On Sat, 27 Oct 2001, Gunther Schadow wrote:

[...]
> Anyway, the bottom line is they have reprogrammed the 8600's
> microcode to implement some Prolog primitives and thus boosted
> the performance of 8600 with Prolog. They used mainly the space
> for the PDP-11 compatibility code for their Prolog additions.

Hmmm, nice stuff.

> Now, while this alone is pretty cool and reason to get an
> 8600 and use that trick, I wonder if the same approach could
> be used (a) on other VAXen (like the 6000's) and (b) if it could
> be used to improve Java performance on the VAX. Wouldn't it
> be cool if the old iron could keep up to pace with the Java
> stuff?

A slight problem is that most modern VAXen don't load the microcode from
disk at every boot. The 86x0 have the microcode on an RL02 disk, so you're
free to hack microcode to your hearts content.
I also happen to have pretty good docuemntation on the KA86 microcode,
along with a running 8650. :-)

> If the CISC design could be put to good use by taking care of
> some of the costly Java primitives?
> 
> Also, I wonder how such additions go together with the
> operating system. But as far as I remember the above-
> mentioned article, all they had to do is load the
> microcode hack from floppy via console before booting the
> OS. I think it even was BSD.

The operating system don't care. As long as you don't reuse existing
opcodes, you're free to play to your hearts content.

The 86x0 machines always load the microcode at power up, so you just
replace the standard microcode files with your modified files, and that's
it.

On a related note: NetBSD on the 11/750 patches the microcode the first
thing it does on boot. That's what the pcs750.bin file is for. :-)
(Yes, the 11/750 have some ram for microcode to allow for patches...)

	Johnny

Johnny Billquist                  || "I'm on a bus
                                  ||  on a psychedelic trip
email: bqt@update.uu.se           ||  Reading murder books
pdp is alive!                     ||  tryin' to stay hip" - B. Idol