CONCEPT

iWarp

  • iWarp

  • iWarp was a high-performance microprocessor architecture and systolic array supercomputer system developed jointly by Intel (under the Supercomputer Systems Division / SSD) and Carnegie Mellon University (CMU) under DARPA funding from 1988 through the early 1990s.
  • Architecture

    • Designed by CMU's H. T. Kung and Intel engineers, the iWarp processor combined a high-performance computation engine (capable of 20 MFLOPS single precision / 64-bit integer processing) with fine-grained, high-bandwidth communication hardware on a single chip.
    • The chips could be connected into a 2D torus or multidimensional mesh to execute parallel computation at supercomputing speeds, rivaling contemporary high-end processors like the early Pentiums.
  • Sysadmin & Infrastructure

    • Randal Schwartz served as the system administrator for the ~20 engineers developing the iWarp chip in Oregon, managing Sun Microsystems workstations and establishing Intel's second direct 56k Internet connection to link Oregon engineers directly to CMU and the broader Internet.
  • See Sysadmin on the iWarp Supercomputer Project and Bringing the Internet to Intel Oregon.
These facts are as Randal recalls them, but much time has passed for most of this. If you find a factual error, please email realmerlyn@gmail.com.