AI/rocm-systems
2
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Merge remote-tracking branch 'nccl/master' into develop

Browse Source
This commit is contained in:
Wenkai Du
2023-06-21 20:54:24 -07:00
parent 0c77c66221 ea38312273
commit abd0615351
83 changed files with 5798 additions and 4090 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/include/devcomm.h
+15 -6
View File
@@ -21,12 +21,13 @@
typedef enum { ncclFuncBroadcast, ncclFuncReduce, ncclFuncAllGather, ncclFuncReduceScatter, ncclFuncAllReduce, ncclFuncSendRecv, ncclFuncSend, ncclFuncRecv, ncclFuncAllToAllPivot, ncclNumFuncs} ncclFunc_t;
extern const char* ncclFuncStr[NCCL_NUM_FUNCTIONS+2];
#define NCCL_NUM_ALGORITHMS 5 // Tree/Ring/CollNet*
#define NCCL_NUM_ALGORITHMS 6 // Tree/Ring/CollNet*
#define NCCL_ALGO_TREE 0
#define NCCL_ALGO_RING 1
#define NCCL_ALGO_COLLNET_DIRECT 2
#define NCCL_ALGO_COLLNET_CHAIN 3
#define NCCL_ALGO_NVLS 4
#define NCCL_ALGO_NVLS_TREE 5
extern const char* ncclAlgoStr[NCCL_NUM_ALGORITHMS];
#define NCCL_NUM_PROTOCOLS 3 // Simple/LL/LL128
@@ -112,10 +113,10 @@ struct ncclConnInfo {
};
struct ncclProxyConnector {
int rank;
int localRank;
int tpRank;
int tpLocalRank;
int sameProcess;
struct ncclProxyConnection* connection;
struct ncclComm* comm;
};
struct ncclConnector {
@@ -124,7 +125,6 @@ struct ncclConnector {
struct ncclTransportComm* transportComm;
void* transportResources;
struct ncclConnInfo conn;
struct ncclComm *comm;
};
struct ncclRing {
@@ -141,6 +141,9 @@ struct ncclRing {
};
// The root of each tree only has one node down (+1 intra-node).
#define NCCL_MAX_TREE_ARITY_TOP 2
// Nodes inside the binary tree can have to two nodes down (+1 intra-node).
#define NCCL_MAX_TREE_ARITY 3
struct ncclTree {
int depth;
@@ -161,18 +164,24 @@ struct ncclDirect {
#define NCCL_CONN_IDX_P2P_NET 2
#define NCCL_MAX_NVLS_ARITY 8
#define NCCL_MAX_NVLS_TREE_ARITY 3
struct ncclNvls {
int out;
int nHeads; // Number of parallel N<->1<->net operations we'll do in parallel; size of up/down
int headRank; // Index in 0..nHeads-1 I am the head rank of. -1 if I'm not a head rank (no local NIC)
int up[NCCL_MAX_NVLS_ARITY];
int down;
int treeUp;
int treeDown[NCCL_MAX_NVLS_TREE_ARITY];
int node;
int nNodes;
};
#define NCCL_MAX_CONNS 3
struct ncclChannelPeer {
struct ncclConnector send[NCCL_MAX_CONNS];
struct ncclConnector recv[NCCL_MAX_CONNS];
int refCount;
};
struct ncclDevComm;
@@ -362,7 +371,7 @@ static_assert(sizeof(struct ncclCollTrace) == 8*sizeof(int), "ncclCollTrace must
#endif
struct alignas(16) ncclDevChannel {
struct ncclDevChannelPeer *peers;
struct ncclDevChannelPeer** peers;
struct ncclRing ring;
struct ncclTree tree;
struct ncclTree collnetChain;