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

2.7.3-1

Browse Source 
Add support for A100 GPU and related platforms.
Add support for CUDA 11.
Add support for send/receive operations (beta).


[ROCm/rccl commit: 5949d96f36]
This commit is contained in:
Sylvain Jeaugey
2020-05-12 14:40:18 -07:00
parent c43022b9d8
commit 89072c82e5
54 changed files with 2051 additions and 1254 deletions
Download Patch File Download Diff File Expand all files Collapse all files
projects/rccl/src/include/alloc.h
+5 -5
View File
@@ -1,5 +1,5 @@
/*************************************************************************
* Copyright (c) 2019, NVIDIA CORPORATION. All rights reserved.
* Copyright (c) 2019-2020, NVIDIA CORPORATION. All rights reserved.
*
* See LICENSE.txt for license information
************************************************************************/
@@ -12,10 +12,10 @@
#include "align.h"
#include <sys/mman.h>
static inline ncclResult_t ncclCudaHostAlloc(void** ptr, void** devPtr, size_t size) {
CUDACHECK(cudaHostAlloc(ptr, size, cudaHostAllocMapped));
memset(*ptr, 0, size);
*devPtr = *ptr;
template <typename T>
static ncclResult_t ncclCudaHostCalloc(T** ptr, size_t nelem) {
CUDACHECK(cudaHostAlloc(ptr, nelem*sizeof(T), cudaHostAllocMapped));
memset(*ptr, 0, nelem*sizeof(T));
return ncclSuccess;
}