:: Re: [DNG] The real reason I like Li…
Inizio della pagina
Delete this message
Reply to this message
Autore: tom
Data:  
To: dng
Oggetto: Re: [DNG] The real reason I like Linux
On Mon, 16 Mar 2020 21:29:51 -0400
Dan Purgert <dan@???> wrote:

> On Mar 16, 2020, tom wrote:
> > On Mon, 16 Mar 2020 08:38:04 -0400
> > Dan Purgert <dan@???> wrote:
> >
> > > On Mar 15, 2020, tom wrote:
> > > > [...] The biggest technical problem is the
> > > > lack of ASIC northbridge, or rather something to interface the
> > > > CPU to an PCIE bus. Currently the best thing available you can
> > > > get is an FPGA and it is a severe bandwidth bottleneck. It's
> > > > also super expensive getting an FPGA that beefy enough. I don't
> > > > see RISCV going anywhere until this is solved except
> > > > microcontroller applications.
> > > >
> > > > The second problem is patents that prevent RISCV developers from
> > > > implementing a lot of popular specs and standards. Just as an
> > > > example look at the licensing cost of implementing HDMI vs
> > > > DisplayPort.
> > >
> > > On the one hand, I understand why a "large market audience" device
> > > would need HDMI or DisplayPort or the newest whizbang 256K DNA
> > > ("Direct Neural Attachment") adapter is ... but why does that
> > > need to be on a small-market / hobby computer?
> > >
> > > I can only speak for myself, but a reasonably open PC at the $400
> > > mark would certainly be competitive to dell or hp; even if it were
> > > "limited" in the peripheral interconnect area (assuming, of
> > > course, the motherboard's peripheral layout were well documented
> > > and people were encouraged to make stuff -- see arduino or rpi
> > > expansion boards )
> > >
> >
> > generally you want to be able to attach a video card or high
> > performance disk controller to a PCIE slot. you /can/ do these
> > things with an FPGA but I wouldn't call it very reliable. You do
> > too many things or send too much data over the bus it exceeds the
> > bandwidth and the system locks up needing a reset.
>
> I think I wasn't clear enough then. For the sake of discussion, let's
> say PCIe is off the table.
>
> What, then, is so bad about PCI? Or hell, even ISA?
>
> Sure, it's super-limiting in terms of what you can buy off the shelf
> -- but then again, so was the "compatible with Arduino(tm)" market
> 5-10 years ago (and now look at that mess!)
>
> I guess what I'm trying to ask is what would be so bad about a "RISC-V
> Hobby Linux Machine(tm)" only offering these "older" peripheral
> connectivity interfaces in interests of being inexpensive and also
> preserving end-user freedom?
>
> Or ... maybe I'm just a bit crazier than I thought.
>


Nothing wrong with that in of itself, it's just a bit odd to pair a
RISCV cpu with only slow peripherals. Now there are RISCV
microcontrollers you can buy that don't have an MMU so you can't run
Linux on them, and they have all the IO you'd come to expect on a
microcontroller, however if you want to run Linux or some other similar
multitasking OS like NetBSD it probably makes a lot more sense to
choose something aarch64 based purely do to how good and mature
aarch64's IO capabilities are.

There is a lot more out there for aarch64 Linux than just the raspberry
Pi computers. Look at Pine64, Beaglebones, and go up further to things
like the Ampere Emag. My personal fav is the RK3399 cpu. very good
all-rounder, lots of IO and hardware offload.

But back to your question why not RISCV with slow IO. That's like of
like pairing a Ferrari with a horse trailer. There are much better
options out there for pulling horse trailers than sports cars.



-- 
 _________________________________________ 
/ Most people exhibit what political      \

| scientists call "the conservatism of    |
| the peasantry." Don't lose what you've  |
| got. Don't change. Don't take a chance, |
| because you might end up starving to    |
| death. Play it safe. Buy just as much   |
| as you need. Don't waste time.          |
|                                         |
| When we think about risk, human beings  |
| and corporations realize in their heads |
| that risks are necessary to grow, to    |
| survive. But when it comes down to      |
| keeping good people when the crunch     |
| comes, or investing money in something  |
| untried, only the brave reach deep into |
| their pockets and play the game as it   |
| must be played.                         |
|                                         |
| - David Lammers, "Yakitori", Electronic |

\ Engineering Times, January 18, 1988     /
 ----------------------------------------- 
\
 \
   /\   /\   
  //\\_//\\     ____
  \_     _/    /   /
   / * * \    /^^^]
   \_\O/_/    [   ]
    /   \_    [   /
    \     \_  /  /
     [ [ /  \/ _/
    _[ [ \  /_/