Hmm. Start with searches in these newsgroups from the 1990s, when I was (among
other things) running sales support and customer support for XKL. Look for
posts from "
alde...@xkl.com".
Note, please, that I have not been an employee of XKL since 2003, and do not
now speak for them in any way, official or un-.
XKL was started by Len Bosack, co-founder of cisco Systems (original spelling),
to build what was originally intended to be cisco's premier product, a desktop
PDP-10 clone which implemented things like the full 30 bit address space in
hardware. The result was the introduction, in 1995, of the XKL Toad-1 System,
an extended clone the size of a two drawer file cabinet.
The Toad-1 featured a processor roughly 2x faster than the KL-10, a 4-port
10baseT Ethernet interface, a 4-port FASTWIDE SCSI-2 interface, and a 32MW
memory card, in a 7 slot backplane. One of the tests we performed was to put 4
memory cards in, to see if our mods to TOPS-20 were correct; the system sat
there happily using 128MW and never even hiccoughed. We sold a few systems, to
MCI, Digital, and Paul Allen (as an individual, not to one of his companies);
others were gifts to friends of the company.
The XKL-1 processor board was built around an FPGA from Xilinx.
The TOAD-2 processor is a reimplementation of the Toad-1 in a single FPGA. The
primary file systems are on 2 microSD cards; the Ethernet is built in, and they
provide a NAS application which runs on Linux to allow for much more than 2GB
of user storage. The early version, which we had at Living Computers: Museum +
Labs, was labeled "TOAD-2"; the current version is embedded in their DarkStar
optical router boxes. The TOAD-2 is used as the configuration driver in their
40G/100G and faster optical networking products (where routing is done in the
fiber header, not a general purpose processor); if no networking interfaces are
detected, the system will assume that it should boot TOPS-20 instead.
They are still using (bigger, faster) FPGAs in their designs. Len has always
held that smart engineers can get the speeds they want from better designs,
instead of wasting time on ASICs which will have to be competely replaced
instead of being modified if a bug, or an enhancement, needs to be updated.
At the time we build the Toad-1 (on which I did some of the operating system
work as well as my other duties), PC clocks had not reached the 1GHz range yet,
and it was a perfectly reasonable product line, given that PDP-10 users had at
least 20 years invested in local software (like most mainframe shops, primary
software loads were local). We even ported Tops-10 to the Toad-1 for a
potential customer; no one apparently has asked for Tops-10 for the TOAD-2.
--
Rich Alderson
ne...@alderson.users.panix.com
Audendum est, et veritas investiganda; quam etiamsi non assequamur,
omnino tamen proprius, quam nunc sumus, ad eam perveniemus.
--Galen