# use the tested library, not the installed one
# (since we want to pass it to the child, it has to be redefined before each command)
# Go for the first test

p Test instrumentation of SMPI

$ ${bindir:=.}/../../../smpi_script/bin/smpirun -hostfile ${srcdir:=.}/../hostfile -platform ${platfdir:=.}/small_platform.xml --cfg=path:${srcdir:=.}/../msg --log=smpi_kernel.thres:warning --log=xbt_cfg.thres:warning --cfg=smpi/host-speed:1 -np 8 ${bindir:=.}/smpi_gemm 1000 native
> You requested to use 8 ranks, but there is only 5 processes in your hostfile...
> Matrix Size : 1000x1000
> Native mode
> Performance= 227.39 GFlop/s, Time= 8.795 sec, Size= 2000000000 Ops

$ ${bindir:=.}/../../../smpi_script/bin/smpirun -hostfile ${srcdir:=.}/../hostfile -platform ${platfdir:=.}/small_platform.xml --cfg=path:${srcdir:=.}/../msg --log=smpi_kernel.thres:warning --log=xbt_cfg.thres:warning --cfg=smpi/host-speed:1 -np 8 ${bindir:=.}/smpi_gemm 1000 sampling
> You requested to use 8 ranks, but there is only 5 processes in your hostfile...
> Matrix Size : 1000x1000
> Sampling mode
> Performance= 227.39 GFlop/s, Time= 8.795 sec, Size= 2000000000 Ops