2.11.4-1

Add new API for creating a reduction operation which multiplies the input by a rank-specific scalar before doing an inter-rank summation (see: ncclRedOpCreatePreMulSum).
Improve CollNet (SHARP) performance of ncclAllReduce when captured in a CUDA Graph via user buffer registration.
Add environment variable NCCL_NET_PLUGIN="<suffix>" to allow user to choose among multiple NCCL net plugins by substituting into "libnccl-net-<suffix>.so".
Fix memory leak of NVB connections.
Fix topology detection of IB Virtual Functions (SR-IOV).


[ROCm/rccl commit: e11238b302]

projects/rccl/src/include/debug.h

@@ -7,17 +7,14 @@
 #ifndef NCCL_DEBUG_H_
 #define NCCL_DEBUG_H_
 
-#include "core.h"
-
+#include "nccl_net.h"
 #include <stdio.h>
 #include <chrono>
 
 #include <sys/syscall.h>
 #include <limits.h>
 #include <string.h>
-#include "nccl_net.h"
-
-#define gettid() (pid_t) syscall(SYS_gettid)
+#include <pthread.h>
 
 extern int ncclDebugLevel;
 extern uint64_t ncclDebugMask;