Battling memory requirements of array programming through streaming

Research output: Contribution to journalConference articleResearchpeer-review

  • Mads Ruben Burgdorff Kristensen
  • Avery, James Emil
  • Troels Blum
  • Simon Andreas Frimann Lund
  • Brian Vinter
A barrier to efficient array programming, for example in Python/NumPy, is that algorithms written as pure array operations completely without loops, while most efficient on small input, can lead to explosions in memory use. The present paper presents a solution to this problem using array streaming, implemented in the automatic parallelization high-performance framework Bohrium. This makes it possible to use array programming in Python/NumPy code directly, even when the apparent memory requirement exceeds the machine capacity, since the automatic streaming eliminates the temporary memory overhead by performing calculations in per-thread registers.

Using Bohrium, we automatically fuse, JIT-compile, and execute NumPy array operations on GPGPUs without modification to the user programs. We present performance evaluations of three benchmarks, all of which show dramatic reductions in memory use from streaming, yielding corresponding improvements in speed and utilization of GPGPU-cores. The streaming-enabled Bohrium effortlessly runs programs on input sizes much beyond sizes that crash on pure NumPy due to exhausting system memory.
Original languageEnglish
Book seriesLecture notes in computer science
Pages (from-to)451-469
Number of pages19
Publication statusPublished - 2016
Event1st International Workshop on Performance Portable Programming Models for Accelerators - Frankfurt, Germany
Duration: 23 Jun 201623 Jun 2016
Conference number: 1


Conference1st International Workshop on Performance Portable Programming Models for Accelerators

ID: 178247789