General Computational Accelerators

HPC Architecture
  1. Shared-memory SIMD machines
  2. Distributed-memory SIMD machines
  3. Shared-memory MIMD machines
  4. Distributed-memory MIMD machines
  5. ccNUMA machines
  6. Clusters
  7. Processors
    1. AMD Opteron
    2. IBM POWER7
    3. IBM BlueGene/Q processor
    4. Intel Xeon
    5. The SPARC processors
  8. Accelerators
    1. GPU accelerators
      1. ATI/AMD
      2. nVIDIA
    2. General computational accelerators
      1. Intel Xeon Phi
    3. FPGA accelerators
      1. Convey
      2. Kuberre
      3. SRC
  9. Interconnects
    1. Infiniband
Available systems
  • The Bull bullx system
  • The Cray XC30
  • The Cray XE6
  • The Cray XK7
  • The Eurotech Aurora
  • The Fujitsu FX10
  • The Hitachi SR16000
  • The IBM BlueGene/Q
  • The IBM eServer p775
  • The NEC SX-9
  • The SGI Altix UV series
  • Systems disappeared from the list
    Systems under development

    Although we have looked at the GPUs in the former section primarily from the point-of-view of computational accelerators, they are also full-blown high-end graphical processors in the first place. Several vendors have developed accelerators that did not have graphical processing in mind as the foremost application to be served (although they might not have been bad in this respect when compared to general CPUs). The future of the general computational accelerators seemed problematic: in principle it is entirely possible to make such accelerators that can compete with GPUs or with the FPGA-based accelerators discussed in section FPGA accelerators but the volume will always be much lower than that of the other two accelerator variants which is reflected in the production cost.

    The IBM/Sony/Toshiba Cell processor and the ClearSpeed accelerator have been discontinued for this reason but recently Intel has come up with its Xeon Phi accelerator which we will discuss on the Xeon Phi page.