nccl-test with nccl2 not run in centos6, crash in init rank

Linux: 2.6.32_1-18-0-0 #1 SMP Fri Jan 29 18:37:55 CST 2016 x86_64 x86_64 x86_64 GNU/Linux
cuda: 8.0
±----------------------------------------------------------------------------+
| NVIDIA-SMI 384.66 Driver Version: 384.66 |
|-------------------------------±---------------------±---------------------+
| GPU Name Persistence-M| Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. |
|===============================+======================+======================|
| 0 Tesla K40m On | 00000000:02:00.0 Off | 0 |
| N/A 30C P8 21W / 235W | 1MiB / 11439MiB | 0% Default |
±------------------------------±---------------------±---------------------+
| 1 Tesla K40m On | 00000000:03:00.0 Off | 0 |
| N/A 31C P8 20W / 235W | 1MiB / 11439MiB | 0% Default |
±------------------------------±---------------------±---------------------+
| 2 Tesla K40m On | 00000000:83:00.0 Off | 0 |
| N/A 29C P8 21W / 235W | 1MiB / 11439MiB | 0% Default |
±------------------------------±---------------------±---------------------+
| 3 Tesla K40m On | 00000000:84:00.0 Off | 0 |
| N/A 30C P8 21W / 235W | 1MiB / 11439MiB | 0% Default |
±------------------------------±---------------------±---------------------+

NCCL:2.1.2, 2.1.4
nccl-test: all_reduce_perf
./build/all_reduce_perf
–nthreads 2
–ngpus 2
–minbytes 32
–maxbytes 64M
–stepfactor 2
–iters 50
–warmup_iters 10
–parallel_init 0
–check 0
–op sum
–datatype float
–root 0
–swap_comms 0
–blocking 0

Program terminated with signal SIGSEGV, Segmentation fault.
#0 0x00007f34e16a20bf in __strncpy_sse2_unaligned () from /opt/compiler/gcc-4.8.2/lib/libc.so.6
(gdb) bt
#0 0x00007f34e16a20bf in __strncpy_sse2_unaligned () from /opt/compiler/gcc-4.8.2/lib/libc.so.6
#1 0x00007f34e2825e4a in getCudaPath (cudaDev=, path=path@entry=0x7fffe5215340)
at include/topo.h:34
#2 0x00007f34e282645d in p2pCanConnect (ret=0x7fffe5215c04, myOpaqueInfo=0x7f34fe123ba0,
peerOpaqueInfo=0x7f34fe123c60) at transport/p2p.cu:129
#3 0x00007f34e2810403 in fillConnect (allInfo=allInfo@entry=0x7f34fe123ba0, nranks=nranks@entry=4,
rank=rank@entry=0, connectTransport=connectTransport@entry=0x7fffe5215c50,
connectValue=connectValue@entry=0x7fffe5215c00) at init.cu:246
#4 0x00007f34e281585a in initTransportsAll (nranks=4, devs=0x7fffe5215ca0, comms=0x7f34f05e7e30)
at init.cu:574
#5 ncclCommInitAll (comms=0x7f34f05e7e30, ndev=4, devlist=) at init.cu:688
#6 0x0000000000405b4d in main (argc=29, argv=0x7fffe5218310) at common.cu:924
#1

one week passed, anybody can help? thanks