesting while not entirely convincing solution imho having tried every previous versions. There're many SBC too and also some x64 ones that might be much more interesting for retrogaming for sure.

Thank you for your reply! ^^ I can't disagree with what you said. I do have a Pi 4 and a the predecessors.
From what I can tell the Pi 3 had the best compatibility and the most modern CPU core so far.
Alas, it wasn't as quick as the Pi 4 and had less memory, also.

The Windows 11 is interesting in so far, I think, because it introduces a change in policy.
Microsoft actually spends some efforts in making both x86/x64 emulation functional and takes the ARM platform seriously.

This kind of reminds me of good old OS/2, which considered itself a "migration platform" at some point if memory serves.
I don't know how things will continue to evolve, but it might be an interesting time. *"Wind of Change" plays*

Simple applications that rely on API calls mainly, but are otherwise not computing-intensive, say IrfanView or Winamp, may run almost as quick as native applications.
Or maybe even slightly faster, if some sort of clever dynamic-recompiling happens along the way.
Anyway, it might be sufficient for Win95 games in terms of performance. *plays Hover* 😉

Also, depending on how Microsofr writes its code, this Windows might be portable once and again.
A RISC-V release of Windows would be interesting, provided that RISC-V manages to compete against ARM.

Anyhow, I'm no fan boy and could be wrong when it comes to Windows.
As an observer, it's just nice to feel a breeze of fresh air in the computing scence! ^^

PS: Here are two videos of running Unigine benchmarks on an M1 Mac (ARM64 based).
The benchmark's exeutable are non-native, ie they used x86 instructions.

Rosetta 2 does "convert" intel binaries before they are executed.
Some could say, Rosetta 2 is like a binary transpiler.

https://www.youtube.com/watch?v=yQga3XLd3rc
https://www.youtube.com/watch?v=i1wxylW0b9Q

Here's also a Windows version (x86) that runs on Windows 10/ARM in Parallels on an M1 Mac.
https://www.youtube.com/watch?v=ZlM96xHGLEs

Sure it's impressive to see the Unigine Valley demo running on an ARM "low power" SoC at such high frame rate! It might not be the best graphic tech demo ever seen but sure is heavy; on my GT610 old video card the passive heatsink couldn't keep the temp low enough even with the case door opened and reaching 110°C easily! I suppose the CPU translation is optimized at best while also not that relevant in a benchmark where the GPU is the real stressed unit used and the API calls might be natively called. There're on other SBC similar sw to run Wine x86 version for example, I tried to run it on a OrangePI board (quad Cortex A7) but I had errors after trying running games. It might be both the Wine version used or the Debian version or the still developed drivers. I heard better results on the Rpi3 SoC.

Sure the M1 SoC is impressive but what impress me more from those videos is its GPU. But also I suppose the wattage is not exactly low as we are used to expect from an ARM SoC and time's passed since the few watts of the early smartphones/Rpi SoC. It'd be more interesting to know that SoC wattage reading while running that bench. 😉

Anyway I suppose the Rpi4 SoC might not be comparable to such proprietary M1 design but also consuming less power and aimed at a low end basic x64 computer replacement while the M1 seems like having higher speed targets and long term expectations.