Talk:Fabritek Core Memory

Still here?

I'm somewhat amazed the one from AI still exists. I would have assumed it would have been junked when HIC did the work to enable LISPM memory to be used on AI to replace the failing core memory. (I forget what hardware was used for the second moby on AI.) Jnc (talk) 21:22, 5 July 2019 (CEST)

I'm amazed too! I had the idea that they should scan the contents, but I suppose the bits have dissapated by now. I think the other moby was an Ampex... but the final configuration had three mobies. The fourth was dedicated to the 10-11 interface and PDP-6. Larsbrinkhoff (talk) 07:27, 6 July 2019 (CEST)

Generally core maintains its contents for a very long time; I've read of other old core memory being read out decades after last use. But in this case I doubt it would be worth it; the contents would either be i) stuff we have already (ITS system/commands), or ii) junk of no interest (personal data).

AI's pager had a hardware address space of 4 mobies, but I'm pretty sure only 2 mobies of actual memory. I think the third was the PDP-6 main memory; the 11's had a moby to themselves. (IIRC, they didn't tweak the memory bus cables, the way they did with the I/O bus; the pagers (on all 3 machines) just had 2 or 4 totally stock memory bus cables coming out of them, which were cabled to the appropriate boxes.) Jnc (talk) 15:48, 6 July 2019 (CEST)

Some time ago I made notes about how the memory sizes of the ITS machines evolved: https://github.com/PDP-10/its/issues/1585

According to this, AI was as you describe at one point. But later a third moby was installed, and the PDP-6 and 10-11 was made to share a moby. ITS has code to accommodate this configuration. Larsbrinkhoff (talk) 18:47, 6 July 2019 (CEST)

I wouldn't put to much weight on the CONFIG > entries as to what machines had; if something's not listed, yes, it probably wasn't there, but the other direction... The memory size setting just allowed for up to that much memory; one could switch memories out at boot time without needing to rebuild the system. (That happened a lot on MC.)

I really don't recall AI having more than the two mobies before HIC did the LISPM memory; I'm pretty sure if they'd added some it would have been noticeable (physically). Maybe check with Greenblatt, or someone? Jnc (talk) 16:17, 7 July 2019 (CEST)

Thanks, that's worth looking into. Larsbrinkhoff (talk) 19:01, 7 July 2019 (CEST)

For what it's worth, here's what I can see in the code.

In ITS 785 from 1973, system memory is 272K, and PDP-6 core is mapped right after this. There is no 10-11 interface.

In ITS 848, system and PDP-6 memories remain the same. The 10-11 interface has been added, and sits in moby 1. There is code to share moby 1 between the system, PDP-6, and 10-11.

In ITS 997 from 1975, the code to share a moby has been commented out. System memory is 512K, mobies 0 and 1. The 10-11 moby is 2, and PDP-6 is 3.

PDP-6 support was dropped from ITS in August 1978.

In 1981, the system memory limit was increased to 768K. At this point, the 10-11 moby was also changed from 2 to 3.

10-11 support was dropped in 1982.

Larsbrinkhoff (talk) 09:34, 8 July 2019 (CEST)

Those all sound plausible; 1981 is about when HIC did the LISPM memory, and the core was de-comissioned. Jnc (talk) 14:36, 8 July 2019 (CEST)