Re: [OFFTOPIC] Re: [bigmem-patch] 4GB with Linux on IA32

Scott Lurndal (slurn@griffin.engr.sgi.com)
Fri, 20 Aug 1999 15:15:13 -0700 (PDT)

> Severely shortened
>
> On Fri, 20 Aug 1999, Scott Lurndal wrote:
> > > On Fri, 20 Aug 1999, Thierry Vignaud wrote:
> > > Power(maybe not the recent ones). So what. The linux hobbyist can run on
> > > a cheap 4GB or less intel, if you need a professional system, just use
> > > another architecture. As all uses plain C(++) paradigm and Unix, it's not
> >
> >
> > Man O' Man!!!! You really can't mean this. You are relegating
> > linux to the hobbiest market with statements like this. It is clear
> > that to compete in the commercial marketplace, you have to provide
> > what the customer wants, not what you think the customer needs. If
> > the customer is Informix, for instance (just try to tell <name db company>
> > they need to rewrite their application because linux won't support PAE
> > because someone thinks it is ugly).
>
> (This is all strictly my opinion with no facts to back it up :-)
> Well, I don't know. One of the reasons, if not the reason, that
> "segmented" memory was devised was because of the lack of support in
> the CPU itself (not enough bits in pointer registers) for flat memory
> models. If we had 32 bit CPU's back when the PC first came out I can
> almost garentee that we would not have "64K segments." But, there was
> not another (financially feasible) alternative available.

I'm certainly not advocating a segmented scheme, the reasons,
not the least of which, are that it won't work to allow access
to more than 4GB.

>
> We have 64 bit CPUs available (cheaply) NOW. There's the Alpha,
> UltraSparc, and PPC (64bit version may not be "cheap" quite yet) at
> least. There is NO REASON to stick with 32bit Intel architecture if
> your requirements dictate over 4GB RAM in a system. Besides, you're

Tell that to a customer that wants (for whatever reason) an intel
system.

> Intel anal then the Merced is supposed to come out Real Soon Now. I'd

Merced has it's own set of problems, not the least of which is the
dearth of applications from ISVs.

> think that by the time any applications that REALLY needed over 4GB RAM
> were ported that the chip would be available.

The question is:

A customer has a choice of OS vendors for the Intel 4P box
with 64GB ram they have bought for 1/2 the price of an equiv
native 64-bit box:

1) NT (Which supports the PAE)
2) Unixware (Which supports the PAE a couple of different ways)
3) *BSD (Dunno PAE support)
4) Linux (no PAE support).

The customer wants to run oracle. Oracle can take advantage of
64 GB of ram under the first 2 choices, but not 3 and 4. Which
OS is the customer going to choose. Probably won't be linux.

If nothing else, implement the same API as Unixware does to get
application capture

>
> Given the horrendous rut the whole industry got stuck in by being
> forced into segmented architecture computers and the fact that we are
> still trying to "get out" of that rut once and for all, I would gladly
> support the "no segmented architecture in Linux" vote (if there was
> such a vote)..

I support it as well. I still believe you need to provide an API
to allow application access to the PAE memory. Can't use the segment
registers anyway for that purpose.

>

scott

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/