1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
|
# memtype=all will iterate all memtype for even processes and set odd processes to device
allred 4 arg=-counts=10,100 arg=-memtype=all
allred 7 arg=-counts=10 arg=-memtype=all
allred2 4 arg=-memtype=all
allred2 12 arg=-memtype=all
reduce 5 arg=-memtype=all
reduce 7 arg=-memtype=all
op_coll 4 arg=-memtype=all
# ZE_AFFINITY_MASK tests
allred2 8 arg=-memtype=all env=MTEST_GPU_VISIBILITY_AFFINITY=CONTIGUOUS_DEVICE
allred2 8 arg=-memtype=all env=MTEST_GPU_VISIBILITY_AFFINITY=CONTIGUOUS_SUBDEVICE
allred2 8 arg=-memtype=all env=MTEST_GPU_VISIBILITY_AFFINITY=CONTIGUOUS_SINGLE_SUBDEVICE
allred2 8 arg=-memtype=all env=MTEST_GPU_VISIBILITY_AFFINITY=CONTIGUOUS_MULTI_SUBDEVICE
# The dtp test will iterate over all typelist and counts, each time will repeat [repeat] times and select seed, testsize, memtypes accordingly
# set MPITEST_VERBOSE=1 to the list of tests being run.
# bcast 4 arg=-typelist=MPI_INT,MPI_INT:4+MPI_DOUBLE:8 arg=-counts=1,17,50,100,512,65530 arg=-seed=100 arg=-testsizes=2,50 arg=-memtype=random timeLimit=600
bcast 4 arg=-typelist=MPI_INT,MPI_INT:4+MPI_DOUBLE:8 arg=-counts=1,17,50,100,512,65530 arg=-seed=100 arg=-testsizes=2,50 arg=-evenmemtype=device arg=-oddmemtype=device
bcast 4 arg=-typelist=MPI_INT,MPI_INT:4+MPI_DOUBLE:8 arg=-counts=1,17,50,100,512,65530 arg=-seed=100 arg=-testsizes=2,50 arg=-evenmemtype=device arg=-oddmemtype=host
bcast 4 arg=-typelist=MPI_INT,MPI_INT:4+MPI_DOUBLE:8 arg=-counts=1,17,50,100,512,65530 arg=-seed=100 arg=-testsizes=2,50 arg=-evenmemtype=device arg=-oddmemtype=reg_host
bcast 4 arg=-typelist=MPI_INT,MPI_INT:4+MPI_DOUBLE:8 arg=-counts=1,17,50,100,512,65530 arg=-seed=100 arg=-testsizes=2,50 arg=-evenmemtype=device arg=-oddmemtype=shared
bcast 4 arg=-typelist=MPI_INT,MPI_INT:4+MPI_DOUBLE:8 arg=-counts=1,17,50,100,512,65530 arg=-seed=100 arg=-testsizes=2,50 arg=-evenmemtype=host arg=-oddmemtype=device
bcast 4 arg=-typelist=MPI_INT,MPI_INT:4+MPI_DOUBLE:8 arg=-counts=1,17,50,100,512,65530 arg=-seed=100 arg=-testsizes=2,50 arg=-evenmemtype=host arg=-oddmemtype=host
bcast 4 arg=-typelist=MPI_INT,MPI_INT:4+MPI_DOUBLE:8 arg=-counts=1,17,50,100,512,65530 arg=-seed=100 arg=-testsizes=2,50 arg=-evenmemtype=host arg=-oddmemtype=reg_host
bcast 4 arg=-typelist=MPI_INT,MPI_INT:4+MPI_DOUBLE:8 arg=-counts=1,17,50,100,512,65530 arg=-seed=100 arg=-testsizes=2,50 arg=-evenmemtype=host arg=-oddmemtype=shared
bcast 4 arg=-typelist=MPI_INT,MPI_INT:4+MPI_DOUBLE:8 arg=-counts=1,17,50,100,512,65530 arg=-seed=100 arg=-testsizes=2,50 arg=-evenmemtype=reg_host arg=-oddmemtype=device
bcast 4 arg=-typelist=MPI_INT,MPI_INT:4+MPI_DOUBLE:8 arg=-counts=1,17,50,100,512,65530 arg=-seed=100 arg=-testsizes=2,50 arg=-evenmemtype=reg_host arg=-oddmemtype=host
bcast 4 arg=-typelist=MPI_INT,MPI_INT:4+MPI_DOUBLE:8 arg=-counts=1,17,50,100,512,65530 arg=-seed=100 arg=-testsizes=2,50 arg=-evenmemtype=reg_host arg=-oddmemtype=reg_host
bcast 4 arg=-typelist=MPI_INT,MPI_INT:4+MPI_DOUBLE:8 arg=-counts=1,17,50,100,512,65530 arg=-seed=100 arg=-testsizes=2,50 arg=-evenmemtype=reg_host arg=-oddmemtype=shared
bcast 4 arg=-typelist=MPI_INT,MPI_INT:4+MPI_DOUBLE:8 arg=-counts=1,17,50,100,512,65530 arg=-seed=100 arg=-testsizes=2,50 arg=-evenmemtype=shared arg=-oddmemtype=device
bcast 4 arg=-typelist=MPI_INT,MPI_INT:4+MPI_DOUBLE:8 arg=-counts=1,17,50,100,512,65530 arg=-seed=100 arg=-testsizes=2,50 arg=-evenmemtype=shared arg=-oddmemtype=host
bcast 4 arg=-typelist=MPI_INT,MPI_INT:4+MPI_DOUBLE:8 arg=-counts=1,17,50,100,512,65530 arg=-seed=100 arg=-testsizes=2,50 arg=-evenmemtype=shared arg=-oddmemtype=reg_host
bcast 4 arg=-typelist=MPI_INT,MPI_INT:4+MPI_DOUBLE:8 arg=-counts=1,17,50,100,512,65530 arg=-seed=100 arg=-testsizes=2,50 arg=-evenmemtype=shared arg=-oddmemtype=shared
# test ipc read bcast and alltoall test
bcast_gpu 6 arg=-memtype=all env=MPIR_CVAR_BCAST_COMPOSITION=1 env=MPIR_CVAR_CH4_IPC_GPU_P2P_THRESHOLD=1 env=MPIR_CVAR_BCAST_POSIX_INTRA_ALGORITHM=ipc_read env=MPIR_CVAR_BCAST_IPC_READ_MSG_SIZE_THRESHOLD=1024 env=MTEST_GPU_VISIBILITY_AFFINITY=CONTIGUOUS_DEVICE
alltoall_gpu 6 arg=-memtype=all env=MPIR_CVAR_ALLTOALL_COMPOSITION=2 env=MPIR_CVAR_CH4_IPC_GPU_P2P_THRESHOLD=1 env=MPIR_CVAR_ALLTOALL_POSIX_INTRA_ALGORITHM=ipc_read env=MPIR_CVAR_ALLTOALL_IPC_READ_MSG_SIZE_THRESHOLD=1024 env=MTEST_GPU_VISIBILITY_AFFINITY=CONTIGUOUS_DEVICE
|
