conferences/ics2022.mdwn

   1 # ICS2022
   2
   3 27 June 2022
   4
   5 * <https://ics2022.github.io/>
   6 * <https://meep-project.eu/form/workshop-on-risc-v-and-openpower>
   7 * <https://meep-project.eu/events/3rd-workshop-risc-v-and-openpower-hpc>
   8 * <https://libre-soc.org/openpower/sv/SimpleV_rationale/>
   9
  10 # Luke Leighton bio
  11
  12 Luke Kenneth Casson Leighton specialises in Libre Ethical Technology.
  13 He has been using, programming and reverse-engineering computing
  14 devices continuously for 44 years, has a BEng (Hons), ACGI, in
  15 Theory of Computing from Imperial College, and recently put that
  16 education to good use in the form of the Libre-SOC
  17 Project: an entirely Libre-Licensed 3D Hybrid CPU-VPU-GPU based on
  18 OpenPOWER. He writes poetry and has been developing a HEP Physics theory
  19 for the past 36 years in his spare time.
  20
  21 # Coherent Distributed Computing: ZOLC, SVP64, OpenCAPI
  22
  23 Deterministic Scheduled Zero-Overhead Loops have a startling property:
  24 their deterministic nature allows them to be distributed. Extra-V began
  25 to illustrate the potential here, by performing near-Memory Coherent
  26 and conditional Graph walking, making full use of OpenCAPI's potential.
  27 Snitch also led the way, bringing back Auto-increment Load/Store from
  28 the CISC era, but hidden behind Tagged Registers connected to
  29 Coherent FIFOs leading indirectly to main Memory.  Where both Snitch
  30 and Extra-V used limited variants of Deterministic Loops as
  31 proof-of-concept to support their overall research, with only rudimentary
  32 processing capability,
  33 ZOLC is a much more deeply extensive and well-defined Deterministic Loop
  34 Control system that can fit directly on top of a standard ISA.
  35
  36 SVP64 takes the Zero-Overhead Loop concept firmly into Supercomputing
  37 Vector Processing territory, currently limited to the register file.
  38 This talk explores the potential of combining Snitch and Extra-V's
  39 pioneering techniques, combining SVP64 and ZOLC, and leveraging OpenCAPI,
  40 on top of the OpenPOWER ISA, to create High-performance Coherent
  41 Distributed Computing with the potential to run large-scale Parallel
  42 Compute tasks at 100% sustained throughput whilst also bringing the potential
  43 of Snitch's 85% power-consumption-reduction to bear, using assembly intrinsics at
  44 in a normal everyday ubiquitous Software environment: no specialist
  45 parallel programming languages or special compilers needed.
  46