HPC White Papers

  • Measures how well something is doing.

  • Isambard  analysis of HPC-optimized Arm processors

    Performance results from Isambard, the first production supercomputer to be based on Arm CPUs that have been optimized specifically for HPC. 

    Learn more
  • An item changing its size.

  • A sneak peek into SVE and VLA programming

    An overview of SVE with information on the new registers and the new instructions, the Vector Length Agnostic (VLA) programming technique with some examples.

    Learn more
  • Measures how well something is doing.

  • HPC Case Study: CFD Applications on Arm

    We examine the readiness and potential of Arm-based platforms for HPC, and have benchmarked two different computational fluid dynamics (CFD) applications.

    Download
  • Applications on a computer.

  • Arm SVE and application to Machine Learning

    Download this white paper to learn how vector-length-agnostic techniques introduced by SVE can be used to efficiently vectorize General Matrix Multiplication (GEMM) and low precision GEMM (GEMMlowp).

    Learn more
  • ARM C/C++ Compiler

  • How the Flang Frontend Works

    The ultimate goal for Flang is to become part of the LLVM ecosystem, receiving the same level of support as the Clang frontend. To support this goal, this paper explains the Flang frontend interior, alongside giving a description of how the source code is structured.

    Download

  • Big Data Case Study: Spark on Armv8

    We examine the readiness of Arm-based platforms for two prominent big data technologies, namely Spark and Hadoop.

    Download
  • Text: arm (logo).

  • GW4 Alliance puts Arm architecture to the test for HPC

    Isambard is a new Tier 2 HPC service for UK-based scientists. It has been designed to provide the first Arm-based Cray XC50 ‘Scout’ system, combining Cavium ThunderX2 CPUs with Cray’s Aries interconnect, as part of the UK’s national HPC service.

    Learn more

Published papers

  • Measures how well something is doing.

  • On the benefits of tasking with OpenMP

    Tasking promises a model to program parallel applications that provides intuitive semantics. In the case of tasks with dependences, it also promises better load balancing by removing global synchronizations (barriers), and potential for improved locality. 

    Learn more