VLSI Revolution Aided Parallel Computing

• Price/density advances in Si => multiprocessor computers were feasible

• SIMD computers continued to reign for technical reasons

  • Memory was still relatively expensive
  • Logic was still not dense enough for a high performance node
  • It’s how most architects were thinking

• Ken Batcher developed the Massively Parallel Procesor (MPP) for NASA with 16K procs

  • Danny Hillis built two machines CM-1,-2 scaling to 64K
  • MASPAR also sold a successful SIMD machine

Previous slide

Next slide

Back to first slide

View graphic version