[ale] Really cool new hardware
jonhall80 at comcast.net
jonhall80 at comcast.net
Mon Jan 18 23:41:35 EST 2021
All,
I have not really inspected the ISA for RISC-V to see if there were any instructions that were "designed for Linux" other than the normal instructions that would be there for almost any modern operating system for things like atomic operations, etc.
Since RISC-V is a RISC computer, I would expect the instructions to be as simple as possible and allow the compilers to do optimization to make the chip run very fast.
On the other hand, as you to up the stack of things necessary for speed, such as good design for cache, data transfer, etc. I would hope that the architects were able to jettison lots of legacy features that are not necessary in a completely new architecture that would not be needed for backward compatibility, and instead build in compatibility with 16-bit, 32-bit, 64-bit and 128-bit instructions.
On the other hand, to talk about a BOARD that is "Linux compatible" I could see lots of things that would make it "Linux friendly", such as hardware interfaces for I/O that would be published and available to software developers, boot code that would have the sources published, having the ability for the kernel to interact with hardware interfaces built into the system, and the ability for hypervisors to work well with the board, again leaving behind some trade-offs done for previous operating systems.
This is not to say that these other operating systems will not change over time to take advantage of this new hardware, it is just that some of the legacy stuff will be left behind.
Just my thoughts on this.
md
> On 01/18/2021 10:05 PM Michael Potter via Ale <ale at ale.org> wrote:
>
>
> I skimmed the article.
>
> I did not see anywhere where it said what made it made for Linux.
>
> Any clue?
>
> I would expect two things....
>
> Some kind of instructions specific for networking.
>
> Instructions specific for fork().
>
>
>
>
> On Sun, Jan 17, 2021 at 11:50 AM Jim Kinney via Ale < ale at ale.org mailto:ale at ale.org > wrote:
>
> > > All true. Spark and Power are (mostly,?)dead and no longer produced outside of industrial gear.
> >
> > I guess my real enthusiasm is for the "designed for Linux" aspect.
> >
> > I don't play games so I'm not sure what runs all the consoles now. I also don't have any beagle toys yet. My electronic skills are really rusty and never were design oriented. I designed and built only one power supply and it was strange, variable current 0-5A at one pin pair, variable voltage 0-1000V on another pin pair, one pin common between the two. Basically a variac driven tube. The 0.5 F capacitor was fun.
> >
> > On January 17, 2021 11:12:55 AM EST, Solomon Peachy < pizza at shaftnet.org mailto:pizza at shaftnet.org > wrote:
> >
> > > > >
> > > On Sun, Jan 17, 2021 at 08:26:18AM -0500, Jim Kinney via Ale wrote:
> > >
> > > > > > > A fully open source cpu design is a game changer.
> > > >
> > > > > > >
> > >
> > > Is it really? SPARC, and POWER have all had fully open ISAs and core
> > > designs since 2005 and 2013, respectively.
> > >
> > > As someone who has done low-level hacking on various architectures over
> > > the years, the actual CPU core/ISA makes very little difference; most of
> > > the headaches are in the rest of the system[-on-chip].
> > >
> > > RISC-V might be fully open from an ISA perspective, but that doesn't
> > > mean that any given SoC built on it has an open source CPU core, or that
> > > any of the various controllers (DMA, display, audio, I2S/I2C/SPI, etc)
> > > or accelerators (video codecs, crypto engines) are open source or even
> > > have documentation available without three-deep NDAs.
> > >
> > > Meanwhile, from the perspective of someone writing software, the
> > > underlying CPU architecture rarely matters if you're not doing low-level
> > > OS/compiler hackery or trying to hand-optimize performance-critical code
> > > that can't be run on a GPU or more specialized accelerator.
> > >
> > >
> > > > > > > I was quite surprised to see fedora is ready to go on riscV. It's
> > > > usually Debian or Ubuntu that's first on the ground for new gear.
> > > >
> > > > > > >
> > >
> > > https://sifivetechsymposium.com/wp-content/uploads/2020/01/Fedora_on_RISC-V_SiFive_BJ_2019.pdf
> > >
> > > TL;DR: Fedora has a very strong "upstream first" mentality, and Red
> > > Hatters contribute heavily to all of the upstream projects that needed
> > > work to enable RISC-V and used Fedora's tooling to work out the kinks.
> > >
> > > ...meanwhile, has Ubuntu _ever_ preceeded Debian when it comes to
> > > architecture/platform support?
> > >
> > > (I don't mean "here's a pre-built image you can put on an SD card to
> > > boot board X and you have to use our special snowflake kernel and
> > > bootloader with binary blobs which preclude getting security updates
> > > from the distribution")
> > >
> > > - Solomon
> > >
> > > > >
> > --
> > Computers amplify human error
> > Super computers are really cool
> > _______________________________________________
> > Ale mailing list
> > Ale at ale.org mailto:Ale at ale.org
> > https://mail.ale.org/mailman/listinfo/ale
> > See JOBS, ANNOUNCE and SCHOOLS lists at
> > http://mail.ale.org/mailman/listinfo
> >
> > >
>
> --
> Michael Potter
> Tapp Solutions, LLC
> http://www.tappsolutions.com
> +1 770 815 6142 ** Atlanta ** michael at potter.name mailto:michael at potter.name ** http://www.linkedin.com/in/michaelpotter
> Schedule a meeting with me: https://calendly.com/michael-potter
> _______________________________________________
> Ale mailing list
> Ale at ale.org
> https://mail.ale.org/mailman/listinfo/ale
> See JOBS, ANNOUNCE and SCHOOLS lists at
> http://mail.ale.org/mailman/listinfo
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.ale.org/pipermail/ale/attachments/20210118/a38d6d72/attachment.html>
More information about the Ale
mailing list