HPLinpack benchmarking

[tle]

Use https://github.com/geerlingguy/top500-benchmark

Code: [Select]

$ lscpu
Architecture:             ppc64le
  Byte Order:             Little Endian
CPU(s):                   32
  On-line CPU(s) list:    0-31
Model name:               POWER9, altivec supported
  Model:                  2.3 (pvr 004e 1203)
  Thread(s) per core:     4
  Core(s) per socket:     8
  Socket(s):              1
  Frequency boost:        enabled
  CPU(s) scaling MHz:     100%
  CPU max MHz:            3800.0000
  CPU min MHz:            2166.0000
Caches (sum of all):     
  L1d:                    256 KiB (8 instances)
  L1i:                    256 KiB (8 instances)
  L2:                     4 MiB (8 instances)
  L3:                     80 MiB (8 instances)
NUMA:                     
  NUMA node(s):           1
  NUMA node0 CPU(s):      0-31
Vulnerabilities:         
  Gather data sampling:   Not affected
  Itlb multihit:          Not affected
  L1tf:                   Mitigation; RFI Flush, L1D private per thread
  Mds:                    Not affected
  Meltdown:               Mitigation; RFI Flush, L1D private per thread
  Mmio stale data:        Not affected
  Reg file data sampling: Not affected
  Retbleed:               Not affected
  Spec rstack overflow:   Not affected
  Spec store bypass:      Mitigation; Kernel entry/exit barrier (eieio)
  Spectre v1:             Mitigation; __user pointer sanitization, ori31 speculation barrier enabled
  Spectre v2:             Mitigation; Software count cache flush (hardware accelerated), Software link stack flush
  Srbds:                  Not affected
  Tsx async abort:        Not affected

Code: [Select]

$ ansible-playbook main.yml --tags "setup,benchmark" --ask-become-pass
  mpirun_output.stdout: |-
    ================================================================================
    HPLinpack 2.3  --  High-Performance Linpack benchmark  --   December 2, 2018
    Written by A. Petitet and R. Clint Whaley,  Innovative Computing Laboratory, UTK
    Modified by Piotr Luszczek, Innovative Computing Laboratory, UTK
    Modified by Julien Langou, University of Colorado Denver
    ================================================================================
 
    An explanation of the input/output parameters follows:
    T/V    : Wall time / encoded variant.
    N      : The order of the coefficient matrix A.
    NB     : The partitioning blocking factor.
    P      : The number of process rows.
    Q      : The number of process columns.
    Time   : Time in seconds to solve the linear system.
    Gflops : Rate of execution for solving the linear system.
 
    The following parameter values will be used:
 
    N      :   70717
    NB     :     256
    PMAP   : Row-major process mapping
    P      :       1
    Q      :      32
    PFACT  :   Right
    NBMIN  :       4
    NDIV   :       2
    RFACT  :   Crout
    BCAST  :  1ringM
    DEPTH  :       1
    SWAP   : Mix (threshold = 64)
    L1     : transposed form
    U      : transposed form
    EQUIL  : yes
    ALIGN  : 8 double precision words
 
    --------------------------------------------------------------------------------
 
    - The matrix A is randomly generated for each test.
    - The following scaled residual check will be computed:
          ||Ax-b||_oo / ( eps * ( || x ||_oo * || A ||_oo + || b ||_oo ) * N )
    - The relative machine precision (eps) is taken to be               1.110223e-16
    - Computational tests pass if scaled residuals are less than                16.0
 
    ================================================================================
    T/V                N    NB     P     Q               Time                 Gflops
    --------------------------------------------------------------------------------
    WR11C2R4       70717   256     1    32            1650.43             1.4286e+02
    HPL_pdgesv() start time Thu Jun 13 15:57:05 2024
 
    HPL_pdgesv() end time   Thu Jun 13 16:24:36 2024
 
    --------------------------------------------------------------------------------
    ||Ax-b||_oo/(eps*(||A||_oo*||x||_oo+||b||_oo)*N)=   2.41238455e-03 ...... PASSED
    ================================================================================
 
    Finished      1 tests with the following results:
                  1 tests completed and passed residual checks,
                  0 tests completed and failed residual checks,
                  0 tests skipped because of illegal input values.
    --------------------------------------------------------------------------------
 
    End of Tests.
    ================================================================================

PLAY RECAP *********************************************************************************************************************************************************************************************************************************************************************
127.0.0.1                  : ok=29   changed=10   unreachable=0    failed=0    skipped=7    rescued=0    ignored=0   

[cy384]

After messing with the parameters a bit, I can get my 16 core blackbird up to 2.1683e+02, I wonder if it's limited by RAM bandwidth (only the two channels).

[tle]

After messing with the parameters a bit, I can get my 16 core blackbird up to 2.1683e+02, I wonder if it's limited by RAM bandwidth (only the two channels).

What is your `P` and `Q` parameters?