#Pentium NOT 486-compat!?
3 messages in this thread
We don't know where else to mention this, but we think this is the
best place, because I can't think of any other group of people
on-line likely to have access to 486s, Pentiums, and a program called
Hi-Res QFX 3.0.
Apparently, the Pentium is NOT 100% upwardly-compatible with the
486 and earlier '86 processors! We have found a utility that, at
least on our own systems, runs fine on all of our 486s, but crashes
completely if run on a Pentium!
The program in question is Altech Type Importer, a utility that
converts Type 1 PostScript and TrueType fonts to a proprietary font
format called "FastFonts." Type Importer is not sold separately, but
is bundled with applications which use the FastFonts format. One
such program which many here use is Hi-Res QFX 3.0. Type Importer is
called "TI.EXE" and is found in the "FONTS" subdirectory under the
directory which Hi-Res QFX 3 is in (usually called "HRQ3" if the
default installation was used).
We have about ten 486s, with varying configurations. Three of them
are part of our 3DS Network Rendering Farm, and contain identical
CONFIG.SYS and AUTOEXEC.BATs, and identical video cards (Trident
8900C), etc. Our new Intel Premio Pentium 60MHz PCI-bus tower
computer has also taken up farming 3DS renders, and shares the
identical CONFIG.SYS, AUTOEXEC.BAT, and video card as the 486s.
One day, we wanted to convert a batch of new fonts we had gotten to
the FastFonts format so HRQFX3 could use them. Since this is a
relatively slow process, we wanted to use the Pentium to do it.
Imagine our surprise when, during the actual conversion of the first
font on the list, the system froze completely without an error
message — and I mean FROZE. [Ctrl] + [Alt] + [Del] did nothing.
[Caps Lock] and [Num Lock] and [Scroll Lock] did not toggle their
respective LEDs. Only the RESET button (or power off and on) would
allow the computer to do anything else once this program had been
run!
And yet, the IDENTICALLY EQUIPPED AND CONFIGURED 486s in the Render
Farm ran the program just fine! The ONLY difference between these
systems are the CPU and the bus (the 486s have VL-Bus, but we don't
have any cards in either machine which uses VL-Bus or PCI-Bus, so
that shouldn't be a factor here). And remember, this is INTEL'S own
Premio machine, designed and assembled by the SAME folks who designed
the Pentium chip and PCI bus!
Of course, it COULD be something specific in our machine — perhaps
a bad RAM SIMM which has somehow managed to escape detection by
CheckIt and other diagnostics. Or perhaps something flakey in the
motherboard.
So, before we contact rumormeister Robert X. Cringely of InfoWorld
about this seeming incompatibility in the Pentium, we would like some
independent corroboration one way or the other. Anyone else who has
this utility AND has access to both 486s and one or more Pentiums,
please try this out. Try to convert a Type 1 font to FastFonts
format, and see if it locks up on your Pentiums but not on your 486s.
If it's just our machine, fine. We'll fix it. We certainly don't
want to do anything which might hurt Intel's rep if it's undeserved.
But IF this is in fact a genuine incompatibility between the 486 and
Pentium, well, The People Have a Right to Know!
Marion – Altech does have a product called "Alltype" wich does exactly
what you describe. It will work in DOS or Windows. I would have more
confidence in your findings with an off the shelf product. You might
want to check and see if this program is available at a place like
"Office Club". Its a very inexpensive program about the cost of a
basic video game. This is not to discount your observations but its
kinda hard to test against a proprietary product that most people
don't have access too. You might also want to contact Atech and ask
if there's anything unusual about the way the code is written. Or Ron
Scott for that matter. I'd be suprised if he didn't have a Pentium or
two laying around. <g>
Marion,
It's too early to say, but I doubt the Pentium is not 486 compatible.
You're font program is probably doing something nasty that it its just
happening to get away with on the 486 but not on the Pentium.
I remember a while back, there was problem with running Sprint (remember
Borland's word processor?) on the 486 because it used self-modifying code, a
real no-no. It worked on the 386, but not on the 486 because the 486 is
pipelined and the instruction they were trying to modify on the fly was already
in the pipeline when they were trying to modify it.
Since the Pentium is even more deeply pipelined than the 486 and since
it is capable of parallel execution of instructions, I would guess
that similar sorts of problems might come up with badly behaved
code. This is the price of progress.