Jump to content

Talk:IBM AS/400

Page contents not supported in other languages.
From Wikipedia, the free encyclopedia

Add additional historical information on AS/400 hardware and OS for the years 1988 to 2006

[edit]

EWLwiki I ask Vt320 and Guy_Harris for your permission, guidance, and consensus, in adding additional content of the rich history of the AS/400 platform, with the hope that you do not revert sincere, honest edits.

The AS/400 article lacks clarity in the chronological, extensive incremental advances to hardware and operating system occurring in the years 1988 to 2006.

The brand names eServer iSeries, System i5, System i, and each processor, can add important clarity to the family that became so different over time, until the platform was discontinued in 2008 with the introduction of IBM Power Systems. Many people lack understanding and appreciation for the significant changes that occurred.

For example, it would be very helpful be more chronological in the article. Example: describe more in dept the Dynamic Logical Partitioning that was introduced with eServer iSeries on POWER4. LPAR was a huge technological advance.

Example: the Integrated File System (IFS) [1] did not exist on the platform in the early years. We should describe in detail at what version release (and processor level) the IFS was implemented into the architecture, and how significant the IFS became to the later generations of AS/400. IFS is just one of the many things that show that AS/400 began as an midrange computer. It was not a server, not connected to the internet, and very proprietary, not easy to connect to other systems.

IBM has always made it so that application code is forward-compatible to a newer machine and OS version, so that customers do not need to recompiled source code IBM RPG II or IBM RPG III or IBM COBOL or IBM ControlLanguage when migrating to a newer machine onnewer IBM POWER microprocessors or OS/400, not IBM i version level.

But as customers incorporate into their application code the hundreds of features of the computer language and hardware, that new code is NOT backward-compatible, that is, cannot run on an older processor or OS level because the those language features and OS features did not exist on the older machines.

What say you? EWLwiki (talk) 18:50, 12 December 2021 (UTC)[reply]

IFS is a software feature, not a hardware feature.
> But as customers incorporate into their application code the hundreds of features of the computer language and hardware, that new code is NOT backward-compatible, that is, cannot run on an older processor or OS level because the those language features and OS features did not exist on the older machines.
That's also a software issue. As time goes on, old hardware goes out of support, and vendors will no longer support old hardware with new versions of the OS or development tools. There is no technical reason why old AS/400s could not support new versions of RPG, IBM simply made the decision not to support them.
I agree that LPAR support could probably be elaborated on. Vt320 (talk) 20:57, 12 December 2021 (UTC)[reply]
There's hardware, there's software, and there are computing platforms. Unless you're toggling raw machine code in from the front panel, or loading hand-encoded raw machine code with a load button, platforms aren't hardware-only - there's some software involved in getting your own software running (even if it's running on another machine as a cross-development tool). And many platforms include at least some hardware, e.g. a personal computer's keyboard and mouse, or other I/O peripherals.
The Macintosh platform has changed its CPU instruction set several times, so, for example, the macOS platform isn't identified by being based on PowerPC, x86, or ARM; to that extent, the CPU is obviously necessary, but, for most developers, isn't a key part of the platform (I suspect most developers don't develop for those instruction sets, unless they need to write CPU-specific low-level code). So it's not necessarily the case that all of the hardware is a key part of the platform.
The "400" platform also changed its CPU instruction set, and the majority of programmers had no reason to care (the main ones who did were the VMC/SLIC developers, who also changed programming languages). The main significant part of the CPU for the 400 platform is, as far as I can tell, the presence of tag bits protecting pointers.
So what hardware changes to the platform added significant capabilities?
And what parts of the platform should be covered by IBM i and what parts covered by AS/400 and IBM Power Systems? Guy Harris (talk) 22:42, 12 December 2021 (UTC)[reply]
To my mind, it is precisely the hardware-independent capabilities which should be covered by the IBM i article. This includes things like TIMI, the file/object system, development tools, PASE, etc.
The hardware articles should cover the specifics of those machines, the various models, CPUs, I/O hardware, etc. LPAR is another good example of something which belongs to hardware, since that is a capability provided by the hardware at a layer below the operating system. Vt320 (talk) 23:28, 12 December 2021 (UTC)[reply]
From a quick look at the Power ISA 3.0 specification, it appears that at least some of the LPAR support is provided by the chip; presumably there's also a hypervisor involved - is that provided in system ROM or in something booted from secondary storage? In either case, that'd count as something belonging on the AS/400, IBM System p, and IBM Power Systems pages; is the LPAR support similar in the RISC AS/400, System p, and Power Systems machines? Currently, AS/400 and IBM System p mention LPARs, mostly relying on logical partition, Dynamic Logical Partitioning, and Micro-Partitioning to provide the details. All three of them should perhaps be synced up.
(And what about the VMC/SLIC? I'd count them both as low-level OS software atop which IBM i run, and put them on the IBM i page, with a note on the AS/400 page about them as well.) Guy Harris (talk) 07:57, 17 December 2021 (UTC)[reply]

References

  1. ^ "Integrated File System". ibm.com. International Business Machines. Retrieved 12 December 2021.

Convergence Section, previous name Consolidation

[edit]

Consider my edit: Convergence title makes more sense than Consolidation, evidenced by words in this section, somewhat modeled on article Technological convergence. EWLwiki (talk) 22:38, 19 March 2022 (UTC)[reply]

Consider the way Linux articles are written

[edit]

I am impressed with the style shown in articles Linux kernel,History of Linux and Linux kernel version history I think the articles on IBM AS/400 and IBM i and IBM Power Systems could benefit by incorporating some of the style shown in the Linux articles. What say you? EWLwiki — Preceding undated comment added 22:26, 19 March 2022 (UTC) EWLwiki (talk) 22:39, 19 March 2022 (UTC)[reply]

CREATE PROGRAM instruction

[edit]

There should be a discussion of the role of the CREATE PROGRAM instruction, which compiles a TIMI template into a program object. I don't know about AS/400, but on S/38 that was the only way to create execurable code. -- Shmuel (Seymour J.) Metz Username:Chatul (talk) 16:51, 24 January 2024 (UTC)[reply]

"Architectural shift developed wholly internally"

[edit]

A citation was requested for "Although announced in 1988, the AS/400 remains IBM's most recent major architectural shift that was developed wholly internally." with the comment "...while POWER 7, 8, 9 and 10 were still developed in-house".

New Power processors aren't architectural shifts - they run newer versions of the Power ISA, but they don't support a completely new ISA.

But I'm not sure what the AS/400 is a shift from. Much of it is the same system architecture as System/38. The instruction set shift did switch from the developed-wholly-internally IMPI to the developed-internally (if you don't count Motorola and Apple's contributions to PowerPC) PowerAS (or whatever term refers to the AS/400-only extensions to PowerPC/Power ISA), but that wasn't part of the switch from System/38 to AS/400, that happened later in AS/400 history.

So the question remains - what does "architectural shift" mean here?

And, in any case, a citation is called for. Guy Harris (talk) 19:19, 21 May 2024 (UTC)[reply]