2.23.4-1

Add scalable init API
 * Add new ncclCommInitRankScalable to allow for passing multiple
   unique IDs to the init function.
 * Spreads the load onto multiple bootstrap roots, allowing for
   constant bootstrap time.
 * Requires multiple ranks to create a unique ID, and the CPU-side
   ID exchange code to call allgather[v] instead of broadcast.

Accelerate init bootstrap operations
 * Reduce the number of calls to allgather.
 * Allow roots to reply early to ranks when information is already
   available.
 * Add an option to use ncclNet instead of sockets to perform
   bootstrap allgather operations.

Add PAT algorithms for Allgather and ReduceScatter
 * Parallel Aggregated Trees, variation of Bruck algorithm.
 * Logarithmic number of network steps for small sizes at scale.
 * Only supports one rank per node at the moment.

Add support for registered buffers for intra-node communication.
 * Allow registered user buffers to be accessed directly intra-node
 * Avoids extra copies in algorithms which permit it, saving
   memory bandwidth and helping with compute overlap.

Add profiler plugin API
 * New plugin API for profiling
 * Supports various levels of profiling, with a hierarchy.

Asynchronous graph allocation
 * Make calls to cudaMalloc and cudaMemcpy during graph allocation
   asynchronous.
 * Significantly speeds up graph capture.

Use fatal IB asynchronous events to stop network operation
 * Avoids many other error messages
 * Only fatal errors are affected; potentially transient errors
   (e.g. port down) do not cause an immediate stop.

Set P2P level to PXB on AMD CPUs when using more than 2 GPUs per node
 * P2P would cause a significant performance degradation when using
   many GPUs, and therefore many interleaved data flows.
 * Disable P2P through the CPU when we have 3+ GPUs per node; keep it
   enabled when we only have 2 GPUs.

Improve the init logs to report the real NCCL function.
 * Make the log report ncclCommInitRank or ncclCommSplit, rather than
   the generic ncclCommInitRankFunc.

Add a parameter to set the location of the user configuration file.
 * Add NCCL_CONF_FILE environment variable to set where the user's
   configuration file resides.

Increase default IB timeout
 * Increase IB timeout value from 18 to 20.
 * Should help avoid fatal errors on large RoCE systems.

Add new check for nvidia peermem
 * On linux kernels 6.6+, /sys/kernel/mm/memory_peers is no longer
   present; check for /sys/module/nvidia_peermem/version instead.

Fix old performance regression when mixing small and large operations.
 * Improves distribution of work on channels.

Fix crash when NUMA IDs are equal to -1.
 * Can happen when a NIC is a virtual NIC, or when linux doesn't
   know which NUMA node a device is attached to
 * Issue NVIDIA/nccl-tests#233

Fix tree graph search when NCCL_CROSS_NIC is set to 1.
 * Would force NCCL to use the balanced_tree pattern, thereby
   disabling LL128 on platforms with 1 GPU+1 NIC per PCI switch.
 * Would also try to use alternate rings even though it was not
   needed.

Compiler tweaks and fixes
 * PR #1177
 * PR #1228

Fix stack smash
 * PR #1325

Fixes for multi-node NVLink + IB operation

Coverity fixes and comments.

src/include/device.h

@@ -128,6 +128,8 @@ struct ncclConnInfo {
 };
 
 struct ncclProxyConnector {
+  bool initialized;
+  int rank;
   int tpRank;
   int tpLocalRank;
   int sameProcess;
@@ -141,6 +143,8 @@ struct ncclConnector {
   struct ncclTransportComm* transportComm;
   void* transportResources;
   struct ncclConnInfo conn;
+  int sendMemSameProcess;
+  int recvMemSameProcess;
 };
 
 struct ncclRing {
@@ -225,6 +229,7 @@ struct alignas(16) ncclDevWorkP2p {
 
   uint8_t sendProtoLL:1, recvProtoLL:1;
   uint8_t sendRegistered:1, recvRegistered:1;
+  uint8_t sendIpcReg:1, recvIpcReg:1;
 };
 
 // Compute the subset of the data transfer corresponding to the given part index.
@@ -266,6 +271,10 @@ struct alignas(16) ncclDevWorkColl {
   uint32_t root;
   void* recvbuff;
   void* sendbuff;
+  uintptr_t sendbuffOffset;
+  uintptr_t recvbuffOffset;
+  uintptr_t* sendbuffRmtAddrs;
+  uintptr_t* recvbuffRmtAddrs;
   union {
     // Continuous-byte-distribution scheduling. The lo and hi channels are of
     // different size than the channels in the middle.
@@ -384,6 +393,7 @@ struct ncclDevComm {
   int nNodes;
   int buffSizes[NCCL_NUM_PROTOCOLS];
   int p2pChunkSize;
+  int isNvlink;
 
   // Work fifo return credits
   uint32_t* workConsumed/*[MAXCHANNELS]*/;
@@ -395,6 +405,7 @@ struct ncclDevComm {
 
   // Channels, device side
   struct ncclDevChannel* channels/*[MAXCHANNELS]*/;
+  int* rankToLocalRank;
 };
 
 struct alignas(16) ncclDevCommAndChannels {
@@ -539,11 +550,12 @@ inline int ncclDevFuncId(int coll, int devRedOp, int type, int algo, int proto)
     if (coll == ncclFuncSendRecv) break;
     row += 1;
 
-    int nAlgos = 3;
+    int nAlgos = 4;
     if (coll == ncclFuncAllGather) {
       int algo1 = algo == NCCL_ALGO_RING ? 0 :
                   algo == NCCL_ALGO_COLLNET_DIRECT ? 1 :
-                /*algo == NCCL_ALGO_NVLS*/ 2;
+                  algo == NCCL_ALGO_NVLS ? 2 :
+                /*algo == NCCL_ALGO_PAT*/ 3;
       row += algo1*NCCL_NUM_PROTOCOLS + proto;
       break;
     }
@@ -556,7 +568,7 @@ inline int ncclDevFuncId(int coll, int devRedOp, int type, int algo, int proto)
     }
     row += nAlgos*NCCL_NUM_PROTOCOLS;
 
-    nAlgos = NCCL_NUM_ALGORITHMS;
+    nAlgos = 6;
     if (coll == ncclFuncAllReduce) {
       row += ((devRedOp*NumTypes + type)*nAlgos + algo)*NCCL_NUM_PROTOCOLS + proto;
       break;
@@ -570,11 +582,12 @@ inline int ncclDevFuncId(int coll, int devRedOp, int type, int algo, int proto)
     }
     row += ncclNumDevRedOps*NumTypes*nAlgos*NCCL_NUM_PROTOCOLS;
 
-    nAlgos = 3;
+    nAlgos = 4;
     if (coll == ncclFuncReduceScatter) {
       int algo1 = algo == NCCL_ALGO_RING ? 0 :
                   algo == NCCL_ALGO_COLLNET_DIRECT ? 1 :
-                /*algo == NCCL_ALGO_NVLS*/ 2;
+                  algo == NCCL_ALGO_NVLS ? 2 :
+                /*algo == NCCL_ALGO_PAT*/ 3;
       row += ((devRedOp*NumTypes + type)*nAlgos + algo1)*NCCL_NUM_PROTOCOLS + proto;
       break;
     }