From 0c8195a8faf2acaa4704408ee6fd364a42ab482d Mon Sep 17 00:00:00 2001 From: Augustin Degomme Date: Tue, 28 Aug 2018 16:32:41 +0200 Subject: [PATCH] Requalify automatic tesh, as another algorithm is used in init_smp now. --- .../coll-allreduce-automatic.tesh | 32 +++++++++---------- 1 file changed, 16 insertions(+), 16 deletions(-) diff --git a/teshsuite/smpi/coll-allreduce/coll-allreduce-automatic.tesh b/teshsuite/smpi/coll-allreduce/coll-allreduce-automatic.tesh index b8b77622c1..2a4d8f913c 100644 --- a/teshsuite/smpi/coll-allreduce/coll-allreduce-automatic.tesh +++ b/teshsuite/smpi/coll-allreduce/coll-allreduce-automatic.tesh @@ -19,22 +19,22 @@ $ ${bindir:=.}/../../../smpi_script/bin/smpirun -map -hostfile ../hostfile_coll > [rank 13] -> Ginette > [rank 14] -> Ginette > [rank 15] -> Ginette -> [ 0.427764] (8:7@Jupiter) The quickest allreduce was redbcast on rank 7 and took 0.007546 -> [ 0.427764] (5:4@Jupiter) The quickest allreduce was redbcast on rank 4 and took 0.007485 -> [ 0.427764] (7:6@Jupiter) The quickest allreduce was redbcast on rank 6 and took 0.007515 -> [ 0.427764] (6:5@Jupiter) The quickest allreduce was redbcast on rank 5 and took 0.007515 -> [ 0.427976] (14:13@Ginette) The quickest allreduce was mvapich2_two_level on rank 13 and took 0.007278 -> [ 0.427976] (13:12@Ginette) The quickest allreduce was mvapich2_two_level on rank 12 and took 0.007247 -> [ 0.427976] (16:15@Ginette) The quickest allreduce was ompi on rank 15 and took 0.007263 -> [ 0.427976] (15:14@Ginette) The quickest allreduce was mvapich2_two_level on rank 14 and took 0.007278 -> [ 0.429367] (2:1@Tremblay) The quickest allreduce was redbcast on rank 1 and took 0.006006 -> [ 0.429367] (3:2@Tremblay) The quickest allreduce was redbcast on rank 2 and took 0.006006 -> [ 0.429367] (4:3@Tremblay) The quickest allreduce was redbcast on rank 3 and took 0.006037 -> [ 0.430519] (12:11@Fafard) The quickest allreduce was mvapich2_two_level on rank 11 and took 0.006523 -> [ 0.430519] (10:9@Fafard) The quickest allreduce was mvapich2_two_level on rank 9 and took 0.006492 -> [ 0.430519] (9:8@Fafard) The quickest allreduce was mvapich2_two_level on rank 8 and took 0.006462 -> [ 0.430519] (11:10@Fafard) The quickest allreduce was mvapich2_two_level on rank 10 and took 0.006492 -> [ 0.434504] (1:0@Tremblay) For rank 0, the quickest was redbcast : 0.005991 , but global was mvapich2_two_level : 0.008672 at max +> [ 0.475403] (5:4@Jupiter) The quickest allreduce was redbcast on rank 4 and took 0.007485 +> [ 0.475403] (6:5@Jupiter) The quickest allreduce was redbcast on rank 5 and took 0.007515 +> [ 0.475403] (7:6@Jupiter) The quickest allreduce was redbcast on rank 6 and took 0.007515 +> [ 0.475403] (8:7@Jupiter) The quickest allreduce was redbcast on rank 7 and took 0.007546 +> [ 0.475616] (13:12@Ginette) The quickest allreduce was mvapich2_two_level on rank 12 and took 0.007247 +> [ 0.475616] (14:13@Ginette) The quickest allreduce was mvapich2_two_level on rank 13 and took 0.007278 +> [ 0.475616] (15:14@Ginette) The quickest allreduce was mvapich2_two_level on rank 14 and took 0.007278 +> [ 0.475616] (16:15@Ginette) The quickest allreduce was ompi on rank 15 and took 0.007263 +> [ 0.477007] (2:1@Tremblay) The quickest allreduce was redbcast on rank 1 and took 0.006006 +> [ 0.477007] (3:2@Tremblay) The quickest allreduce was redbcast on rank 2 and took 0.006006 +> [ 0.477007] (4:3@Tremblay) The quickest allreduce was redbcast on rank 3 and took 0.006037 +> [ 0.478158] (10:9@Fafard) The quickest allreduce was mvapich2_two_level on rank 9 and took 0.006492 +> [ 0.478158] (11:10@Fafard) The quickest allreduce was mvapich2_two_level on rank 10 and took 0.006492 +> [ 0.478158] (12:11@Fafard) The quickest allreduce was mvapich2_two_level on rank 11 and took 0.006523 +> [ 0.478158] (9:8@Fafard) The quickest allreduce was mvapich2_two_level on rank 8 and took 0.006462 +> [ 0.482143] (1:0@Tremblay) For rank 0, the quickest was redbcast : 0.005991 , 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 ] -- 2.20.1