XPost: sci.physics, comp.lang.prolog   
   From: janburse@fastmail.fm   
      
   Hi,   
      
   You wrote:   
      
    > No, they don't, they just add one (or some)   
    > more layer on top of it.   
      
   Techically they are not von Neuman architecture.   
   Unified Memory with Multiple Tensor Cores is   
   not von Neuman architecture. But the architecture   
      
   is possibly toned down by Data Flow, so that   
   in principle one can run the same thing on a   
   von Neuman architecture.   
      
   But in principle the architecture is rather:   
      
   parallel random-access machine (parallel RAM   
   or PRAM) is a shared-memory abstract machine.   
   https://en.wikipedia.org/wiki/Parallel_RAM   
      
   The above class of machines is not widely know.   
   But PRAM has been also studied, already in the 80's.   
      
   Bye   
      
   Maciej Woźniak schrieb:   
   > On 12/1/2025 11:25 AM, Mild Shock wrote:   
   >> Hi,   
   >>   
   >> 1) Classical computing = Boolean logic + von Neumann architecture   
   >   
   >   
   >   
   >>   
   >> For decades, all mainstream computation was built on:   
   >> Boolean algebra   
   >> Logic gates   
   >> Scalar operations executed sequentially   
   >> Memory and compute as separate blocks   
   >> Even floating-point arithmetic was implemented on top of Boolean logic.   
   >>   
   >> This shaped how programmers think — algorithms expressed   
   >> as symbolic operations, control flow, and discrete steps.   
   >>   
   >> 2) AI accelerators break from that model   
   >   
   > No, they don't, they just add one (or some)   
   > more layer on top of it.   
   >   
   > On the other hand, neural networks were   
   > always outside. So were quantum computers.   
   > It was never the only one and never the   
   > most powerful one.   
   >   
   >   
   >   
   >   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)