Home/ Questions/Q 8764191
In Process

The Archive Base Latest Questions

Editorial Team
  • 0
Asked: 2026-06-13T15:58:17+00:00

I’m benchmarking my computing cluster with High-Performance Linpack 2.0. I just completed 16 runs

  • 0

I’m benchmarking my computing cluster with High-Performance Linpack 2.0. I just completed 16 runs for which certain tuning parameters vary, but others do not. In particular, the four parameters listed with the FLOPS score in the output (N, NB, P, and Q) do not vary. Because of this, I do not know how to distinguish the results in terms of those parameters that do vary (PFACT, RFACT, BCAST, and DEPTH).

The HPL output in question can be found here

The corresponding input file

Any insights or ideas appreciated!

Share

Leave an answer

You must login to add an answer.

Need An Account,

1 Answer

  1. Editorial Team

    The parameters of each run are encoded in the value of the T/V field, e.g.

    WR00L2L4
 |||||||
 ||||||+-- NBMIN = 4
 |||||+--- PFACT = Left
 ||||+---- NDIV = 2
 |||+----- RFACT = Left
 ||+------ BCAST = 0 (1ring)
 |+------- DEPTH = 0
 +-------- possibly PMAP = Row-major

    Parameter values are rotated starting with the last one. When all possible values are tested, the value of the previous one is changed and so forth.

    • 0