! setenv LD_LIBRARY_PATH=../../lib
p Test allreduce
-$ ${bindir:=.}/../../../bin/smpirun -map -hostfile ../hostfile_coll -platform ../../../examples/platforms/small_platform.xml -np 16 --log=xbt_cfg.thres:critical ${bindir:=.}/coll-allreduce --log=smpi_kernel.thres:warning --log=smpi_coll.thres:error --cfg=smpi/allreduce:automatic --cfg=smpi/async_small_thresh:65536 --cfg=smpi/send_is_detached_thresh:128000 --cfg=smpi/running_power:-1 "--log=root.fmt:[%10.6r]%e(%i:%P@%h)%e%m%n"
+! output sort
+$ ${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"
> [rank 0] -> Tremblay
> [rank 1] -> Tremblay
> [rank 2] -> Tremblay
> [rank 13] -> Ginette
> [rank 14] -> Ginette
> [rank 15] -> Ginette
-> [ 0.420156] (8:7@Jupiter) The quickest allreduce was redbcast on rank 7 and took 0.007545
-> [ 0.420156] (5:4@Jupiter) The quickest allreduce was redbcast on rank 4 and took 0.007484
-> [ 0.420156] (7:6@Jupiter) The quickest allreduce was redbcast on rank 6 and took 0.007514
-> [ 0.420156] (6:5@Jupiter) The quickest allreduce was redbcast on rank 5 and took 0.007514
-> [ 0.420368] (14:13@Ginette) The quickest allreduce was mvapich2_two_level on rank 13 and took 0.007278
-> [ 0.420368] (13:12@Ginette) The quickest allreduce was mvapich2_two_level on rank 12 and took 0.007247
-> [ 0.420368] (16:15@Ginette) The quickest allreduce was ompi on rank 15 and took 0.007263
-> [ 0.420368] (15:14@Ginette) The quickest allreduce was mvapich2_two_level on rank 14 and took 0.007278
-> [ 0.421760] (2:1@Tremblay) The quickest allreduce was redbcast on rank 1 and took 0.006005
-> [ 0.421760] (3:2@Tremblay) The quickest allreduce was redbcast on rank 2 and took 0.006005
-> [ 0.421760] (4:3@Tremblay) The quickest allreduce was redbcast on rank 3 and took 0.006036
-> [ 0.422911] (12:11@Fafard) The quickest allreduce was mvapich2_two_level on rank 11 and took 0.006523
-> [ 0.422911] (10:9@Fafard) The quickest allreduce was mvapich2_two_level on rank 9 and took 0.006492
-> [ 0.422911] (9:8@Fafard) The quickest allreduce was mvapich2_two_level on rank 8 and took 0.006462
-> [ 0.422911] (11:10@Fafard) The quickest allreduce was mvapich2_two_level on rank 10 and took 0.006492
-> [ 0.426896] (1:0@Tremblay) For rank 0, the quickest was redbcast : 0.005990 , but global was mvapich2_two_level : 0.008672 at max
+> [ 0.427747] (5:4@Jupiter) The quickest allreduce was redbcast on rank 4 and took 0.007484
+> [ 0.427747] (6:5@Jupiter) The quickest allreduce was redbcast on rank 5 and took 0.007514
+> [ 0.427747] (7:6@Jupiter) The quickest allreduce was redbcast on rank 6 and took 0.007514
+> [ 0.427747] (8:7@Jupiter) The quickest allreduce was redbcast on rank 7 and took 0.007545
+> [ 0.427959] (13:12@Ginette) The quickest allreduce was mvapich2_two_level on rank 12 and took 0.007247
+> [ 0.427959] (14:13@Ginette) The quickest allreduce was mvapich2_two_level on rank 13 and took 0.007278
+> [ 0.427959] (15:14@Ginette) The quickest allreduce was mvapich2_two_level on rank 14 and took 0.007278
+> [ 0.427959] (16:15@Ginette) The quickest allreduce was ompi on rank 15 and took 0.007263
+> [ 0.429351] (2:1@Tremblay) The quickest allreduce was redbcast on rank 1 and took 0.006005
+> [ 0.429351] (3:2@Tremblay) The quickest allreduce was redbcast on rank 2 and took 0.006005
+> [ 0.429351] (4:3@Tremblay) The quickest allreduce was redbcast on rank 3 and took 0.006036
+> [ 0.430502] (10:9@Fafard) The quickest allreduce was mvapich2_two_level on rank 9 and took 0.006492
+> [ 0.430502] (11:10@Fafard) The quickest allreduce was mvapich2_two_level on rank 10 and took 0.006492
+> [ 0.430502] (12:11@Fafard) The quickest allreduce was mvapich2_two_level on rank 11 and took 0.006523
+> [ 0.430502] (9:8@Fafard) The quickest allreduce was mvapich2_two_level on rank 8 and took 0.006462
+> [ 0.434487] (1:0@Tremblay) For rank 0, the quickest was redbcast : 0.005990 , but global was mvapich2_two_level : 0.008672 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 ]