From: ldo@nz.invalid   
      
   On Sat, 20 Sep 2025 21:13 +0100 (BST), John Dallman wrote:   
      
   > ... the IBM Z instruction set only has three instruction lengths -   
   > 2, 4 and 6 bytes, which has not changed since System/360 - and you   
   > can always discover the length of each instruction from its first   
   > two bytes. That makes having multiple instructions being decoded   
   > simultaneously easier, which is a bottleneck in x86 and x86-64, the   
   > other long-lasting CISC instruction set.   
      
   Mainframes were never designed for high CPU performance.   
      
   Look at the current Top500 list of the world’s fastest machines; what   
   architectures do you see? IBM POWER offers a few contenders; also ARM,   
   I think MIPS, and of course the most common is x86-64. At some point   
   no doubt a RISC-V machine is likely to make an appearance.   
      
   No IBM Z. Not before, not now, not ever.   
      
   > I talked to a colleague, who returned to my employer after a   
   > takeover, and remembers our business in the early 1980s. He's   
   > perfectly clear that VMS was a far better OS for technical computing   
   > than any of the proprietary minicomputer OSes of the time, all of   
   > which are dead. But VAX couldn't match the performance of high-end   
   > 68000 Unix machines, followed by the RISCs, and the rest is history.   
      
   Presumably your colleague was talking only about non-Unix systems?   
   After all, what were “Unix workstations” if not platforms for   
   “technical computing”?   
      
   And while official “Unix” did last into the 1990s and early ’00s   
   before (mostly) expiring, the spirit lives on in Linux today, to the   
   extent that both Microsoft and Apple have decided that they need to   
   add support for running actual Linux kernels on their respective   
   proprietary platforms.   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)