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

[ROCm/rccl commit: a4ea1fed5b]
This commit is contained in:
Wenkai Du
2021-04-30 16:57:36 -07:00
melakukan a76bebf8b6
80 mengubah file dengan 3136 tambahan dan 1675 penghapusan
+6 -4
Melihat File
@@ -1,6 +1,6 @@
/*************************************************************************
* Copyright (c) 2016-2020, NVIDIA CORPORATION. All rights reserved.
* Modifications Copyright (c) 2019-2020 Advanced Micro Devices, Inc. All rights reserved.
* Copyright (c) 2016-2021, NVIDIA CORPORATION. All rights reserved.
* Modifications Copyright (c) 2019-2021 Advanced Micro Devices, Inc. All rights reserved.
*
* See LICENSE.txt for license information
************************************************************************/
@@ -29,8 +29,7 @@
// Intel CPU convert GPU P2P traffic into 64B PCI TLPs, so GPU
// to GPU traffic consumes more PCI bandwidth.
#define INTEL_P2P(speed) (speed*9/12)
#define INTEL_P2P_OVERHEAD(speed) (speed*12/9)
#define INTEL_P2P_OVERHEAD(speed) (speed*6/5)
#define NCCL_TOPO_NODE_TYPES 7
#define GPU 0
@@ -114,6 +113,9 @@ struct ncclTopoNode {
int model;
cpu_set_t affinity;
}cpu;
struct {
uint64_t device;
}pci;
};
int nlinks;
struct ncclTopoLink links[NCCL_TOPO_MAX_LINKS];