On Fri, 24 Dec 2004 11:49:13 -0500, George Neuner
<gneuner2/@comcast.net> wrote:
>On Fri, 24 Dec 2004 01:26:19 -0800, Jeff Liebermann
><(E-Mail Removed)> wrote:
>
>> The 486 is a 16 bit processor. 32 bit requires a Pentium.
>Sorry, but you're completely wrong.
Argh. You're right and I screwed up. However, I have a bad excuse.
Intel released the 486 in 1989 and the Pentium 60 in 1993. CardBus
was released in 1995. By the time CardBus started appearing in
laptops, the 486 was long obsolete and those laptops used Pentium
processors. At the time, I was told by one of the CardBus chip
vendors that they would only support development for the Pentium.
I've never bothered to check if this was true or determine if there
were any 486 laptops with CardBus controllers. Apparently, you have
one. As I said before...I'm amazed that the card fits.
Incidentally, one potential problem is that 16 bit PCMCIA ran on 5VDC
while CardBus runs on 3.3VDC. That's the reason for the card keying
slot. For a while, the controllers were only one voltage. You could
plug an older 16bit PCMCIA card into a CardBus slot. It would fit,
but not run. Eventually, the CardBus controllers became dual voltage
to accommodate older cards. However, there a few controllers that
didn't. Before you try the Orinioco card, which is 5VDC, check the
CardBus to PCI chip specs on your laptop to make sure it will work
with 5VDC cards.
>I've been writing code since 1982
>and I've worked with most of the major Intel chips and various of
>their clones (AMD, NEC, etc.).
Well, I've been pretending to be the worlds worst programmist for much
longer to avoid having to write code. Politics, hardware, and RF is
my forte. At this time, I couldn't program my way out of a paper bag
and suspect that inflicting my code upon the world might precipitate
the demise of civilization.
>The 386 introduced 32-bit mode in the x86 line (integer only - there
>was still a separate 387 numeric coporocessor). The 486 combined the
>integer and FP cores in a single chip [ initially the chips had
>problems - the "sx" models were chips sold with nonworking FP units ].
>The Pentium line introduced dual (but assymetric) integer pipelines
>and double wide (64-bit) memory buses.
>
>
>One of the more interesting tricks for 486 programming was to optimize
>for the Pentium instead. The P5's primary integer pipeline was nearly
>identical to the 486's, but the secondary pipeline only executed a
>subset of integer instructions - simple, fast register ops. Keeping
>both pipelines busy meant using lots of these simple instructions.
>The similarity of the primary pipeline to the 486 meant the 486 could
>also benefit from using lots of simple instructions.
>
>At the time 486 optimizers were really reworked 386/87 optimizers.
>They tended to favor complex instructions rather than simple ones
>because that was the way to get best performance on a 386. If you
>*knew* for certain your code would be running on a 486, optimizing it
>for the Pentium instead could give a 5-10% performance boost
>essentially for free.
Nice trick. As I recall, Intel was in the CISC (complex instruction
set computah) camp, while everyone else was pushing RISC (reduced
instruction set computah). Intel introduced some "RISC-like"
instructions claiming the best of both worlds. They may have been
right, but they couldn't convince the designers to do much with RISC.
Eventually Intel contrived the i960 chip and licensed the ARM chip
core, which are true RISC chips. In the heyday of the 486 processor
(1989 thru 1995), Intel was a single CPU chip company trying to place
their general purpose 486 processor into every conceivable and often
inappropriate application. Their main competitor was the Motorola
68000 series, which also was a CISC chip, but had some big holes in
their glue chip lineup (no MMU chip for quite a while) and was doing
their best to do the same thing. Intel eventually came out with the
Pentium, which has a RISC core. Motorola went a step further with the
PowerPC chip, which is mostly RISC. Anyways (as the topic wanders),
if you knew how to deal with Intel's "RISC-like" instructions, there
were some substantial performance benefits.
If you have problems with the Orinoco card under Windoze 3.11, bug me.
I have some Orinoco "Classic" Silver cards and a few old laptops
running Windoze 3.11 (mostly for radio programming).
--
Jeff Liebermann
(E-Mail Removed)
150 Felker St #D
http://www.LearnByDestroying.com
Santa Cruz CA 95060 AE6KS 831-336-2558