I actually traveled to Japan to work on an Amdahl machine installed there. We visited the factory where the base machines were built and then sent to Amdahl for their modifications.
My time at Amdahl was fantastic. Best technology (PERIOD) and some of the best people I ever worked with. We pushed like crazy to have Fujitsu move from 31-bit to 64-bit and keep competing with the new CMOS machines. However, FJs fascination with high-end unix ended that dream. I left Amdahl/Fujitsu America after a small amount of time working on the unit stuff. Jon Nolting System Administrator Engineering IT jon.nolt...@oracle.com 425-295-1733 (Cell) -----Original Message----- From: IBM Mainframe Discussion List <IBM-MAIN@LISTSERV.UA.EDU> On Behalf Of Tom Marchant Sent: Monday, July 31, 2023 4:30 PM To: IBM-MAIN@LISTSERV.UA.EDU Subject: [External] : Re: Mainframe Makers.... WAS: Ars Technica: The IBM mainframe: How it runs and why it survives On Mon, 31 Jul 2023 16:29:22 -0400, Steve Thompson <ste...@wkyr.net> wrote: >Fujitsu did not "buy" Amdahl machines, Phil didn't say that Fujitsu bought Amdahl machines. He said that they bought Amdahl. This is true. >Fujitsu supplied Amdahl with their machines I worked for Amdahl too, from 1978 to 1984. I started as a field Systems Engineer, often finding and occasionally fixing bugs in MVS. When MVS abended on an Amdahl machine, IBM would take the position that it must be the hardware, unless the customer could reproduce it in an IBM machine.Then I cross-trained to hardware, then an SE Specialist. During that time frame, Fujitsu did not supply Amdahl with their machines. Amdahl designed and built their own machines. IIRC Amdahl designed the chips. I don't remember who fabricated the chips, but it might have been Fujitsu. Probably other components were supplied by Fujitsu as well. >with the MODs we (yeah, I worked for Amdahl >prior to 1990) asked for/needed, and then for instructions we >didn't have micro-store for, Micro-store? There was no micro-store on the 470 or 580 (5850,5860, 5870 and 5880) systems. All instructions were implemented entirely in hardware. On the 470 series, that caused Amdahl to be at a disadvantage when IBM added new instructions. Instead, Amdahl used software emulation for the new instructions. The first of these was MVS/SE Assist, an enhancement to the Program Interruption First-Level Interrupt Handler. It would detect the PIC 1 and emulate the instruction in software if it was something that it handled. The 580 had a radical new design. During a three month stint at headquarters, I worked with the 580 console project and I had my own, numbered and registered copy of the ALTA Principles of operation. Among other things, it defined a mechanism to permit another level of virtualization, allowing 4 Domains to be defined and mapping System storage to domain storage. Sometimes I wish I had "forgotten" to return it when I left... To manage it, there was a new state, System state, in addition to Problem and Supervisor state. System state registers registers for use only when in System state. Special System state instructions to do things like moving data between the normal registers and the system state registers. The design included 31 or 32 bit memory (I forget which) and a much improved channel subsystem. When 370/XA came out a year or two later, I looked in the XA POO for anything that didn't more or less fit in the ALTA design and didn't find anything. Macrocode ran in System state IIRC it was loaded into System storage by the console processor. The console on the 580 was a separate 370 processor on one MCC that ran the UTS flavor of Unix. That was my first exposure to Unix. Macrocode mapped System storage to domain storage and system channels to domain channels. All interruptions went through Macrocode. New instructions could be simulated by Macrocode >we used FAM (Fast Assist Mode) which >we then emulated instructions (part of MacroCode). > ... > >And I still think my time at Amdahl was the best job and >education in machine hardware I could have ever had for the short >time I was there. Same here. And the training was top notch. -- Tom Marchant ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN
