Talk:IBM System/370

Archives (Index)

1

This page is archived by ClueBot III.

PSW tables

The article currently has these tables for PSW format

• S/370 BC mode
• S/370 EC mode
• S/370-XA
• ESA/370

The format for ESA/390 is identical to that for ESA/370, but the names of two bits in the program mask have been changed to reflect that they apply only to hexadecimal (legacy) floating point as opposed to binary (IEEE) floating point. I'm considering three options:

1. Add an ESA/390 table that is almost identical to the ESA/370 table.
2. Add footnotes to bits 22 and 23
3. Add text to the ESA/370 table to show both names for bits 22 and 23

Which option is best?

Also, the tables should be copied to Program status word once complete. Shmuel (Seymour J.) Metz Username:Chatul (talk) 14:35, 22 December 2020 (UTC)

I vote for 2 or 3. Creating a completely new table for two name changes seems like overkill. Peter Flass (talk) 15:38, 22 December 2020 (UTC)

How about a variation on 3; relabel the table as ESA/370 and ES/390 and distinguish the two bits with one local footnote Tom94022 (talk) 18:51, 22 December 2020 (UTC)

I think it's clearer the way that it is now, but if you want to change the name to ESA and change the Program Mask table to present the two nomenclatures in a different fashion, go ahead. Note that I've added layouts to Program Status Word, although I need to decide how to handle z. Shmuel (Seymour J.) Metz Username:Chatul (talk) 01:16, 27 December 2020 (UTC)

Separate pages for the S/370 family and the S/370 architecture?

We have two separate pages for S/360 - IBM System/360, which talks about the history of S/360, enumerates models, gives release dates, etc., and IBM System/360 architecture, which purely discusses architectural details, mentioning only the Models 20, 44, and 67 in passing because of their exceptions/extensions.

Should we do something similar for S/370?

A similar split for the 64-bit machines arguably exists, with IBM Z and z/Architecture. Guy Harris (talk) 06:54, 28 December 2020 (UTC)

Yes, if someone is willing to do the work a split is desirable. Shmuel (Seymour J.) Metz Username:Chatul (talk) 05:18, 1 January 2021 (UTC)

I/O Evolutions

There is an I/O Evolutions section, but it does not discuss the I/O changes in the S/360 line and has a {{main article|IBM System/360#Channels}} hatnote; It does not discuss, e.g., channel set switching, 4.5 MB/s B&T, ESCON, FICON, the new channel susbsystem of XA; essentially the section only discusses what changed from the initial S/360 to the initial S/370. Shmuel (Seymour J.) Metz Username:Chatul (talk) 05:34, 1 January 2021 (UTC)

@Chatul: good point. I added ESCON to the section and will add FICON later. Note I had to first improve ESCON before linking here. May I suggest u take a hack at the other evolutions? Tom94022 (talk) 18:18, 1 January 2021 (UTC)

Unexplained change to citation

@Guy Harris: Edit https://en.wikipedia.org/w/index.php?title=IBM_System/370&oldid=1153105243 changed a citation for the block multiplexor channel from section Types of Channels on pages 13-4 – 13-5 of PoOps to Start I/O Fast Release on pages 26-27. The new citation belongs on Start-I/O-Fast Queuing, not on the block mx channel. The original {{sfn}} should have specified the URL http://bitsavers.org/pdf/ibm/370/princOps/GA22-7000-10_370_Principles_of_Operation_Sep87.pdf#page=355 for the -11 PoOps but was otherwise appropriate. The -0 PoOps does not include the relevant material, although a citation of the channel to CU OEMI would have been legitimate. -- Shmuel (Seymour J.) Metz Username:Chatul (talk) 12:16, 4 May 2023 (UTC)

So the problem is that IBM didn't give enough details about block multiplexor channels in the -0 PoOps? Guy Harris (talk) 19:06, 4 May 2023 (UTC)

I guess so - the first edition is, according to its cover, a manual that "describes extensions to the functional design of the System/360 that are incorporated in models of the System/370", so it's an addendum to a (sufficiently-recent) S/360 Principles of Operation, rather than a full independent Principles of Operation manual, as later editions (which had a lot more "extensions" to describe, such as "I haz MMUDAT box").

I'm not sure why they didn't describe block multiplexor channels, though. For what it's worth, the 2880 Block Multiplexor Channel was announced in February 1970 for the Models 85 and 195; the Model 85 was announce in 1968 and shipped in late 1969, so that was before the announcement of the first block multiplexor channel, and System/370 was announced in June 1970, which was after the announcement of the first block multiplexor channel. The block multiplexor channel appears to be mentioned in "IBM's 360 and Early 370 Systems" only in a footnote, and that book also speaks of S/370 being intended to come out in two phases, with phase 1 being a "small delta from S/360" (and a smaller delta from S/360 Model 85) phase and phase 2 being the larger "now with added DAT!" phase, so perhaps either 1) the block multiplexor channel was a late addition and there wasn't time to document it more fully, 2) they documented it in a manual about the 2880, or 3) they thought it wasn't important enough to mention in the first edition and deferred it to a later edition that was a stand-alone Principles of Operation for phrase 2.

In any case, I've updated the reference to point to the Eleventh Edition section describing block multiplexor channel operation. Guy Harris (talk) 10:04, 13 May 2023 (UTC)

Right pages, wrong section - I've corrected that. Meanwhile, I ran into a problem: {{snd}} was displaying what looked like an HTML character attribute; I'm assuming that under some circumstances it doesn't generate trailing semicolns. IAC, I used a - to bypass the problem. -- Shmuel (Seymour J.) Metz Username:Chatul (talk) 01:24, 14 May 2023 (UTC)

References

Multiple operating systems

The statement The 390, which was based on a new ESA/390 model, expanded the multiple memory concept to include full hardware virtualization that allowed it to run multiple operating systems at the same time. is anachronistic; Virtual Machine Facility/370 (VM/370) allowed running multiple operating systems on the same machine in 1972 and PR/SM on the 3090 offered similar support in 1988 as a processor option, while ESA/390 didn't come out until 1990.

There were certainly incremental improvements in PR/SM for S/390, but the basic capability had been around for two decades. -- Shmuel (Seymour J.) Metz Username:Chatul (talk) 17:55, 26 April 2024 (UTC)

Notes