Add local user buffer registration for NVLink SHARP.
Add tuning plugin support.
Increase net API to v7 to allow for device-side packet reordering;
remove support for v4 plugins.
Add support for RoCE ECE.
Add support for C2C links.
Better detect SHM allocation failures to avoid crash with Bus Error.
Fix missing thread unlocks in bootstrap (Fixes #936).
Disable network flush by default on H100.
Move device code from src/collectives/device to src/device.
This commit is contained in:
Sylvain Jeaugey
2023-09-26 05:47:28 -07:00
parent 559b70f86c
commit f9c3dc251e
108 changed files with 4749 additions and 2059 deletions
+15
View File
@@ -1040,4 +1040,19 @@ static inline int ibv_post_send(struct ibv_qp *qp, struct ibv_send_wr *wr, struc
return qp->context->ops.post_send(qp, wr, bad_wr);
}
struct ibv_ece {
/*
* Unique identifier of the provider vendor on the network.
* The providers will set IEEE OUI here to distinguish
* itself in non-homogenius network.
*/
uint32_t vendor_id;
/*
* Provider specific attributes which are supported or
* needed to be enabled by ECE users.
*/
uint32_t options;
uint32_t comp_mask;
};
#endif // NCCL_IBV_CORE_H_