p Test allreduce
! output sort
-$ ${bindir:=.}/../../../smpi_script/bin/smpirun -map -hostfile ../hostfile_coll -platform ../../../examples/platforms/small_platform.xml -np 16 --log=xbt_cfg.thres:critical ${bindir:=.}/coll-allreduce --log=smpi_kernel.thres:warning --log=smpi_coll.thres:error --cfg=smpi/allreduce:automatic --cfg=smpi/async-small-thresh:65536 --cfg=smpi/send-is-detached-thresh:128000 --cfg=smpi/simulate-computation:no "--log=root.fmt:[%10.6r]%e(%i:%P@%h)%e%m%n" --log=smpi_mpi.thres:error
-> [rank 0] -> Tremblay
-> [rank 1] -> Tremblay
-> [rank 2] -> Tremblay
-> [rank 3] -> Tremblay
-> [rank 4] -> Jupiter
-> [rank 5] -> Jupiter
-> [rank 6] -> Jupiter
-> [rank 7] -> Jupiter
-> [rank 8] -> Fafard
-> [rank 9] -> Fafard
-> [rank 10] -> Fafard
-> [rank 11] -> Fafard
-> [rank 12] -> Ginette
-> [rank 13] -> Ginette
-> [rank 14] -> Ginette
-> [rank 15] -> Ginette
-> [ 0.475378] (5:4@Jupiter) The quickest allreduce was redbcast on rank 4 and took 0.007485
-> [ 0.475378] (6:5@Jupiter) The quickest allreduce was redbcast on rank 5 and took 0.007515
-> [ 0.475378] (7:6@Jupiter) The quickest allreduce was redbcast on rank 6 and took 0.007515
-> [ 0.475378] (8:7@Jupiter) The quickest allreduce was redbcast on rank 7 and took 0.007546
-> [ 0.475591] (13:12@Ginette) The quickest allreduce was mvapich2 on rank 12 and took 0.007247
-> [ 0.475591] (14:13@Ginette) The quickest allreduce was mvapich2 on rank 13 and took 0.007278
-> [ 0.475591] (15:14@Ginette) The quickest allreduce was mvapich2 on rank 14 and took 0.007278
-> [ 0.475591] (16:15@Ginette) The quickest allreduce was ompi on rank 15 and took 0.007263
-> [ 0.476982] (2:1@Tremblay) The quickest allreduce was redbcast on rank 1 and took 0.006006
-> [ 0.476982] (3:2@Tremblay) The quickest allreduce was redbcast on rank 2 and took 0.006006
-> [ 0.476982] (4:3@Tremblay) The quickest allreduce was redbcast on rank 3 and took 0.006037
-> [ 0.478133] (10:9@Fafard) The quickest allreduce was mvapich2 on rank 9 and took 0.006492
-> [ 0.478133] (11:10@Fafard) The quickest allreduce was mvapich2 on rank 10 and took 0.006492
-> [ 0.478133] (12:11@Fafard) The quickest allreduce was mvapich2 on rank 11 and took 0.006523
-> [ 0.478133] (9:8@Fafard) The quickest allreduce was mvapich2 on rank 8 and took 0.006462
-> [ 0.482118] (1:0@Tremblay) For rank 0, the quickest was redbcast : 0.005991 , but global was mvapich2 : 0.008672 at max
+$ ${bindir:=.}/../../../smpi_script/bin/smpirun -map -hostfile ../hostfile_coll -platform ${platfdir:=.}/small_platform.xml -np 16 --log=xbt_cfg.thres:critical ${bindir:=.}/coll-allreduce --log=smpi_config.thres:warning --log=smpi_coll.thres:error --cfg=smpi/allreduce:automatic --cfg=smpi/async-small-thresh:65536 --cfg=smpi/send-is-detached-thresh:128000 --cfg=smpi/simulate-computation:no "--log=root.fmt:[%10.6r]%e(%i:%a@%h)%e%m%n" --log=smpi_mpi.thres:error --log=smpi_pmpi.thres:error --log=root.app:stdout
+> [ 0.000000] (0:maestro@) [rank 0] -> Tremblay
+> [ 0.000000] (0:maestro@) [rank 1] -> Tremblay
+> [ 0.000000] (0:maestro@) [rank 2] -> Tremblay
+> [ 0.000000] (0:maestro@) [rank 3] -> Tremblay
+> [ 0.000000] (0:maestro@) [rank 4] -> Jupiter
+> [ 0.000000] (0:maestro@) [rank 5] -> Jupiter
+> [ 0.000000] (0:maestro@) [rank 6] -> Jupiter
+> [ 0.000000] (0:maestro@) [rank 7] -> Jupiter
+> [ 0.000000] (0:maestro@) [rank 8] -> Fafard
+> [ 0.000000] (0:maestro@) [rank 9] -> Fafard
+> [ 0.000000] (0:maestro@) [rank 10] -> Fafard
+> [ 0.000000] (0:maestro@) [rank 11] -> Fafard
+> [ 0.000000] (0:maestro@) [rank 12] -> Ginette
+> [ 0.000000] (0:maestro@) [rank 13] -> Ginette
+> [ 0.000000] (0:maestro@) [rank 14] -> Ginette
+> [ 0.000000] (0:maestro@) [rank 15] -> Ginette
+> [ 0.516218] (8:7@Jupiter) The quickest allreduce was redbcast on rank 7 and took 0.008087
+> [ 0.516218] (7:6@Jupiter) The quickest allreduce was redbcast on rank 6 and took 0.008056
+> [ 0.516218] (6:5@Jupiter) The quickest allreduce was redbcast on rank 5 and took 0.008056
+> [ 0.516218] (5:4@Jupiter) The quickest allreduce was redbcast on rank 4 and took 0.008026
+> [ 0.516333] (4:3@Tremblay) The quickest allreduce was redbcast on rank 3 and took 0.008054
+> [ 0.516333] (3:2@Tremblay) The quickest allreduce was redbcast on rank 2 and took 0.008023
+> [ 0.516333] (2:1@Tremblay) The quickest allreduce was redbcast on rank 1 and took 0.008023
+> [ 0.516940] (13:12@Ginette) The quickest allreduce was mvapich2 on rank 12 and took 0.005970
+> [ 0.516940] (14:13@Ginette) The quickest allreduce was mvapich2 on rank 13 and took 0.006001
+> [ 0.516940] (15:14@Ginette) The quickest allreduce was mvapich2 on rank 14 and took 0.006001
+> [ 0.516940] (16:15@Ginette) The quickest allreduce was ompi on rank 15 and took 0.005970
+> [ 0.519494] (12:11@Fafard) The quickest allreduce was mvapich2 on rank 11 and took 0.006009
+> [ 0.519494] (11:10@Fafard) The quickest allreduce was mvapich2 on rank 10 and took 0.005978
+> [ 0.519494] (10:9@Fafard) The quickest allreduce was mvapich2 on rank 9 and took 0.005978
+> [ 0.519494] (9:8@Fafard) The quickest allreduce was mvapich2 on rank 8 and took 0.005948
+> [ 0.523490] (1:0@Tremblay) For rank 0, the quickest was redbcast : 0.008023 , but global was mvapich2 : 0.009199 at max
> [0] sndbuf=[0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 ]
> [1] sndbuf=[16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 ]
> [2] sndbuf=[32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 ]
