Optimization for Tree allreduce on A100.
Improve aggregation performance.
Use shared buffers for inter-node send/recv.
Add NVTX profiling hooks.
Accelerate alltoall connections by merging communication for all
channels.
Add support for one hop communication through NVLink, for faster
send/recv communication on cubemesh topologies like DGX-1.
Improve alltoall scheduling to better balance intra/inter node
communication.
Increase send/recv parallelism by 8x, each warp sending or
receiving to a different peer.
Net: move to v4.
Net: make flush operation asynchronous to accelerate alltoall.
Net: define maximum number of requests.
Fix hang when using LL128 protocol after 2^31 steps.
Fix #379 : topology injection failing when using less GPUs than
described in the XML.
Fix #394 : protocol mismatch causing hangs or crashes when using
one GPU per node.
Tento commit je obsažen v:
Sylvain Jeaugey
2020-09-04 14:35:05 -07:00
rodič 084207e685
revize 920dbe5b35
90 změnil soubory, kde provedl 11172 přidání a 3209 odebrání
+2 -2
Zobrazit soubor
@@ -9,7 +9,7 @@ Package: libnccl${nccl:Major}
Section: libs
Architecture: ${pkg:Arch}
Depends: ${misc:Depends}, ${shlibs:Depends}
Description: NVIDIA Collectives Communication Library (NCCL) Runtime
Description: NVIDIA Collective Communication Library (NCCL) Runtime
NCCL (pronounced "Nickel") is a stand-alone library of standard collective
communication routines for GPUs, implementing all-reduce, all-gather, reduce,
broadcast, and reduce-scatter.
@@ -21,7 +21,7 @@ Package: libnccl-dev
Section: libdevel
Architecture: ${pkg:Arch}
Depends: ${misc:Depends}, ${shlibs:Depends}, libnccl${nccl:Major} (= ${binary:Version})
Description: NVIDIA Collectives Communication Library (NCCL) Development Files
Description: NVIDIA Collective Communication Library (NCCL) Development Files
NCCL (pronounced "Nickel") is a stand-alone library of standard collective
communication routines for GPUs, implementing all-reduce, all-gather, reduce,
broadcast, and reduce-scatter.
+3 -3
Zobrazit soubor
@@ -1,7 +1,7 @@
Name: libnccl
Version: ${nccl:Major}.${nccl:Minor}.${nccl:Patch}${nccl:Suffix}
Release: ${pkg:Revision}+cuda${cuda:Major}.${cuda:Minor}
Summary: NVIDIA Collectives Communication Library (NCCL) Runtime
Summary: NVIDIA Collective Communication Library (NCCL) Runtime
Group: Development/Libraries
License: BSD
@@ -18,13 +18,13 @@ NVLink, NVswitch, as well as networking using InfiniBand Verbs or TCP/IP
sockets.
%package devel
Summary: NVIDIA Collectives Communication Library (NCCL) Runtime
Summary: NVIDIA Collective Communication Library (NCCL) Runtime
Group: Development/Libraries
%description devel
NCCL development files
%package static
Summary: NVIDIA Collectives Communication Library (NCCL) Runtime
Summary: NVIDIA Collective Communication Library (NCCL) Runtime
Group: Development/Libraries
%description static
NCCL static library