7fa81a40a1
Fix #163 : remove warnings
...
[ROCm/rccl commit: 469b69a5d0
2018-12-11 09:19:16 -08:00
e74cca3317
Fix dummy plugin
...
[ROCm/rccl commit: 8606cdb8b2
2018-12-05 17:25:23 -08:00
e2f73fea64
Remove error logging from a normal path
...
When initNet fails, we should not print the backtrace as it is
supposed to be normal operation (falling back to sockets)
[ROCm/rccl commit: 57368189e1
2018-12-04 14:47:41 -08:00
49819bcb7c
Fix GPU Direct RDMA detection.
...
Whether the network supported GPU Direct RDMA or not was ignored,
causing sockets to break when cards were local enough that NCCL
tried to use it.
[ROCm/rccl commit: 4b39a4cf91
2018-12-04 14:42:28 -08:00
274fd3756a
Add NCCL_NET flag to many debug lines.
...
[ROCm/rccl commit: b8a9a32ccb
2018-12-04 13:10:19 -08:00
245786e9a5
Improve INFO message when external network is not found.
...
Fix #162
[ROCm/rccl commit: cdae05b277
2018-12-04 12:10:58 -08:00
f842c81503
Fixed some compilation errors when TRACE=1 set
...
[ROCm/rccl commit: 5fe2618c0e
2018-11-29 14:12:14 -08:00
5dffba1287
Rework shared memory code to use SYSCHECK macros.
...
This is to handle EINTR/EGAIN properly (issue #137 ), and also
make the code consistent with the rest.
Unfortunately posix_fallocate and mmap do not follow the classic
return code/errno pattern, so we need to write wrappers around those
functions.
[ROCm/rccl commit: eed8218e17
2018-11-29 12:52:13 -08:00
57bff87e1d
Rework SYSCHECK macros to better handle retries.
...
SYSCHECKVAL was not retrying when a retry was needed. Since not all
calls are inside a loop, that means we could silently miss an
EINTR/EAGAIN return code.
Also rework the socket connection code and improve error reporting.
[ROCm/rccl commit: 302d538b73
2018-11-29 12:52:13 -08:00
f0e160728f
Improve net API description
...
[ROCm/rccl commit: 61b50a63ef
2018-11-26 16:24:31 -08:00
ec1247221e
Make network isend/irecv non blocking
...
[ROCm/rccl commit: 98adf2fe11
2018-11-26 16:24:31 -08:00
378a103248
Add support for external network.
...
Dynamically load external network from libnccl-net.so.
Add init function in networks.
Move PCI scoring to net.cu, only ask transport to provide a path.
Simplify CUDA PCI path detection.
Add dummy external network
[ROCm/rccl commit: 0d3a20f96d
2018-11-26 16:24:31 -08:00
b458b3b337
Generate host-hash for P2P and SHM based on $(readlink /proc/self/ns/uts) + $(readlink /proc/self/ns/mnt) ( #156 )
...
[ROCm/rccl commit: d7a58cfa58
2018-11-19 17:39:44 -08:00
726c60336d
Generate nccl.h in build instead of src
...
Generating nccl.h in src makes source directories dirty after builds.
[ROCm/rccl commit: 3c6e25210b
2018-11-09 14:00:41 -08:00
ec81b3ad0e
Add official builds download link
...
[ROCm/rccl commit: 21d9a877be
2018-11-08 11:22:28 -08:00
287e36a92d
Add instructions to install packaging toolchain
...
Address #143 and #150 : debuild not installed.
[ROCm/rccl commit: f7d31919d7
2018-11-05 11:42:33 -08:00
1cc88accd5
Add install target
...
Fix issue #145
[ROCm/rccl commit: bed43524cc
2018-11-05 09:53:59 -08:00
8c945929af
2.3.7-1
...
Improved LL tuning for multi-node jobs.
Improved bootstrap for large job scaling.
Fixed a hang during bootstrap due to socket reuse.
Added operation name to the COLL INFO logging.
[ROCm/rccl commit: b56650c7f5
2018-10-24 14:44:59 -07:00
8507b66cfd
Fix nccl-tests all_reduce_perf path
...
It's `all_reduce_perf` not `allreduce_perf`
[ROCm/rccl commit: 3202d6b393
2018-10-14 00:53:17 -07:00
8ffcfac437
2.3.5-5
...
Add support for inter-node communication using sockets and InfiniBand/RoCE.
Improve latency.
Add support for aggregation.
Improve LL/regular tuning.
Remove tests as those are now at github.com/nvidia/nccl-tests .
[ROCm/rccl commit: f93fe9bfd9
2018-09-25 14:12:01 -07:00
42037d978f
fix tests on maxwell
...
[ROCm/rccl commit: 680a35c6b7
2017-11-11 19:22:06 -08:00
5047915874
Update README to link to NCCL2
...
[ROCm/rccl commit: 03d856977e
2017-08-04 09:44:37 -07:00
eadeff7e4c
Update README to link to NCCL2 part 3
...
[ROCm/rccl commit: 4a33f66e27
2017-08-04 09:44:09 -07:00
39f8f3a3a9
Update README to link to NCCL2 #2
...
[ROCm/rccl commit: d66fb63679
2017-08-04 09:43:29 -07:00
5ef74d8b80
Update README to link to NCCL2
...
[ROCm/rccl commit: 80ae43b443
2017-08-04 09:42:25 -07:00
338f6eaae1
Add support for CUDA9 half semantics
...
[ROCm/rccl commit: 29a1a916dc
2017-06-14 11:20:24 -07:00
498770f57b
Merge pull request #78 from ilya-biryukov/master
...
Fix compilation error when compiling with 'clang -x cuda'.
[ROCm/rccl commit: ccfc4567dc
2017-04-04 09:47:52 -07:00
12e9519ed0
Added Pascal nvcc flags, bumped version
...
[ROCm/rccl commit: 649f04d077
2017-03-24 11:58:14 -07:00
fa9d609d5d
Fix compilation error when compiling with 'clang -x cuda'.
...
Functions vFetch and vStore are not found by ADL with clang,
so they need to be declared before usage in ReduceCopy.
[ROCm/rccl commit: 8241cd7b6e
2017-03-16 12:01:11 +01:00
4644095088
Bumping version to 1.3.3
...
[ROCm/rccl commit: 7fef264bfa
2017-03-01 16:44:27 -08:00
72ab3de6f6
Only enable peer access for ring neighbors.
...
This enables support for systems with more than 9 GPUs attached to a single PCIe root complex.
[ROCm/rccl commit: 8996811936
2017-03-01 16:42:38 -08:00
f93f7b5d52
Fix copy/paste typo in error message
...
[ROCm/rccl commit: c219a183d0
2017-03-01 16:42:38 -08:00
d2a9c5d52e
Fix crash in Reduce when non-root ranks have invalid recvbuff
...
[ROCm/rccl commit: 8e1d6f9b60
2017-03-01 16:42:38 -08:00
030edd0507
Qualify nullptr_t with std::.
...
[ROCm/rccl commit: 5eab428294
2017-02-08 07:06:31 -08:00
851663654d
Fix 1.3.2 compilation
...
[ROCm/rccl commit: 2a974f5ca2
2016-12-08 09:11:43 -08:00
56e1a845e9
Adding missing file
...
[ROCm/rccl commit: 648e9fbb58
2016-12-05 18:06:24 -08:00
db3103d96a
1.3.2 release
...
Broadcast tuning
Better checking of inputs
Copy/reduce code simplification
[ROCm/rccl commit: 34d27771c6
2016-12-01 15:17:50 -08:00
9ec80e0873
Replace min BW by average BW in tests
...
[ROCm/rccl commit: 1093821c33
2016-12-01 15:16:35 -08:00
95860f0721
Add a static library target "staticlib" to the Makefile.
...
Rename the static library "libnccl_static.a" to disambiguate from the
dynamic libraries.
[ROCm/rccl commit: 5765d608cc
2016-11-24 11:31:03 -08:00
11e4da5bc0
Remove irrelevant output from ncclReduce Fortran tests
...
[ROCm/rccl commit: c2c515516b
2016-11-21 10:18:04 -08:00
daffd4e0ee
Add Copyright header to Fortran bindings source files
...
[ROCm/rccl commit: 9c18468fe2
2016-11-21 10:17:58 -08:00
a35c827aa6
Add Fortran bindings
...
[ROCm/rccl commit: 5f2b32e45b
2016-11-17 15:33:34 -08:00
5d496d605d
Bump to 1.3.1
...
[ROCm/rccl commit: 534b9a1697
2016-10-13 10:33:05 -07:00
ea6feab31f
Fix primitives function prototype
...
[ROCm/rccl commit: b2781d0501
2016-10-13 10:32:42 -07:00
ad50fded63
NVML (libwrap) : import the needed definitions
...
[ROCm/rccl commit: bf7d1514f7
2016-10-13 10:28:59 -07:00
827119c43e
Improved allreduce segmentation for small sizes
...
[ROCm/rccl commit: 8bb06c94be
2016-10-07 12:42:23 -07:00
d4c7d8014a
Add scan tests
...
[ROCm/rccl commit: ca330b110a
2016-09-22 11:58:33 -07:00
5dd42ff223
Make tests check for deltas and report bandwidth
...
[ROCm/rccl commit: 6c77476cc1
2016-09-22 11:58:28 -07:00
b540c2b5cc
Heavy code refactoring to remove a lot of code in collectives (~1000 lines).
...
Have all collectives use the same args, the same ring, and the same primitives for synchronization between threads with the same pattern.
[ROCm/rccl commit: cabd6848e4
2016-09-22 11:57:56 -07:00
93f5211c22
Add profiling API
...
[ROCm/rccl commit: e3dbc6110e
2016-09-22 11:56:51 -07:00
Sylvain Jeaugey
Ke Wen
David Addison
Alex Sergeev
Obihörnchen
sclarkson
Boris Fomitchev
Ilya Biryukov
Nathan Luehr
Chad Whipkey
Peter Jin
Kyle Fernandes, ne Jacobs