From: "Frode van der Meeren" Subject: Re: Intel ISIS 1 executable file format to: Herb Johnson date Oct 26 2023 On 10/26/2023 7:34 PM, Frode van der Meeren wrote: --------------------------------------------------- I have looked a little around on your webpage, and it seems like you know a great deal about the early Intel ISIS history. As of late I have been interested in a vintage Norwegian computer by Tandberg Data, the TDV-2114. The more and more I dig, the more I see they seems to have been very much inspired by the Intel Intellec series II and ISIS. For instance, there is a ROM monitor with many of the same commands than the Intellec monitor (incl. file-support for the ISIS-II ss-sd disk format), the operating system built on top of this has the same syntax and device-files as ISIS-II, it runs a 8080 CPU, and the physical layout of the machine looks a bit like some of the Series II workstations. However, it seems like the executable file structure, among other things, is different. There was some mention somewhere that the executables in TOS ("Tandberg OS") might be an earlier "fastload" format used by ISIS-1.0 and 1.2 instead of the object module files commonly used in ISIS-II. There are no "record type" bytes, an executable file is just a series of records all headed with nothing but each their start address pointer and record length word-pair. The last record usually has the size of zero, and when the load of the file ends the address pointer stored in this empty record is used as a call-pointer for where to start execution. Now, I have been trying to find technical information about ISIS 1.x, but I find about nothing. Do you have any more information here? Could you confirm if my suspicion is correct? There are other clues that there certainly was some collaboration between Tandberg and Intel. For instance, Intel made or licensed the design of one of the memory-cards used in some of the TDV-2114. These cards support both the 2109 and 2117 (4116) type DRAM, and of course has the Intel logo in the corner. It would be very interesting to see if there were more similarities between these two architectures, for instance in the early versions of the Tandberg monitor system-call API. However, I have too little references of the inner workings of ISIS to tell. With best regards. -Frode van der Meeren on 27. oktober 2023 Herbert Johnson said ------------------------------------------ Thanks for your inquiry. There's been some work in the last year or two about early Intel ISIS. also the Tandberg computer. It's been discussed in the Google Group https://groups.google.com/g/intel-devsys I see your Facebook/Tandberg post was discussed in the Google Group. I also found by search in that group, a subject line "ISIS 1.1 and 2.2" Apr 28, 2021 and later. There's reference to Mark Ogden's archive in an Apr 29, 2021 post. I participated in the discussion. I don't think I migrated these results to my retrotechnology Web page. Mark Ogden has an old WEb site, not often maintained currently: http://mark-ogden.uk/files/intel/publications/The%20Greatest%20Microcomputer%20Fair%20Ever-May76.pdf A fair amount of current content is in Mark's github site: Mark comments there about ISIS V1.1: https://github.com/ogdenpm A recent addition to the repository is ISIS v1.1 (16k ISIS). I am not aware of this being elsewhere on the internet. It was uncovered in early September 2020 by Jon Hales a volunteer at the Cambridge Centre for Computing History... https://github.com/ogdenpm/intel80tools/tree/master/itools/isis The disk image for V1.1 is there. So the file contents are as80, attrib, copy, delete, dir, edit, format, etc., and the disk image. https://github.com/ogdenpm/intel80tools/tree/master/src/isis_1.1 Is apparently the disassembled/decompiled PL/M-80 sources. Hopefully you can examine the ISIS OS calls from their destinations in the OS source, and from calls by various OS utilities. I don't know if Mark reconstructed any ISIS 1.1 API documentation. Maybe you can do that while looking at the source. Maybe the Tanberg documentation will inform you on ISIS! ?? The PL/M sources will be hard to recompile, they may need various versions of Intel's PL/M compiler (like the Fortran cross compiler) and so on. But you can run the image with (whatever emulator Mark discusses, maybe other emulators) and execute the programs... https://github.com/ogdenpm/thames-src .. and read the source to understand them. Mark has become adept in reading 8080 binaries to extract the PL/M-80 structures in them. Of course - talk to Mark Ogden about all this, it's his work, I'm just reading files tonight. Join the intel devsys discussion. You said in your email: For instance, Intel made or licensed the design of one of the memory-cards used in some of the TDV-2114. Well, Intel was started as a memory company, for minicomputers and mainframes, later microcomputers. They made many memory boards for different computers of the era, to sell their memory chips! So did other semiconductor companies. Microprocessors were ways to sell more memory! Regards Herb Johnson On 10/27/2023 12:36 AM, Frode van der Meeren wrote: -------------------------------------------------- Thanks a lot for the links! I will go through them and have a look. What I can tell right off the bat, looking at the files from the dump of the ISIS 1.1 disk, yes: Those files unmistakingly use the same format as executable files in Tandberg OS. From here on, it will be interesting to see to what extent the Monitor API is cross-compatible. Also, looking at the standard accessories on the disk, while most are named the same as in TOS, most of the versions from Intel seems to be compiled from high-level language, each executable containing a lot of library boiler-plate code, while the TOS versions are tiny in comparison (probably written directly in assembler). For example, Intel's hexbin is 3.4KB in size while Tandberg's hexbin is merely just over 500 bytes. With best regards -Frode ON 27. oktober 2023 Herbert Johnson said ------------------------------------------- Thanks for the quick analysis of Mark's work. With your permission I'll add your notes and mine to my ISIS Web page. Again, I encourage you to discuss these matters in the Google Group, or at least contact Mark for his informed responses. I'll want to point-to your Tandberg work on the Web. I found the links below, and I found a few of your posts in some vintage computing groups. I did not find a Web site with your content. Do you have Tandberg content somewhere on the Web? http://www.datormuseum.se/documentation-software/isis-ii-floppy-disks.html (Mattis Lind Web site) which appears to be two Tandberg disk images; http://heim.bitraf.no/tingo/files/other/ http://heim.bitraf.no/tingo/td/tandberg-data.html scans of Tanberg Data newsletters and some photos of products (a private archive) On your observation that "assembler is smaller than high-level-language (HLL)". That was a common observation in the era where HLL's were first in use on microprocessors. One consideration, is how compilers and linkers included libraries. If an entire library was included, that's more code than only linking as-used library modules. You can confirm this, but a 3.4Kb executable in PL/M, was likely mostly library code, plus dozens of lines of actual program. That may have been faster & easier to produce, than a tight .5K 8080 assembler program of a few hundred developer-written instructions. Kildall promoted PL/M as more efficient *for programmers* than assembly. Regards Herb On 10/27/2023 12:36 AM, Frode van der Meeren wrote: --------------------------------------------------- Of course, please post this info. I will look into join the Google group when I have the time, this weekend is super busy. Most of my posting has been in this thread overfor at a German forum: https://forum.classic-computing.de/forum/index.php?thread/25639-siemens-6-610-reparatur-gegl%C3%BCckt/ Tandberg as one single company went bankrupt in 1978, just two years after this computer was developed. In that time, Siemens had gotten interested, and bought majority ownership in the computer-division of the old Tandberg company after the bankrupcy. A result of this is the Siemens OEM of the TDV-2114 (called "System 6.610"), but Tandberg kept selling their own machines still. Later on the Tandberg terminals would use a lot more parts from Siemens, obviously. Most of my focus till now has been on the hardware, and that forum thread does link to some scanned scematics (but not all). It also links to the datormuseum page, but I think only one of those two disks are for the TDV-2114. The programs on the other disk seems to be OMF format and from Intel. -Frode [Note: Frode also posted something in Facebook in early Sept 2023, about his Tandberg Dta TDV-2114. So there's some discussions there, somewhere. - Herb]