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

2.5.6-1 (#255)

Browse Source 
Add LL128 Protocol.

Rewrite the topology detection and tree/ring creation (#179). Improve
tree performance by sending/receiving from different GPUs. Add
model-based tuning to switch between the different algorithms and
protocols.

Rework P2P/SHM detection in containers (#155, #248).

Detect duplicated devices and return an error (#231).

Add tuning for GCP

[ROCm/rccl commit: 299c554dcc]
This commit is contained in:
Sylvain Jeaugey
2019-11-19 14:57:39 -08:00
committed by GitHub
parent 221b65bee1
commit 71560fd67b
65 changed files with 4783 additions and 2832 deletions
Download Patch File Download Diff File Expand all files Collapse all files
projects/rccl/src/include/group.h
+3 -3
View File
@@ -8,14 +8,14 @@
#define NCCL_GROUP_H_
#include "nccl.h"
#include "core.h"
#include "comm.h"
bool ncclAsyncMode();
ncclResult_t ncclAsyncErrCheck(ncclResult_t ret);
typedef ncclResult_t(*ncclInitFunc_t)(ncclComm_t* newcomm, int ndev, ncclUniqueId commId, int myrank);
typedef ncclResult_t(*ncclInitFunc_t)(ncclComm_t* newcomm, int ndev, ncclUniqueId commId, int myrank, int cudaDev);
ncclResult_t ncclAsyncInit(ncclInitFunc_t func, int cudaDev, ncclComm_t* newcomm, int ndev, ncclUniqueId commId, int myrank);
ncclResult_t ncclAsyncInit(ncclInitFunc_t func, ncclComm_t* newcomm, int ndev, ncclUniqueId commId, int myrank, int cudaDev);
typedef ncclResult_t(*ncclCollFunc_t)(const void* sendbuff, void* recvbuff, size_t count,
ncclDataType_t type, ncclRedOp_t op, int root, ncclComm_t comm, cudaStream_t stream);