From: r...@clark.net (Ryan Rafferty)
Subject: Time for a 64-bit LINUX
Date: 1995/07/03
Message-ID: <3t9218$4re@clarknet.clark.net>#1/1
X-Deja-AN: 105625466
content-type: TEXT/PLAIN; charset=ISO-8859-1
organization: Clark Internet Services, Inc., Ellicott City, MD USA
mime-version: 1.0
newsgroups: comp.os.linux.misc
since 1992, Linux was the only OS to fully take advantage of the 386/486
32-bit capability (I don't really count Windows). The 386 was a
technology that had been around since 1987 at least; it took 5-YEARS for
a decent OS to come out that handled the new 32-bit capability.
Now, however, the limit has been pushed up again, by the 64-bit
capability of the Pentium. At the same time, while Linux fulfilled its role
for 32-bit
Intel machines beautifully, it is being ported to all sorts of new,
powerful platforms like Suns, ALPHAs,and PowerPC's, all of which are more
powerful than the original i386 platform Linux was designed for. And
because the versions of Linux for these other platforms has been designed
to take full advantage of their respective capacities, I feel the Intel
version of Linux may be in danger of gradually becoming obselete even as
Linux for other architectures come to the fore!
Thus, I ask of the Linux community wether or not it is a good idea to
begin a new project for Linux--creating a 64-bit version of Linux for the
new line of Intel platforms. This new version would possibly cause a
rift--binaries from 32-bit Linux would be incompatible with the new
64-bit version, though any source code could be recompiled. Also,
the original intent of Linux was to provide an advanced OS that
would work with any 386 or better. However, the
advantage would be in maintaining the vitality of the Linux project as a
whole. The new version would take full advantage of P5 piping and 64-bit
addressing capabilities, and anything else the Pentium can do that I
don't know of (but would be willing to learn). This would keep Linux on
the cutting edge of techonology, as it was when the 486 was the top of
the line.
Please let me know how you feel about this proposal. If the primary goal
should be to provide binary backwards-compatibility, then Linux is fine
as it is on its current path. If people feel there is a need for a Linux
upgrade, then I would be willing to coordinate the project (it would be a
GREAT learning experience).
Ryan Rafferty
r...@takara.clark.net
r...@clark.net
From: tdgil...@best.com (Tim Gilman)
Subject: Re: Time for a 64-bit LINUX
Date: 1995/07/03
Message-ID: <3t9gs1$1i2@shell1.best.com>#1/1
X-Deja-AN: 105625520
distribution: world
references: <3t9218$4re@clarknet.clark.net> <3t9e8b$dcn@ixnews2.ix.netcom.com>
organization: Best Internet Communications, Inc. (i...@best.com)
newsgroups: comp.os.linux.misc
L. Donovan Heinz < ldhe...@ix.netcom.com> wrote:
>Last I checked, the Pentium doesn't add much from a programming point
>of view. No 64-bit registers or operations, at least. It's just
>faster. So what is wrong with the Linux that we've got?
And you'd be better off getting the PowerPC RISC-based chip to work
with Linux if you were going to go that route. Some people were working
on it, but since the Linux code is closely optimized for a 32-bit Intel
chip, I think the effort basically fizzled and they gave up.
--
Tim D. Gilman < tdgil...@best.com> < tdgil...@ce.berkeley.edu>
http://www.best.com/~tdgilman
.fvwmrc Archive - ftp://ftp.best.com/pub/tdgilman/Fvwmrcs
From: torva...@cc.Helsinki.FI (Linus Torvalds)
Subject: Re: Time for a 64-bit LINUX
Date: 1995/07/04
Message-ID: <3taj9q$l5v@kruuna.helsinki.fi>#1/1
X-Deja-AN: 105625854
sender: torva...@cc.helsinki.fi
references: <3t9218$4re@clarknet.clark.net> <3t9e8b$dcn@ixnews2.ix.netcom.com>
<3t9gs1$1i2@shell1.best.com>
content-type: text/plain; charset=ISO-8859-1
organization: University of Helsinki
mime-version: 1.0
newsgroups: comp.os.linux.misc
In article <3t9gs1$...@shell1.best.com>, Tim Gilman < tdgil...@best.com> wrote:
>L. Donovan Heinz < ldhe...@ix.netcom.com> wrote:
>>Last I checked, the Pentium doesn't add much from a programming point
>>of view. No 64-bit registers or operations, at least. It's just
>>faster. So what is wrong with the Linux that we've got?
>
>And you'd be better off getting the PowerPC RISC-based chip to work
>with Linux if you were going to go that route. Some people were working
>on it, but since the Linux code is closely optimized for a 32-bit Intel
>chip, I think the effort basically fizzled and they gave up.
As others have said, the Pentium is indeed not a 64 bit chip, even
though intel tried to do some PR by claiming it was (it has a 64-bit
external interface to fill in the cache quicker, that's all, I do
believe).
But if you want to go 64 bits, don't go for the PowerPC: it's still 32
bit (well, maybe the 620 is out now, but even so the PowerPC linux will
be only 32-bit due to that being the only mode supported by all PPC
chips).
If you want to go 64 bit, get the DEC Alpha chip: linux for it already
exists, and it's fully 64-bit clean. Linux has been cleaned up a lot
the last year or two, and it's not tied to the intel architecture any
more.
Linus
From: rmo...@bach.seattleu.edu (Robert Mobbs)
Subject: Re: Time for a 64-bit LINUX
Date: 1995/07/10
Message-ID: <3tqiau$gll@dns1.seattleu.edu>#1/1
X-Deja-AN: 105906874
distribution: world
references: <3t9218$4re@clarknet.clark.net> <3t9e8b$dcn@ixnews2.ix.netcom.com>
<3t9gs1$1i2@shell1.best.com>
organization: Seattle University
newsgroups: comp.os.linux.misc
Tim Gilman (tdgil...@best.com) wrote:
: And you'd be better off getting the PowerPC RISC-based chip to work
: with Linux if you were going to go that route. Some people were working
: on it, but since the Linux code is closely optimized for a 32-bit Intel
: chip, I think the effort basically fizzled and they gave up.
Precisely. That's the problem with this field -- everything written
risks the danger of becoming obsolete. Therefore, what we do need is a new
OS for the RISC processors. Once a standard comes out, I'm certain there
will be some clever young person like Linus who will create a brand new
form of home-Unix, and widen the freeware market. And, six years later,
probably the same thing will happen again.
Robert Mobbs
rmo...@seattleu.edu
From: r...@planetx.com (root)
Subject: Re: Time for a 64-bit LINUX
Date: 1995/07/11
Message-ID: <3ttv4m$bek@mars.earthlink.net>#1/1
X-Deja-AN: 106006083
distribution: world
references: <3t9218$4re@clarknet.clark.net> <3t9e8b$dcn@ixnews2.ix.netcom.com>
<3t9gs1$1i2@shell1.best.com> <3tqiau$gll@dns1.seattleu.edu>
organization: Planet-Xware Systems/Solutions
reply-to: rept...@earthlink.net
newsgroups: comp.os.linux.misc
Robert Mobbs (rmo...@bach.seattleu.edu) wrote:
: Tim Gilman (tdgil...@best.com) wrote:
: : And you'd be better off getting the PowerPC RISC-based chip to work
: : with Linux if you were going to go that route. Some people were working
: : on it, but since the Linux code is closely optimized for a 32-bit Intel
: : chip, I think the effort basically fizzled and they gave up.
: Precisely. That's the problem with this field -- everything written
: risks the danger of becoming obsolete. Therefore, what we do need is a new
: OS for the RISC processors. Once a standard comes out, I'm certain there
: will be some clever young person like Linus who will create a brand new
: form of home-Unix, and widen the freeware market. And, six years later,
: probably the same thing will happen again.
: Robert Mobbs
: rmo...@seattleu.edu
DEC is fully supporting Linux. This was announced at Last years DECAS
convention. They gave Linus Torvalds a system with the new ALPHA CPU to
port Linux. This will be a full blown 64-bit OS. For more complete info
grab a copy of the July Linux Journal. They may have some info at
http://www.digital.com but I cannot remember. As soon as Linux is ported
to the ALPHA, I am getting one!
Dave - rept...@earthlink.net
From: torva...@cc.Helsinki.FI (Linus Torvalds)
Subject: Re: Time for a 64-bit LINUX
Date: 1995/07/12
Message-ID: <3tvqq8$dba@kruuna.helsinki.fi>#1/1
X-Deja-AN: 106006394
sender: torva...@cc.helsinki.fi
references: <3t9218$4re@clarknet.clark.net> <3t9gs1$1i2@shell1.best.com>
<3tqiau$gll@dns1.seattleu.edu> <3ttv4m$bek@mars.earthlink.net>
content-type: text/plain; charset=ISO-8859-1
organization: University of Helsinki
mime-version: 1.0
newsgroups: comp.os.linux.misc
In article <3ttv4m$...@mars.earthlink.net>, root < rept...@earthlink.net> wrote:
>: Tim Gilman (tdgil...@best.com) wrote:
>: : And you'd be better off getting the PowerPC RISC-based chip to work
>: : with Linux if you were going to go that route. Some people were working
>: : on it, but since the Linux code is closely optimized for a 32-bit Intel
>: : chip, I think the effort basically fizzled and they gave up.
Linux is no longer tied to the i386: the reason the PPC port has had
lots of problems is the lack of documentation on the PowerMacs and the
lack of machinery for anything else. We'll get a PPC linux eventually,
but it will probably be at the end of this year at the earliest (and
that without X11 etc).
> DEC is fully supporting Linux. This was announced at Last years DECAS
>convention. They gave Linus Torvalds a system with the new ALPHA CPU to
>port Linux. This will be a full blown 64-bit OS. For more complete info
>grab a copy of the July Linux Journal. They may have some info at
>http://www.digital.com but I cannot remember. As soon as Linux is ported
>to the ALPHA, I am getting one!
My alpha (and others) already do run linux. The major lack right now is
that there is no good and easy distribution (but people inside DEC has
done some of the work already), and there is no X server yet. And to be
quite frank, the axp-linux hasn't gotten nearly the amount of testing
that the normal x86-linux has, so I expect that it won't be quite that
stable for a while.
That said, Linux/axp already runs most OSF/1 binaries, and as of this
week the networking also works so I can telnet out from it (I haven't
installed any inetd/telnetd etc yet, so I can't telnet in yet). The
worst problem right now is the lack of X: I can't work without X any
longer, so I still use OSF/1 for Linux/axp development.
Linus