Thank you Henk, no I hadn't come across that system3/basic folder at bitsavers. But it is exactly what I was looking for! One intent was to try to verify that indeed the IBM 5100 is running the same "version" of BASIC as was on the native System/3 itself, and approximately when that version of BASIC was available (which seems to be 1971). Although we confirmed the 5100 is (when in "BASIC mode") running the S/3 emulation (we found the instruction jump table in the ROM), I was never really sure where that BASIC code had come from (the one implemented using S/3 instructions). [ now it might be interesting to do a hex comparison of the actual S/3 native BASIC -- what that S/3 emulator is running, and what we've extracted out of the IBM 5100; note the 5100 arranges things in 6KB segments, not every segment is fully used but code generally does straddle across two segments -- but the size is approximately 54KB, which any early microcomputer would struggle to handle that ] Cory is doing a talk about BASIC at June VCF - mostly from the microcomputer perspective, but I wasn't sure if he'd be interested in touching on any pre-1975 aspects (aside from its Dartmouth origin). The S/3 isn't quite a "boot to BASIC" system -- I was curious on the process of how it BASIC was started on this system, and page 12-14 of that Guide to BASIC PDF in the link you provided describes exactly that sequence (which I see followed in your example1/2.txt outputs-- any chance that S/3 emulator runs on a PC? Imagine a MAME front end with a Model 6 CRT depicted). Anyway, it leads to a question of what was the first boot-to-BASIC system? And tentatively I'm thinking that might have been the Wang 2200 (but I've no idea what the "lineage" of the Wang 2200 BASIC was). Yes, yes, there was the HP9830 (if the BASIC ROM cartridge was inserted) - sales of both systems seemed to have started really in '74. Thanks! Steve On Wed, May 7, 2025 at 12:43 AM Henk Stegeman via cctalk < cctalk(a)classiccmp.org&gt; wrote:

Makes sense, IPL (instruction program load?) - how were the disk packs initialized to indicate it was an IPL media? I assume the content to be loaded had to be on a specific sector or of a file-type indicating it is to be loaded on startup? And it is "boot to BASIC" (in quotes) mainly because the operator has to do the effort of at least inserting the proper disk (but it was a "pack", not just a single disk?) I assume SCP is a serial line of some fashion, to support that idea of remote-loading of an OS or startup software? Kind-of sort of an early form of PXE to "boot from a network"? While we're here, another note: on the SCAMP, they emulated the IBM 1130's version of APL (so I may have misspoken earlier - it wasn't an IBM S/3 they had trouble sourcing, but rather during the SCAMP dev it was an IBM 1130 they had trouble finding to borrow/lease). This SCAMP was the prototype leading to the IBM 5100 (where in the IBM 5100, they switched to using an S/360 based APL). On an off-chance of any IBMers out there, I'm still looking for Kitty Price or Patrick Smith (two known experts of the PALM processor, they wrote a paper referring to it in 1974). In the appendix of that paper, they refer to a "1130 PALM Simulator" (i.e. before the PALM was available to them, someone had developed a simulator of it on the 1130). Then in the next section, they refer to a "1130 Simulator which runs under VM/370". Finding any of that software would be incredible. <SCAMP/IBM_SCAMP_1of3_May1974_AnArchitecturalAndDesignOverviewOfScamp_PatrickSmith_KittyPrice_Part1.pdf at main · voidstar78/SCAMP <https://github.com/voidstar78/SCAMP/blob/main/IBM_SCAMP_1of3_May1974_AnArchitecturalAndDesignOverviewOfScamp_PatrickSmith_KittyPrice_Part1.pdf On Thu, May 8, 2025 at 2:38 PM Henk Stegeman via cctalk < cctalk(a)classiccmp.org&gt; wrote: > A few corrections: > The BASIC on the S/3 model 6 was a "boot to BASIC" system. You could IPL > BASIC from R1 > or business SCP from F1. > This BASIC system had virtual memory implemented. Real storage was 8 or > 16k but in the > BASIC environment approximately 54KB was available during execution. > Only thing IBM did was reuse of the Psuedo machine code definition in > the 5100. > Another nice feature of this BASIC was the online helpfiles. > Regards Henk

"Initial Program Load" I don't know any system other than the 360 and that one only very vaguely. I think it would load one record (a card, if IPL from the card reader; a sector, probably the first one, if disk; don't know about tape) and execute that. And I think that the loaded data is a channel program, so it's handled to the IPL channel to execute. A special case is "Emulator IPL" on the 360 model 44, which loads the special memory (inaccessible from normal execution) that contains the code to emulate SS class instrutions. The emulator is provided as a card deck, in standard assembler object file output format, one object data record per card. And the loader is a self-modifying channel program that reads in the entire emulator deck by an impressive mix of data and command chaining, picking up the data load address and length fields of each object card to modify the CCW that then reads the data bytes into the correct memory location. paul

That paper is what led me to asking the Smithsonian if I could see the Joe George collection (since in their inventory of documents, they listed said 1974 "Overview of SCAMP" paper, so theirs is the only place I've seen a copy of it - marked as ZZ20-6426 of the Palo Alto Scientific Center (largely an IBM internal document at the time, other than that - I'm not sure what the intended publication audience was). Kitty Price was apparently an IBM 1130 expert, helping Patrick Smith host that system using PALM (so they were figuratively "tied at the hip" figuring out that integration). Kitty was hand writing advise about this (aspects of shadow memory and improving floating point precision), right as she was departing for maternity leave in Winter of '73 (per the journal notes). Curious Marc once mentioned the SC/MP largely contributing to his early interest in computing - it was a later 1970s kit, right? But altogether unrelated to the 1973 IBM SCAMP. On Fri, May 9, 2025 at 12:13 PM Bill Degnan via cctalk < cctalk(a)classiccmp.org&gt; wrote:

On the HP9830 - I'm not sure if it was really available in 1972 (I suspect it was announced in that year, but not for general public sale until the following year). I say this because in the SCAMP prototype development journals, Joe George (of IBM) doesn't mention the HP9830 until October 1973 (and in his notes, he refers to it being mentioned in "Electronic News page 31", but I've never been able to find that reference). I believe Paul Friedl was aware of some early HP9830 concepts of prototypes (maybe in 1972), motivating him to try to incorporate the "Bathia" style printer into their "portable computer" concept (but ended up never pulling that off -- but rather they incorporated a more functional CRT). My notes on this are about 3/4th down on this page. https://voidstar.blog/ibm-scamp-joe-george-notebook/ We had an operational HP9830 last year at VCF SW (with the attached printer, it is a wonderful system). While it does "boot to BASIC", as I recall one expansion cartridge is necessary to support some additional keywords (like possibly the TRIG functions or some certain operations). So it may be more fair to also consider "level of BASIC" that is being supported (e.g. Microsoft's own 4K BASIC being quite limited, or similar systems trying to do some form of BASIC in under 4K of ROM). My notes about the HP9830 experience at the end of this page: https://voidstar.blog/exploring-project-ahl-in-2024/ Someone mentioned the IBM System/3 also had a form of boot of BASIC. I don't think it was 2 seconds of startup, as I think it did load from a disk pack. But the disk pack could be setup to auto-load something on startup, so you get the effect of getting to a "READY" prompt on power up. Speaking of Microsoft BASIC's - I suppose none of the Altair offerings were really "boot to BASIC" (maybe you could rig it to auto-load from the paper tape). In the microcomputer domain, does the PET remain really as the first boot to BASIC system? (and fairly full featured, with floating point and trig functions) -Steve On Wed, Jun 18, 2025 at 4:19 PM Brent Hilpert via cctalk < cctalk(a)classiccmp.org&gt; wrote:

On Tue, Jun 24, 2025 at 5:19 AM Steve Lewis via cctalk &lt;cctalk(a)classiccmp.org&gt; wrote: Trig functions (and floating point numbers/variables) are standard on the HP9830. As a aside, the HP9810 and HP9820 calculators (related machines in that they used the same bit-serial processor) did need an extra ROM module for the trig functions There are many expansion ROM modules for the HP9830, including : String Variables. Without this you can print string constants (to label inputs and outputs. for example) but nothing more. This ROM adds the ability to store and manipulate character strings in variables Extended I/O : Essential if you want to use any I/O devices other than the printer and cassette drive, really. Allows binary I/O, use of HPIB (with the appropriate interface), etc Matrices : Adds the MAT statements that were present in original BASIC and missing in most microcomputer versions. This ROM for the HP9830 is surprising common in my experience. -tony

Ah yes, it was the MAT features that the stock 9830 lacked (without an expansion ROM slotted into the side), likewise with STRING support. As for the Apple II being a first boot-to-BASIC microcomputer... Maybe. Story goes that Chuck Peddle pulled all nighters getting the earliest PET ready for CES in January 1977 (the one with a wooden case), which included a boot-to-BASIC ROM - and per a VCF presentation by Leonard Tramiel (IIRC), this (now archived into a vault) prototype went on tour over to Europe also. Apple followed suit by mid-year (around May or June), but with INTEGER (only) BASIC. So while the PET had technically made sales earlier (at the CES show in January), from what I've read Apple II's were first to actually get delivered into customer hands. And good find of the HP9830 HP journal, I do recall coming across that in the past, and the price sheet near the end settles it for me (although that's the very end Dec 1972, so might be a little unclear exactly when deliveries happened -- I believe it was the same for the Altair? the Jan article was actually available Dec 1974, but orders still didn't arrive till a few months later into 1975?). The Wang 2200 was around mid-1973 (as a desktop boot to BASIC system, initially with that clunky external CPU and external PSU -- but a decent multi-row screen). There was a BASIC for one of the Intel 8008 systems in 1974, but I don't think any of them were boot-to-BASIC. On Tue, Jun 24, 2025 at 2:50 AM Tony Duell &lt;ard.p850ug1(a)gmail.com&gt; wrote:

Interesting - then I wonder why in 1974 they bothered to build a S/3 emulator into the IBM 5100. They had already done the S/360 emulator and gotten APL working, so why not use this OS/360 version of BASIC? (well, OS/360 is the OS -- maybe there were more hooks and baggage in getting that working; i.e. they didn't want to pull over the entire OS/360 environment) Or maybe there were more features they liked about the S/3 version? Did the TSO BASIC have floating point, MAT, PRINT USING, etc? On Wed, May 7, 2025 at 12:43 AM Henk Stegeman via cctalk < cctalk(a)classiccmp.org&gt; wrote:

of what is encompassed by the emulation. It’s at least plausible that the > state of the emulation required for implementing the APL interpreter was missing some non-trivial set of features required by the BASIC interpreter. Certainly possible - we've found the jump tables in both emulators that gives an inventory of what opcodes they supported, and to fit in the capacity and cost (of what amount of software could be fitted into a slotted card at the time) that they were targeting, some compromises had to be made [ so neither were a "general 100% cycle accurate" kind of emulation, rather the focus was on supporting the essentials necessary for those respective applications of APL and BASIC, adapted from line-printer over to kbd/CRT fashion ] I recall in the SCAMP Joe George engineering docs, there was an entry where they had trouble funding or obtaining a "spare" System/3 for their development (that being IBMs own internal developers). I'm on travel and will have to lookup the exact reference later. Perhaps similarly, whatever S/360 they borrowed or used previously was no longer available. So (maybe) just being pragmatic on what was available to them at the time, they went the S/3 route. Another reason might possibly be: iirc, 1975 the trial portion of IBMs long antitrust suite was starting (the decade long one ending right at the release of the IBM PC in 1981). That being the issue of coupling software to their hardware. A thought was that by at least showing the concept that "some other language" could essentially be plugged into the system, it at least gives the impression that a third-party could technically do the same. Of course no one ever did (I'm not sure if they'd have access to enough info then to do so), but there was mention of a FORTRAN card concept for the system. Aspects along these lines are also implied in the dev journal mentioned above (if not for third party, but even in internal debate and justification by Friedl to get the project funded, a la "you want to build WHAT?" kind of discussions to a mainframe oriented business unit). Dennis Robinson might still be available to chime in on this aspect. -Steve On Wed, May 7, 2025 at 5:18 AM r.stricklin via cctalk &lt;cctalk(a)classiccmp.org&gt; wrote: > > > > On May 6, 2025, at 11:32 PM, Steve Lewis via cctalk < > cctalk(a)classiccmp.org&gt; wrote: > > > >>>> OS/360 TSO BASIC, pgm-nr 5734-RC2 round 1971 > > > > Interesting - then I wonder why in 1974 they bothered to build a S/3 > > emulator into the IBM 5100. They had already done the S/360 emulator > and > > gotten APL working, so why not use this OS/360 version of BASIC? > > of what is encompassed by the emulation. It’s at least plausible that the > state of the emulation required for implementing the APL interpreter was > missing some non-trivial set of features required by the BASIC interpreter. > > It’s also worth considering, as you point out, that perhaps the BASIC > inerpreter depended on some OS services. APL\360 had originally been its > own standalone timesharing environment. I’m unsure of whether that > situation was fully at an end by the time of APL\360 on DOS and MVT > (1969ish) but I don’t doubt the standalone code was likely still reasonably > fresh by the time work was underway on the 5100. The S/3 BASIC may have > been in a more reasonable state for adapting to standalone operation on the > 5100. I haven’t kept up with all the scholarship around the history of the > 5100 but it’s also not exactly implausible that there was some phase of the > project where the thing had been two separate products. > > ok > bear. > >

APL on the 5100/5110/5120 was based on the IBM mainframe APL/SV, (follow-on to APL\360), with several added features. The "360 emulator" was not a full emulator of 360/370, but only emulated exactly those instructions needed by the APL interpreter (that was written in 360/370 assembler language). Similarly, the System/3 emulator that was developed for the 5100/5110/5120 BASIC implemented only those instructions that were needed by the S/3 BASIC software. System/3 was a machine with limited amounts of real CORE memory, so developers recognized early on that they needed to implement a kind of a "virtual machine" for BASIC, and that was coded in S/3 assembler language. This layer also provided the "virtual" memory for BASIC use, done via "software paging" on a machine (S/3) that had no hardware for virtual memory, relocation, etc. See also: https://www.glennsmuseum.com/items/s3m6/ Glenn Henry was instrumental in the development of the S/3 Model 6 BASIC, and later led the effort to create the IBM System/38. He was eventually promoted to become an IBM Fellow. The design of the virtual instruction set used for S/3 BASIC was used to create the "scientific instructions" for the S/32, S/34 and S/36 that was used by BASIC and FORTRAN on those systems. Mark S Waterbury

On Wed, 14 May 2025, mark.s.waterbury(a)gmail.com wrote: I had disassembled and partially documented the 5110 S/3 (basicros.asm) and S/360 (aplros.asm) emulators many years ago. The listings are publicly available on our servers since 2008. Christian