2.22.3-1

Rework core for NVIDIA Trusted Computing
 * Compress work structs so that they are shared between channels
 * Utilize the full amount of kernel argument space permitted (4k)
   before resorting to work fifo.
 * Rework the task preprocessing phase.
 * Use a separate abortDevFlag which is kept in sync with abortFlag
   using cudaMemcpy operations.
 * Rename src/include/align.h to src/include/bitops.h

Add lazy connection establishment for collective operations
 * Move buffer allocation and connection establishment to the first
   collective operation using that algorithm.
 * Accelerate init time and reduce memory usage.
 * Avoid allocating NVLS buffers if all calls are registered.
 * Compute algo/proto in ncclLaunchCollTasksInfo early on.
 * Connect peers in ncclCollPreconnectFunc if not connected already.
 * Also move shared buffer creation to the first send/recv call.

Accelerate intra-node NVLink detection
 * Make each rank only detect NVLinks attached to its GPU.
 * Fuse XMLs to reconstruct the full NVLink topology

Add init profiling to report time spend in different init phases.
 * Report timings of bootstrap, allgather, search, connect, etc.
 * Add new "PROFILE" category for NCCL_DEBUG_SUBSYS.

Add support for PCI p2p on split PCI switches
 * Detect split PCI switches through a kernel module exposing
   switch information.
 * Update the topology XML and graph to add those inter-switch
   connections.

Add cost estimation API
 * Add a new ncclGroupEndSimulate primitive to return the estimated
   time a group would take.

Net/IB: Add separate traffic class for fifo messages
 * Add NCCL_IB_FIFO_TC to control the traffic class of fifo messages
   independently from NCCL_IB_TC.
   Merges PR #1194

Net/IB: Add support for IB router
 * Use flid instead of lid if subnets do not match
 * Warn if flid is 0

Optimizations and fixes for device network offload (unpack)
 * Double the default number of channels
 * Cache netDeviceType
 * Fix save/increment head logic to enable Tree support.

Support ncclGroupStart/End for ncclCommAbort/Destroy
 * Allow Abort/Destroy to be called within a group when managing
   multiple GPUs with a single process.

Improve Tuner API
 * Provide to the plugin the original cost table so that the plugin
   can leave unknown or disabled algo/proto combinations untouched.
 * Remove nvlsSupport and collnetSupport.

Do not print version to stdout when using a debug file
 * Also print version from all processes with INFO debug level.
   Fixes issue #1271

Fix clang warnings in NVTX headers
 * Update NVTX headers to the latest version
   Fixes issue #1270

Disable port fusion in heterogeneous systems
 * Do not fuse ports if a mix of multi-port and single port are detected.

Fix NVLS graphs search for dual NICs.
 * Fix NVLS graph search when we have more than one NIC per GPU.

Fix crash with collnetDirect
 * Add separate graph search for collnetDirect, testing alltoall paths
   and working similarly to the NVLS search.

Fix hang when nodes have different CPU types
 * Add the CPU type to the rank peer info.
 * Align all ranks on the CPU type after the first allgather.
 * Only use the aligned CPU type for all tuning operations.
   Fixes issue #1136
   Fixes issue #1184

Fix performance of registered send/recv operations
 * Allow for single full size operations
 * Add INFO to confirm the registration of send/recv buffers.

Move all sync ops to finalize stage
 * Ensure ncclCommDestroy is non-blocking if ncclCommFinalize has
   been called.

Improve error reporting during SHM segment creation

Improve support of various compilers
   Merges PR #1177
   Merges PR #1228

Allow net and tuner plugins to be statically linked
 * Search for ncclNet or ncclTuner symbols in the main binary.
   Merges PR #979

Plugin examples includes cleanup
 * Harmonize err.h and common.h usage.
 * Add mixed plugin with both net and tuner.


[ROCm/rccl commit: 178b6b7590]

projects/rccl/src/graph/paths.cc

@@ -10,6 +10,8 @@
 #include "comm.h"
 #include "net.h"
 #include "channel.h"
+#include "transport.h"
+#include "device.h"
 
 // Pre-compute GPU->NIC, GPU->GPU and NIC->GPU paths
 
@@ -732,12 +734,7 @@ static ncclResult_t ncclTopoGetNchannels(struct ncclComm* comm, int g /*local gp
 
 NCCL_PARAM(MinP2pNChannels, "MIN_P2P_NCHANNELS", 1);
 NCCL_PARAM(MaxP2pNChannels, "MAX_P2P_NCHANNELS", MAXCHANNELS);
-
-static int nextPow2(int v) {
-  int pow2 = 1;
-  while (pow2 < v) pow2 <<= 1;
-  return pow2;
-}
+extern int64_t ncclParamWorkArgsBytes();
 
 ncclResult_t ncclTopoComputeP2pChannels(struct ncclComm* comm) {
   /* here we already honor comm->max/minCTAs for p2pnChannels. */
@@ -759,19 +756,17 @@ ncclResult_t ncclTopoComputeP2pChannels(struct ncclComm* comm) {
     }
   }
 
-  // Round to next pow2 nChannelsPerPeer and nChannels
-  comm->p2pnChannelsPerPeer = nextPow2(minChannels);
-  comm->p2pnChannels = nextPow2(comm->p2pnChannels);
+  // Make nChannelsPerPeer and nChannels powers of 2. This is relied on when
+  // mapping p2p peers to channels.
+  comm->p2pnChannelsPerPeer = pow2Up(minChannels);
+  comm->p2pnChannels = pow2Up(comm->p2pnChannels);
+
+  comm->p2pnChannels = std::min(comm->p2pnChannels, pow2Down(ncclDevMaxChannelsForArgsBytes(ncclParamWorkArgsBytes())));
+  comm->p2pnChannelsPerPeer = std::min(comm->p2pnChannelsPerPeer, comm->p2pnChannels);
 
   // Init channels that weren't used so far
   for (int c=comm->nChannels; c<comm->p2pnChannels; c++) NCCLCHECK(initChannel(comm, c));
 
-  // We want to spread channels used when there aren't many and progressively
-  // fill the whole space of nChannels. To do so we mirror the bits in the
-  // nChannels space.
-  for (int c=0; c<comm->p2pnChannels; c++) {
-    comm->p2pChannels[c] = mirrorBits(c, comm->p2pnChannels);
-  }
   return ncclSuccess;
 }