[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Proposal for new architecture support/distribution



-----BEGIN PGP SIGNED MESSAGE-----

On 1 Feb 1999, Jim Pick wrote:

> Wouldn't that make more sense as a subarchitecture of the PowerPC
> port.  I gather that the userspace component would be the same.  You'd
> just need work on the kernel and installation process.  Or are the
> instruction sets somehow incompatible?

My apologies for replying out of order and a bit slowly; I'm still sorting
out several small bugs (just took my x86 box to slink, and managed to
break much in the process! *GRIN*;) and things like that.

Perhaps it would, but there are some serious differences that can be named
offhand. Here's a quick reference table. Pardon my lack of ASCII skills.
;)

Feature             | Apple & Clones PPC  | RS/6000 PPC
- --------------------+---------------------+------------
ISA bus             | No                  | Yes
PCI 2.1 bus         | Very Few            | Yes
PCI 2.0 bus         | Very Few            | Yes
MCA 80/160 bus      | No                  | Yes
Onboard SCSI        | Pre-G3 Only (Apple) | Yes (IBM)
Onboard Sound       | Yes (Apple)         | Yes (IBM)
Tablet Port         | No                  | Yes (IBM)
Parallel Port       | Most                | Yes
Standard Serial Port| No                  | Yes
SMP Hardware        | *VERY* few          | Most
Maximum SMP CPUs    | 2                   | 12 (S70 AS)
Works with PC PCI   | No                  | Yes
Works with PC ISA   | No                  | Yes

As you can see, there's a lot of things that IBM has, that Apple doesn't.
Or does differently. There are many things that a PC (PC being defined as
an x86/IA-32 system) can use, such as pciutils, that an RS/6000 will
likely also be able to use, that an Apple or clone will not be able to
use. You wonder why MacOS doesn't run on RS/6000's? That's why. ;) 

I honestly can't make any accurate guesses at just how much x86 kernel
code, which is the current *core* problem next to hardware, will easily
transfer to an RS/6000. My bet is that the majority of PCI 2.(0,1) code,
ISA, and serial will work. However, we are still limited to 8 processors,
due to kernel limitations, we also do not have support for the onboard
SCSI or the tablet port. (Which I have not been able to find much
information on, incidentally. Possibly an obsolete feature that is too
difficult to remove?)

So, yes, the instruction sets are for the most part, entirely compatible.
However, the second tier hardware that is driven by that instruction set,
is not entirely compatible. It does work in some systems, but not entirely
reliably, or effectively.

*grin* I'm seriously considering throwing up a little webpage with an
itemized list, by priority, of the work that would need to be done. :)

- -Phillip R. Jaenke (prj@nls.net | InterNIC: PRJ5)
 "Look. It works this way." "Why?" "Because the designer said so."
 "Why?" "Because the designer is a moron. Let's fix it." --anon.


-----BEGIN PGP SIGNATURE-----
Version: 2.6.3a
Charset: noconv

iQCVAwUBNrYhkztUbefKjv6dAQFKggQAkCt/1mRDMOaC6clIaRb60tuh5d8Q1UG7
UCP6OQOZ2mW4OpEG8yMDP+2C8WV5uIl5WVCM+Ne/0aOubX/MyFpmjteYH2+GDZqX
oWxgnZKBT/2IS7xlxV2KoRWUF7HIZ8PdagKZ4qkYGwS2YFf1ahuEskFkXkTZHtXX
XrPQH6XXTJM=
=8KGo
-----END PGP SIGNATURE-----


Reply to: