how about a LINUX CHIP?

how about a LINUX CHIP?

Post by Christopher B. Brow » Thu, 22 Oct 1998 04:00:00




Quote:>The sun 'java chip' concept is not really making any headway -
>in other words there is not a finished one anyway - and sun
>aren't going to manufacture it themselves - they're just licensing the
>design.

>Why can't there be a linux chip?  Couldn't the shell be put onto a chip?
>Or maybe a scaled down version of it - like a 'Linux CE' sort of thing?

It would be much more interesting to build a chip optimized to provide the
ring services required to implement Multics.

"All" Linux needs is things like:
- Memory protection
- Hopefully a FPU
which is all provided fairly nicely by many families of CPUs including
IA-32, MIPS, ARM, PPC, Alpha, 68030 (and greater), SPARC.

There have been some rumors that Transmeta's chip may be somewhat "optimized
for Linux."  

Rumor only...  The only thing that appears clear is that they are doing CPU
design...

At any rate, in order to do a "scaled down" Linux system is not particularly
dependent on the CPU, but rather on the overall system.

The fact that IA-32 "accessory hardware" such as motherboard,
SCSI/IDE/Serial/Parallel I/O controllers, and video cards are widely
available deployed in the "Windoze world" results in their being cheap.

In contrast, a StrongARM-based system may use a dirt-cheap CPU, but since
you don't have 15 vendors hawking motherboards, the secondary hardware is
much less available and relatively more costly.

Design and build an 8-chip SMP motherboard using $25 MIPS CPUs that uses PCI
and sells for under $500 and you'll see SMP MIPS get popular real quick now.

Design a "Linux chip" that requires a custom-built motherboard and nobody
will care...

--
Those who do not understand Unix are condemned to reinvent it, poorly.  
-- Henry Spencer          <http://www.hex.net/~cbbrowne/lsf.html>

 
 
 

how about a LINUX CHIP?

Post by Darren Steve » Fri, 23 Oct 1998 04:00:00




> >The sun 'java chip' concept is not really making any headway -
> >in other words there is not a finished one anyway - and sun
> >aren't going to manufacture it themselves - they're just licensing the
> >design.

> >Why can't there be a linux chip?  Couldn't the shell be put onto a chip?
> >Or maybe a scaled down version of it - like a 'Linux CE' sort of thing?

I think the Java Chip is a microSparc with mods to allow it to execute Java Byte
code directly (or very efficiently). It does not implement to Java OS, just the raw
Virtual Machine.

Linux is an Operating system, you may be able to design a CPU that has features
that would make implementation of Linux easy (I have no idea what you need other
than what is available on modern CPU's - VM, protection, etc...)

Darren Steven

UNIX Administrator

Australian Maritime College
Information Technology Services Department
Newnham Campus
Launceston, Tasmania, Australia

Phone (03) 6335 4842



 
 
 

how about a LINUX CHIP?

Post by cball.. » Fri, 23 Oct 1998 04:00:00



Quote:> I think the Java Chip is a microSparc with mods to allow it to execute Java Byte
> code directly (or very efficiently). It does not implement to Java OS, just the raw
> Virtual Machine.

What would be very cool would be to port Linux to a Java VM so that when
the Java Chip comes out, it will have a good OS to run :)  I thought once
of trying this, but then decided it would be crazy to spend that much time
on it...   I would help out if anyone wanted to do it, though...

--
Chris Ballard              
Systems Administrator

 
 
 

how about a LINUX CHIP?

Post by Chong Liang Oo » Sat, 24 Oct 1998 04:00:00


    I think the Java chip will be too slow to get anyone interested in it (other than
embeded market).

-Chong



> > I think the Java Chip is a microSparc with mods to allow it to execute Java Byte
> > code directly (or very efficiently). It does not implement to Java OS, just the raw
> > Virtual Machine.

> What would be very cool would be to port Linux to a Java VM so that when
> the Java Chip comes out, it will have a good OS to run :)  I thought once
> of trying this, but then decided it would be crazy to spend that much time
> on it...   I would help out if anyone wanted to do it, though...

> --
> Chris Ballard
> Systems Administrator


 
 
 

how about a LINUX CHIP?

Post by Richard Horto » Sat, 24 Oct 1998 04:00:00


The Java chip too slow???

I thought the whole point (or most of the point - prehaps three quaters
of the circle ;> ) was that a chip executing java byte codes natively
would be faster than your WinTel chip doing the same thing via VM... a
Java chip won't need a virtual machine - it is a none virtual - virtual
machine (ie you can physically hit the thing when it crashes)

And as to Chris suggesting porting Linux to a chip  -- a good idea but
define how much of linux you intend to base on the chip and how much in
loadable rom form...


>     I think the Java chip will be too slow to get anyone interested in it (other than
> embeded market).

> -Chong



> > > I think the Java Chip is a microSparc with mods to allow it to execute Java Byte
> > > code directly (or very efficiently). It does not implement to Java OS, just the raw
> > > Virtual Machine.

> > What would be very cool would be to port Linux to a Java VM so that when
> > the Java Chip comes out, it will have a good OS to run :)  I thought once
> > of trying this, but then decided it would be crazy to spend that much time
> > on it...   I would help out if anyone wanted to do it, though...

> > --
> > Chris Ballard
> > Systems Administrator


--
"In world without fences who needs Gates?"

My opinions - when I have any - are my own and unless stated explicitly
in no way represent
those of my employers, friends, collegues, enemies.

Richard Horton.                 Graduate Software Engineer GEC-Easams.

 
 
 

how about a LINUX CHIP?

Post by bill davids » Sat, 24 Oct 1998 04:00:00




|

|

| > >The sun 'java chip' concept is not really making any headway -
| > >in other words there is not a finished one anyway - and sun
| > >aren't going to manufacture it themselves - they're just licensing the
| > >design.
| > >
| > >Why can't there be a linux chip?  Couldn't the shell be put onto a chip?
| > >Or maybe a scaled down version of it - like a 'Linux CE' sort of thing?
| >
|
| I think the Java Chip is a microSparc with mods to allow it to execute Java Byte
| code directly (or very efficiently). It does not implement to Java OS, just the raw
| Virtual Machine.

Once upon a time there was a project at UCSD called "UCSD Pascal." It
was a compiler which compiled to tokens which could be executed on any
machine by a virtual machine. Then front end FORTRAN and Ada compilers
were written to the same VM. Then a hardware chip (from WD I believe)
was designed to execute the tokens directly, and compilers were written
to compile the tokens into machine code. Mid 1980's.

Once upon a time there was a project at MIT to make a small subset of
the BCPL language which compiled to tokens. These tokens were then
executed on a VM on any hardware... soon people wrote compilers for the
tokens to generate machine code. Late 1960's.

History repeats itself. And the subset of BCPL was called B, which was
later rethought as C, which looked like a good language for writing an
operating system in a high level language, which was called UNIX. I
believe the first o/s written in hll which was not just a research
project was MULTICS, written in PL/1.
--

Never underestimate the power of stupid people in large groups

 
 
 

how about a LINUX CHIP?

Post by bill davids » Sat, 24 Oct 1998 04:00:00




| The Java chip too slow???
|
| I thought the whole point (or most of the point - prehaps three quaters
| of the circle ;> ) was that a chip executing java byte codes natively
| would be faster than your WinTel chip doing the same thing via VM... a
| Java chip won't need a virtual machine - it is a none virtual - virtual
| machine (ie you can physically hit the thing when it crashes)

A proper VM will execute only slightly slower than a nativce machine. I
see no reason to think that hardware will buy much, and the cost of
going to new technology for speed is high. The idea of an embedded chip
is possible, but with chip densities, I think you could do a 386 with VM
firmware at very low cost, with a cheap RISC chip, like 29000, for high
end, if there is one.

The 68000 is another candidate. It's old, but Motorola did a IBM360 in
firmware back in the 80's, I'm sure JAVA is easier ;-)

I don't think a JAVA chip from scratch can combine fast and cheap, and
cheap is more import for most embedded applications. Obviously there are
exceptions, consider it said.
--

Never underestimate the power of stupid people in large groups

 
 
 

how about a LINUX CHIP?

Post by Alexander Vi » Sat, 24 Oct 1998 04:00:00





>>The sun 'java chip' concept is not really making any headway -
>>in other words there is not a finished one anyway - and sun
>>aren't going to manufacture it themselves - they're just licensing the
>>design.

>>Why can't there be a linux chip?  Couldn't the shell be put onto a chip?
>>Or maybe a scaled down version of it - like a 'Linux CE' sort of thing?

>It would be much more interesting to build a chip optimized to provide the
>ring services required to implement Multics.

>"All" Linux needs is things like:
>- Memory protection
>- Hopefully a FPU
>which is all provided fairly nicely by many families of CPUs including
>IA-32, MIPS, ARM, PPC, Alpha, 68030 (and greater), SPARC.

        There is one thing that would be _very_ nice: MMU able to keep
several contexts at the same time. That is, all you get in user mode
is one of them, but in the priv mode you can use additional commands
able to address things from any context. That would mean that you
don't have to reload the context on each c/s, only when you are switching
to something currently not loaded. Think of 8 contexts simultaneously.
That would make 90% of context switches almost zero-cost. Add cache
colouring, indeed. Sigh... If XKL would be cheaper... OTOH I suspect
that we are _not_ 36-bit clean, so XKL would be _very_ painful.

--
"You're one of those condescending Unix computer users!"
"Here's a nickel, kid.  Get yourself a better computer" - Dilbert.

 
 
 

how about a LINUX CHIP?

Post by William McBri » Mon, 26 Oct 1998 03:00:00


Well, it's not quite a "Linux Chip", but here's a related idea I've just
heard about on Fidonet. Has anyone else heard of this?

==========================================================================
 System: Doc's Place Bbs Online.
   Area: UNIX
   Date: 10-22-98 14:58
   From: MATHIEU BOUCHARD
     To: WILLIAM MCBRINE
   Subj: FreeCPU?
--------------------------------------------------------------------------
 MB>> year 2000 are projected the "Merced" and "FreeCPU" architectures.
 WM>                                            ^^^^^^^
 WM> What is this?

A CPU whose plans are protected by the GPL, and made by a team of
hackers over the Net, just like Linux. 10000 units will be produced and
it will be sold 100$US each in 2000, while Merced's projected price
will be 5000$.

Its speed will be roughly equivalent to Merced except it will lack
floating point instructions. It uses a radically new kind of RISC; it
has no general-purpose registers: RAM is directly referenced at every
instruction, because registers are not important since the apparition
of internal cache. The result is you don't need MOV instructions (also
known as load (LD) and store (ST) for Motorolish people out there), and
because of this, compilers will be simpler and object code will be
smaller than other VVLIW (fixed 64-bit instruction size) processors.
Needless to say that it's not compatible PC.

Projected OS'es: Linux. However, any kernel compilable with GCC will be
very easy to port. The chip will fit on PC motherboards. (Socket six or
seven I think)

My Prediction: If they can drag some attention in 2000/2001, some
company will sell chips of that design independently from the FreeCPU
team, if Linux's user base outgrows Mac's user base as expected.

matju

--- Terminate 4.00/Pro
 * Origin: The Lost Remains Of SatelliteSoft BBS (1:163/215.42)

--
William McBrine    | http://www.clark.net/~wmcbrine/

 
 
 

how about a LINUX CHIP?

Post by Ivan Sumne » Mon, 26 Oct 1998 03:00:00


This is not a new idea.  TI designed its micro processors in the late 70's
without registers, saying that it made sense to put as much memory on chip
as
possible and software should not need to know exactly where the boundary
between "register" memory and ordinary"slower" memory was, so it would be
compatible on whole families of processors.


> Well, it's not quite a "Linux Chip", but here's a related idea I've just
> heard about on Fidonet. Has anyone else heard of this?

> ==========================================================================
>  System: Doc's Place Bbs Online.
>    Area: UNIX
>    Date: 10-22-98 14:58
>    From: MATHIEU BOUCHARD
>      To: WILLIAM MCBRINE
>    Subj: FreeCPU?
> --------------------------------------------------------------------------
>  MB>> year 2000 are projected the "Merced" and "FreeCPU" architectures.
>  WM>                                            ^^^^^^^
>  WM> What is this?

> A CPU whose plans are protected by the GPL, and made by a team of
> hackers over the Net, just like Linux. 10000 units will be produced and
> it will be sold 100$US each in 2000, while Merced's projected price
> will be 5000$.

> Its speed will be roughly equivalent to Merced except it will lack
> floating point instructions. It uses a radically new kind of RISC; it
> has no general-purpose registers: RAM is directly referenced at every
> instruction, because registers are not important since the apparition
> of internal cache. The result is you don't need MOV instructions (also
> known as load (LD) and store (ST) for Motorolish people out there), and
> because of this, compilers will be simpler and object code will be
> smaller than other VVLIW (fixed 64-bit instruction size) processors.
> Needless to say that it's not compatible PC.

> Projected OS'es: Linux. However, any kernel compilable with GCC will be
> very easy to port. The chip will fit on PC motherboards. (Socket six or
> seven I think)

> My Prediction: If they can drag some attention in 2000/2001, some
> company will sell chips of that design independently from the FreeCPU
> team, if Linux's user base outgrows Mac's user base as expected.

> matju

> --- Terminate 4.00/Pro
>  * Origin: The Lost Remains Of SatelliteSoft BBS (1:163/215.42)

> --
> William McBrine    | http://www.clark.net/~wmcbrine/


--
==========================================================
                              R. Ivan Sumner                E-Mail:

University of Kentucky        Office Ph.: (606) 257-9348
Department of Computer Science
141 ASTeCC                    Home page:  http://sac.uky.edu/~risumn0
Lexington, KY  40506
 
 
 

how about a LINUX CHIP?

Post by Ivan Sumne » Tue, 27 Oct 1998 04:00:00


This is not a new idea.  TI designed its micro processors in the late 70's
without registers, saying that it made sense to put as much memory on chip as
possible and software should not need to know exactly where the boundary
between "register" memory and ordinary"slower" memory was, so it would be
compatible on whole families of processors.


> Well, it's not quite a "Linux Chip", but here's a related idea I've just
> heard about on Fidonet. Has anyone else heard of this?

> ==========================================================================
>  System: Doc's Place Bbs Online.
>    Area: UNIX
>    Date: 10-22-98 14:58
>    From: MATHIEU BOUCHARD
>      To: WILLIAM MCBRINE
>    Subj: FreeCPU?
> --------------------------------------------------------------------------
>  MB>> year 2000 are projected the "Merced" and "FreeCPU" architectures.
>  WM>                                            ^^^^^^^
>  WM> What is this?

> A CPU whose plans are protected by the GPL, and made by a team of
> hackers over the Net, just like Linux. 10000 units will be produced and
> it will be sold 100$US each in 2000, while Merced's projected price
> will be 5000$.

> Its speed will be roughly equivalent to Merced except it will lack
> floating point instructions. It uses a radically new kind of RISC; it
> has no general-purpose registers: RAM is directly referenced at every
> instruction, because registers are not important since the apparition
> of internal cache. The result is you don't need MOV instructions (also
> known as load (LD) and store (ST) for Motorolish people out there), and
> because of this, compilers will be simpler and object code will be
> smaller than other VVLIW (fixed 64-bit instruction size) processors.
> Needless to say that it's not compatible PC.

> Projected OS'es: Linux. However, any kernel compilable with GCC will be
> very easy to port. The chip will fit on PC motherboards. (Socket six or
> seven I think)

> My Prediction: If they can drag some attention in 2000/2001, some
> company will sell chips of that design independently from the FreeCPU
> team, if Linux's user base outgrows Mac's user base as expected.

> matju

> --- Terminate 4.00/Pro
>  * Origin: The Lost Remains Of SatelliteSoft BBS (1:163/215.42)

> --
> William McBrine    | http://www.clark.net/~wmcbrine/


--
==========================================================
                              R. Ivan Sumner                E-Mail:

University of Kentucky        Office Ph.: (606) 257-9348
Department of Computer Science
141 ASTeCC                    Home page:  http://sac.uky.edu/~risumn0
Lexington, KY  40506
 
 
 

how about a LINUX CHIP?

Post by Christopher Brow » Wed, 28 Oct 1998 04:00:00



Quote:>Well, it's not quite a "Linux Chip", but here's a related idea I've just
>heard about on Fidonet. Has anyone else heard of this?

Material elided...

It only will "matter" if it can be deployed *IN COMPLETE SYSTEMS* at
interestingly low prices.

A system is not solely comprised of a CPU; it also requires motherboard,
memory, and various other I/O circuitry.

In order for the project to be of any interest, that set of other
hardware must *also* be cheap and fast.  Of which I am somewhat more
skeptical...

--
quit   When the quit statement is read, the  bc  processor
       is  terminated, regardless of where the quit state-
       ment is found.  For example, "if  (0  ==  1)  quit"
       will cause bc to terminate.
(Seen in the manpage for "bc". Note the "if" statement's logic)

 
 
 

how about a LINUX CHIP?

Post by Richard Jone » Thu, 29 Oct 1998 04:00:00


: A proper VM will execute only slightly slower than a nativce machine. I
: see no reason to think that hardware will buy much, and the cost of
: going to new technology for speed is high.

Not necessarily true.

Current RISC chips [1] are optimized to run C and C-like
languages very fast. Of course, that's because C is probably
the most frequently used language judging by the amount of
time a chip spends running it. If you believe that Java,
Smalltalk, Haskell or another language could actually become
common, then you need to rethink your assumptions and may
need to add some primitives to optimize, for example,
garbage collection or functional evaluation or typed
pointers. This obviously requires a leap of faith - none of
these languages look like they might overtake C (or C++) any
time soon, which is a shame :-(

Rich.

[1] I say ``RISC'' chips to exclude x86. What's this
optimized for? It's anyone's guess ...

--
-      Richard Jones. Linux contractor London and SE areas.        -
-    Very boring homepage at: http://www.annexia.demon.co.uk/      -
- You are currently the 1,991,243,100th visitor to this signature. -
-    Original message content Copyright (C) 1998 Richard Jones.    -

 
 
 

how about a LINUX CHIP?

Post by Richard Jone » Thu, 29 Oct 1998 04:00:00


:       There is one thing that would be _very_ nice: MMU able to keep
: several contexts at the same time. That is, all you get in user mode
: is one of them, but in the priv mode you can use additional commands
: able to address things from any context. That would mean that you
: don't have to reload the context on each c/s, only when you are switching
: to something currently not loaded. Think of 8 contexts simultaneously.
[...]

NO! Fallacy!

Please read Hennessy & Patterson: ``Computer Architecture -
A Quantitative Approach'' and rethink your statement. While
you may well reduce your context switch time to zero, your
scheme will certainly increase the clock cycle time, resulting
in an overall increase in runtime for all but the most
I/O intensive application.

I wish the well-meaning people doing F-CPU would read this
book too ...

Rich.

--
-      Richard Jones. Linux contractor London and SE areas.        -
-    Very boring homepage at: http://www.annexia.demon.co.uk/      -
- You are currently the 1,991,243,100th visitor to this signature. -
-    Original message content Copyright (C) 1998 Richard Jones.    -

 
 
 

1. chip-0.2 / driver for MARALU chip-card-reader/writer v1.0

I have placed chip02.tgz on
  ftp.thp.uni-koeln.de(134.95.64.1):/pub/linux/chip/chip02.tgz

It contains an loadable device driver and some sample application to be
used with MARALU's chip-card-reader/writer.

The reader/writer can access I2C-cards and other formats.


If you don't have ftp-access, drop me a note and I'll mail the code to
you (in uuencoded form, about 35k size, if nothing different is specified).

Martin

PLEASE NOTE, THAT NEITHER I NOR MARALU TAKE ANY RESPONSIBILITY FOR THE
FUNCTIONALITY OR RELIABILITY OF THE PROVIDED CODE AND INFORMATION.

--

Be sure to include Keywords: and a short description of your software.

2. JDK 1.1.7A on RH5.2- hangs when running GUI apps

3. Chips & Tech 65550 Grapics Chip!

4. Install on Toshiba laptop

5. Chips & Tech 65555 video chip supported?

6. I don't understand Linux graphics modes!

7. Toshiba PC-100; 4 chip vs 8 chip?

8. linux to win2000 gateway

9. XF86Setup - Epson ActionNote 650C - Chips&Tech65440 chip set?

10. How to setup Sound for ES1948F chip (also known as Maestro-1 chip)

11. SVGA Driver for Chips and Technology 452 chip set

12. Chips & Tech 65550 Grapics Chip!

13. Sound chips on PC-Chips boards