[rescue] Sun crushes the used Sun market
Jonathan C. Patschke
jp at celestrion.net
Tue Jul 5 22:25:39 CDT 2005
On Tue, 5 Jul 2005, Charles Shannon Hendrix wrote:
>> Because this is for IA64, not IA32. Wintel/Itanium systems already
>> have this. No one's put sufficient pressure on the IA32 (or even
>> AMD64) people to fix the BIOS.
> I was thinking that Intel has the connections to generate that kind of
I don't think so. The BIOS<->hardware dependency only goes one way.
You could run any standalone program as firmware on an arbitrary PC, but
the BIOS is dependent on the hardware underneath it to present that 1984
AT abstraction. Intel can't easily force BIOS writers to make something
better because there's nothing with which to force.
Microsoft, on the other hand, can easily wave the "Windows $N+1 will
require $feature, and it will require it to be present in $way at system
initialization" axe, and you'd get trampled if you were standing between
a BIOS writer and his workstation.
> It would be nice to see something like the DEC Alphastation had: a
> dual-personality BIOS.
It would. But, there's no software demand for it. SRM grew out of
certain needs (ie: the ability to pass persistent arguments to loaded
programs and for sane serial-console management). Typical PCs run
Windows, and, therefore, don't have those needs. Those who run Linux or
BSD or Solaris are all too happy to find software that works around it,
because they're not a viable-enough market force to exert pressure on
BIOS writers, and they'd like to have computers on which to run their
Linux or BSD or Solaris.
There are halfway-decent BIOSes out there. COMBIOS (I think that's the
name of what runs on the Soekris systems, anyway) is quite nice, even if
it looks like a BIOS to the software it loads.
> Normal users would see the same old crud, but there would be an option
> to use a good boot ROM for those who wanted it.
I guess what I'm getting at is that there's really only a reason to do
that if the software will take advantage of it. SPARCs took advantage
of OpenBoot to use third-party devices before loading drivers for them
and to get a complete hardware map (with parameters) without having to
probe the whole system. PCs don't have software that'll make use of
that. One or two vendors making a change won't make much of a
difference either; how many PCs switched to something like the (much
better) underpinning that SGI used in their x86 Visual Workstations?
With computers being a commodity, and money being the final decision-
maker, only very well-to-do private corporations (and enterprising
individuals) can take the risk of doing Something Different. Unless it
shakes the world, chances are that no one else will take notice.
Jonathan Patschke ) "Beatniks and politics, nothing is new.
Elgin, TX ( A yardstick for lunatics: one point of view."
USA ) --Strawberry Alarm Clock, "Incense and Peppermints"
More information about the rescue