######################################################################################
# Saiph, scalability experiment:
#
# App:Heat 3D - garlic/tampi+isend+oss+task+simd branch
# App details:
#   3D mesh of ~400*400*400 points
#   nbgx = global blocks in the X dimension
#   nbgy = global blocks in the Y dimension
#   nbgz = global blocks in the Z dimension
#     --> nbgx*nbgy*nbgz = global distributed blocks
#   nbly = local blocks in the Y dimension
#   nblz = local blocks in the Z dimension
#     --> nbly*nblz = local blocks (#tasks)
#   
# Scalability experiment configuration:
#   From a single-core granularity experiment, use a suited local blocking set:
#     --> nbly*nblz >= 48   (at least 3tasks/proc) 
#   MPI binded to sockets: MPI procs = 2*nodes
#   Mesh distributed across third dimension to ensure contiguous communications
#     --> nbgx = 1, nbgy = 1
#   Global distribution limited by global mesh size 
#   First dimension cannot be locally blocked (simd reasons)
#   Second and third dimension local blocking limited by local mesh size 
# 
######################################################################################

# Common packages, tools and options
{ 
  stdenv
, stdexp
, bsc
, targetMachine
, stages
, garlicTools
}:

with stdenv.lib;
with garlicTools;

let

  #*** Variable configurations ***
  varConf = with targetMachine.config; {
    # FIXME: None of those selected nbl* and problem size is able to give good
    # efficiency when testing strong scaling. We should find better values.
    # Local blocks per dimension
    nblx = [ 1 ]; # SIMD
    nbly = [ 32 ];
    nblz = [ 8 ];
    sizex = [ 3 6 ];
    gitBranch = [ "garlic/tampi+isend+oss+task+simd" ];
    nodes = range2 1 8;
  };

  #*** Generate the complete configuration for each unit ***
  genConf = c: targetMachine.config // rec {

    # Experiment, units and job names 
    expName = "saiph-ss";
    unitName = "${expName}"
      + "-N${toString nodes}"
      + "-nbg.x${toString nbgx}.y${toString nbgy}.z${toString nbgz}"
      + "-nbl.x${toString nblx}.y${toString nbly}.z${toString nblz}";
    jobName = unitName;

    # saiph options
    enableManualDist = true;    # allows to manually set nbg{x-y-z}
    nbgx = 1;
    nbgy = 1;
    nbgz = nodes * ntasksPerNode;    # forcing distribution by last dim

    inherit (c) nblx nbly nblz nodes sizex;

    gitBranch = "garlic/tampi+isend+oss+task+simd";
    gitCommit = "8052494d7dc62bef95ebaca9938e82fb029686f6";  # fix a specific commit 

    blocksPerTask = nblx * nbly * nblz * 1.0;
    blocksPerCpu = blocksPerTask / cpusPerTask;

    # Repeat the execution of each unit 10 times
    loops = 10;

    # Resources
    inherit (targetMachine.config) hw;

    qos = "debug";
    ntasksPerNode = hw.socketsPerNode;   # MPI binded to sockets
    cpusPerTask = hw.cpusPerSocket;      # Using the 24 CPUs of each socket
  };

  #*** Compute the final set of configurations ***
  # Compute the array of configurations: cartesian product of all factors
  configs = stdexp.buildConfigs {
    inherit varConf genConf;
  };

  #*** Sets the env/argv of the program ***
  exec = {nextStage, conf, ...}: with conf; stages.exec {
    inherit nextStage;
    env = ''
      export OMP_NUM_THREADS=${toString hw.cpusPerSocket}
    '';
  };

  #*** Configure the program according to the app ***
  program = {nextStage, conf, ...}: bsc.apps.saiph.override {
    inherit (conf) enableManualDist
      nbgx nbgy nbgz nblx nbly nblz
      sizex
      gitBranch gitCommit;

    L3SizeKB = conf.hw.cacheSizeKB.L3;
    cachelineBytes = conf.hw.cachelineBytes;
  };

  #*** Add stages to the pipeline ***
  pipeline = stdexp.stdPipeline ++ [ exec program ];

in

  stdexp.genExperiment { inherit configs pipeline; }