X-Received: by 2002:a05:6214:1bcd:b0:4af:646a:9793 with SMTP id m13-20020a0562141bcd00b004af646a9793mr519172qvc.94.1664409116042; Wed, 28 Sep 2022 16:51:56 -0700 (PDT)
X-Received: by 2002:a37:395:0:b0:6cd:cca6:4b8a with SMTP id 143-20020a370395000000b006cdcca64b8amr388714qkd.573.1664409115648; Wed, 28 Sep 2022 16:51:55 -0700 (PDT)
Path: csiph.com!1.us.feeder.erje.net!feeder.erje.net!usenet.blueworldhosting.com!feed1.usenet.blueworldhosting.com!peer03.iad!feed-me.highwinds-media.com!news.highwinds-media.com!news-out.google.com!nntp.google.com!postnews.google.com!google-groups.googlegroups.com!not-for-mail
Newsgroups: comp.sys.transputer
Date: Wed, 28 Sep 2022 16:51:55 -0700 (PDT)
Injection-Info: google-groups.googlegroups.com; posting-host=82.31.125.46; posting-account=l3olFwkAAAB7_YlbF47jpoRgdTkZ-iGF
NNTP-Posting-Host: 82.31.125.46
User-Agent: G2/1.0
MIME-Version: 1.0
Message-ID: <d7189237-e335-44c2-b763-e77cee63fdaan@googlegroups.com>
Subject: Hacking CSA Mandelzoom to work on a B008
From: Tom Stepleton <stepleton@gmail.com>
Injection-Date: Wed, 28 Sep 2022 23:51:56 +0000
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
X-Received-Bytes: 2908
Xref: csiph.com comp.sys.transputer:452

There, that's more like it:
https://photos.app.goo.gl/d4HmNMHEwW8Jry6x8

As noted in the previous thread and elsewhere, the CSA Mandelbrot program w=
on't work on a B008 as it's natively configured. The program freezes when i=
t attempts to execute some code on the T212 installed on the board.

I've been studying the worm program that distributes the software to the Tr=
ansputers, and there's a pretty easy (if kludgy) alteration to the source c=
ode that sidesteps this problem. Just above this line: https://github.com/a=
xelmuhr/T-Mandel/blob/master/FLBOOT.TAL#L179

add these instructions:

        ldl     LOOPA
        adc     -1
        cj      @R3

What it's doing is pretty unsophisticated. Note in this diagram: http://tra=
nsputer.net/mg/b008ug/figure/096/fig01.png
that the T212 is reached from TRAM0 (natively the host TRAM) via Link 1. Si=
nce the software distributing worm starts from TRAM0, the T212 is discovere=
d right away and joins the Transputer network assembled by the worm.

The Transputer network is never represented completely in one place --- ins=
tead, each node records its upstream node ("BOOTIN", the neighbour that "in=
fected" the node with the worm) and up to three downstream nodes ("BOOTOUT[=
0..2]"). It identifies these nodes by the memory location of their channel =
control word (if that's confusing to any reader, you can replace it with "l=
ink ID" to get the idea).

What my hack does is prevent Link 1 from being the downstream connection to=
 ANY node. It's quite crude, and it prevents the network from getting the m=
ost out of some topologies, but there is no ordinary topology you can set u=
p on a B008 in its default hardware configuration that will fail to operate=
 --- provided you stick to 32-bit Transputers.

I'll share my notes on the worm program before long.

--Tom