-
Notifications
You must be signed in to change notification settings - Fork 31
Description
git clone https://github.com/microsoft/msccl.git
Cloning into 'msccl'...
fatal: unable to access 'https://github.com/microsoft/msccl.git/': Failed to connect to github.com port 443 after 129347 ms: Connection timed out
root@10-10-14-107:/opt# git clone https://github.com/microsoft/msccl.git
Cloning into 'msccl'...
remote: Enumerating objects: 4379, done.
remote: Counting objects: 100% (43/43), done.
remote: Compressing objects: 100% (31/31), done.
remote: Total 4379 (delta 22), reused 25 (delta 12), pack-reused 4336
Receiving objects: 100% (4379/4379), 3.00 MiB | 3.19 MiB/s, done.
Resolving deltas: 100% (3195/3195), done.
root@10-10-14-107:/opt# cd msccl
root@10-10-14-107:/opt/msccl# make -j src.build
make -C src build BUILDDIR=/opt/msccl/build
make[1]: Entering directory '/opt/msccl/src'
Grabbing include/nccl_net.h > /opt/msccl/build/include/nccl_net.h
Generating nccl.h.in > /opt/msccl/build/include/nccl.h
Generating nccl.pc.in > /opt/msccl/build/lib/pkgconfig/nccl.pc
Compiling init.cc > /opt/msccl/build/obj/init.o
Compiling channel.cc > /opt/msccl/build/obj/channel.o
Compiling bootstrap.cc > /opt/msccl/build/obj/bootstrap.o
Compiling transport.cc > /opt/msccl/build/obj/transport.o
Compiling enqueue.cc > /opt/msccl/build/obj/enqueue.o
Compiling group.cc > /opt/msccl/build/obj/group.o
Compiling debug.cc > /opt/msccl/build/obj/debug.o
Compiling proxy.cc > /opt/msccl/build/obj/proxy.o
Compiling enhcompat.cc > /opt/msccl/build/obj/enhcompat.o
Compiling net.cc > /opt/msccl/build/obj/net.o
Compiling misc/nvmlwrap.cc > /opt/msccl/build/obj/misc/nvmlwrap.o
Compiling misc/ibvwrap.cc > /opt/msccl/build/obj/misc/ibvwrap.o
Compiling misc/gdrwrap.cc > /opt/msccl/build/obj/misc/gdrwrap.o
Compiling misc/utils.cc > /opt/msccl/build/obj/misc/utils.o
Compiling misc/argcheck.cc > /opt/msccl/build/obj/misc/argcheck.o
Compiling misc/socket.cc > /opt/msccl/build/obj/misc/socket.o
Compiling misc/shmutils.cc > /opt/msccl/build/obj/misc/shmutils.o
Compiling misc/profiler.cc > /opt/msccl/build/obj/misc/profiler.o
Compiling misc/param.cc > /opt/msccl/build/obj/misc/param.o
Compiling misc/npkit.cc > /opt/msccl/build/obj/misc/npkit.o
Compiling transport/p2p.cc > /opt/msccl/build/obj/transport/p2p.o
Compiling transport/shm.cc > /opt/msccl/build/obj/transport/shm.o
Compiling transport/net.cc > /opt/msccl/build/obj/transport/net.o
Compiling transport/net_socket.cc > /opt/msccl/build/obj/transport/net_socket.o
Compiling transport/net_ib.cc > /opt/msccl/build/obj/transport/net_ib.o
Compiling transport/coll_net.cc > /opt/msccl/build/obj/transport/coll_net.o
Compiling collectives/sendrecv.cc > /opt/msccl/build/obj/collectives/sendrecv.o
Compiling collectives/all_reduce.cc > /opt/msccl/build/obj/collectives/all_reduce.o
Compiling collectives/all_gather.cc > /opt/msccl/build/obj/collectives/all_gather.o
Compiling collectives/broadcast.cc > /opt/msccl/build/obj/collectives/broadcast.o
Compiling collectives/reduce.cc > /opt/msccl/build/obj/collectives/reduce.o
Compiling collectives/reduce_scatter.cc > /opt/msccl/build/obj/collectives/reduce_scatter.o
Compiling collectives/all_to_all.cc > /opt/msccl/build/obj/collectives/all_to_all.o
Compiling collectives/custom_collective.cc > /opt/msccl/build/obj/collectives/custom_collective.o
Compiling graph/topo.cc > /opt/msccl/build/obj/graph/topo.o
Compiling graph/paths.cc > /opt/msccl/build/obj/graph/paths.o
Compiling graph/search.cc > /opt/msccl/build/obj/graph/search.o
Compiling graph/connect.cc > /opt/msccl/build/obj/graph/connect.o
Compiling graph/rings.cc > /opt/msccl/build/obj/graph/rings.o
Compiling graph/tuning.cc > /opt/msccl/build/obj/graph/tuning.o
Compiling graph/trees.cc > /opt/msccl/build/obj/graph/trees.o
Compiling graph/xml.cc > /opt/msccl/build/obj/graph/xml.o
make[2]: Entering directory '/opt/msccl/src/collectives/device'
Generating rules > /opt/msccl/build/obj/collectives/device/Makefile.rules
nvcc fatal : Unsupported gpu architecture 'compute_35'
nvcc fatal : Unsupported gpu architecture 'compute_35'
make[2]: *** [Makefile:55: /opt/msccl/build/obj/collectives/device/sendrecv.dep] Error 1
make[2]: *** Waiting for unfinished jobs....
nvcc fatal : Unsupported gpu architecture 'compute_35'
make[2]: *** [Makefile:56: /opt/msccl/build/obj/collectives/device/all_reduce.dep] Error 1
nvcc fatal : Unsupported gpu architecture 'compute_35'
make[2]: *** [Makefile:56: /opt/msccl/build/obj/collectives/device/all_gather.dep] Error 1
make[2]: *** [Makefile:56: /opt/msccl/build/obj/collectives/device/broadcast.dep] Error 1
nvcc fatal : Unsupported gpu architecture 'compute_35'
nvcc fatal : Unsupported gpu architecture 'compute_35'
make[2]: *** [Makefile:56: /opt/msccl/build/obj/collectives/device/reduce.dep] Error 1
make[2]: *** [Makefile:56: /opt/msccl/build/obj/collectives/device/reduce_scatter.dep] Error 1
make[2]: Leaving directory '/opt/msccl/src/collectives/device'
make[1]: *** [Makefile:50: /opt/msccl/build/obj/collectives/device/colldevice.a] Error 2
make[1]: *** Waiting for unfinished jobs....
misc/ibvwrap.cc: In function ‘ncclResult_t wrap_ibv_symbols()’:
misc/ibvwrap.cc:52:54: warning: ‘int pthread_yield()’ is deprecated: pthread_yield is deprecated, use sched_yield instead [-Wdeprecated-declarations]
52 | while (ibvState == ibvInitializing) pthread_yield();
| ~~~~~~~~~~~~~^~
In file included from /usr/include/features.h:486,
from /usr/include/pthread.h:21,
from include/core.h:10,
from include/ibvwrap.h:15,
from misc/ibvwrap.cc:7:
/usr/include/pthread.h:479:12: note: declared here
479 | extern int __REDIRECT_NTH (pthread_yield, (void), sched_yield)
| ^~~~~~~~~~~~~~
misc/gdrwrap.cc: In function ‘ncclResult_t wrap_gdr_symbols()’:
misc/gdrwrap.cc:60:54: warning: ‘int pthread_yield()’ is deprecated: pthread_yield is deprecated, use sched_yield instead [-Wdeprecated-declarations]
60 | while (gdrState == gdrInitializing) pthread_yield();
| ~~~~~~~~~~~~~^~
In file included from /usr/include/features.h:486,
from /usr/include/x86_64-linux-gnu/bits/libc-header-start.h:33,
from /usr/include/limits.h:26,
from /usr/lib/gcc/x86_64-linux-gnu/11/include/limits.h:203,
from /usr/lib/gcc/x86_64-linux-gnu/11/include/syslimits.h:7,
from /usr/lib/gcc/x86_64-linux-gnu/11/include/limits.h:34,
from /usr/local/cuda/include/driver_types.h:81,
from /usr/local/cuda/include/builtin_types.h:59,
from /usr/local/cuda/include/cuda_runtime.h:90,
from /opt/msccl/build/include/nccl.h:10,
from include/gdrwrap.h:10,
from misc/gdrwrap.cc:7:
/usr/include/pthread.h:479:12: note: declared here
479 | extern int __REDIRECT_NTH (pthread_yield, (void), sched_yield)
| ^~~~~~~~~~~~~~
In file included from include/core.h:57,
from include/transport.h:13,
from include/comm.h:10,
from include/enqueue.h:10,
from enqueue.cc:7:
enqueue.cc: In function ‘ncclResult_t ncclLaunchCooperativeKernelMultiDevice(cudaLaunchParams*, int*, int, int)’:
enqueue.cc:152:53: warning: ‘cudaError_t cudaLaunchCooperativeKernelMultiDevice(cudaLaunchParams*, unsigned int, unsigned int)’ is deprecated [-Wdeprecated-declarations]
152 | CUDACHECK(cudaLaunchCooperativeKernelMultiDevice(paramsList, numDevices,
| ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~
153 | // These flags are to reduce the latency of using this API
| ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
154 | cudaCooperativeLaunchMultiDeviceNoPreSync|cudaCooperativeLaunchMultiDeviceNoPostSync));
| ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
include/checks.h:14:23: note: in definition of macro ‘CUDACHECK’
14 | cudaError_t err = cmd;
| ^~~
In file included from /usr/local/cuda/include/channel_descriptor.h:61,
from /usr/local/cuda/include/cuda_runtime.h:94,
from /opt/msccl/build/include/nccl.h:10,
from include/devcomm.h:10,
from include/transport.h:10,
from include/comm.h:10,
from include/enqueue.h:10,
from enqueue.cc:7:
/usr/local/cuda/include/cuda_runtime_api.h:4384:57: note: declared here
4384 | extern __CUDA_DEPRECATED host cudaError_t CUDARTAPI cudaLaunchCooperativeKernelMultiDevice(struct cudaLaunchParams launchParamsList, unsigned int numDevices, unsigned int flags __dv(0));
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
enqueue.cc: In function ‘ncclResult_t ncclCpuBarrierOut(ncclComm)’:
enqueue.cc:294:48: warning: ‘int pthread_yield()’ is deprecated: pthread_yield is deprecated, use sched_yield instead [-Wdeprecated-declarations]
294 | while (*ptr < comm->intraRanks) pthread_yield();
| ~~~~~~~~~~~~~^~
In file included from /usr/include/features.h:486,
from /usr/include/x86_64-linux-gnu/bits/libc-header-start.h:33,
from /usr/include/limits.h:26,
from /usr/lib/gcc/x86_64-linux-gnu/11/include/limits.h:203,
from /usr/lib/gcc/x86_64-linux-gnu/11/include/syslimits.h:7,
from /usr/lib/gcc/x86_64-linux-gnu/11/include/limits.h:34,
from /usr/local/cuda/include/driver_types.h:81,
from /usr/local/cuda/include/builtin_types.h:59,
from /usr/local/cuda/include/cuda_runtime.h:90,
from /opt/msccl/build/include/nccl.h:10,
from include/devcomm.h:10,
from include/transport.h:10,
from include/comm.h:10,
from include/enqueue.h:10,
from enqueue.cc:7:
/usr/include/pthread.h:479:12: note: declared here
479 | extern int __REDIRECT_NTH (pthread_yield, (void), sched_yield)
| ^~~~~~~~~~~~~~
make[1]: Leaving directory '/opt/msccl/src'
make: *** [Makefile:25: src.build] Error 2