Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Modify 'netComms' to 'sharedNetComms' to void misunderstanding. #1376

Open
wants to merge 1 commit into
base: master
Choose a base branch
from
Open

Modify 'netComms' to 'sharedNetComms' to void misunderstanding. #1376

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion src/include/proxy.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -194,7 +194,7 @@ struct ncclProxyProgressState {
pthread_t thread;
volatile int stop;
struct ncclProxyPeer** localPeers;
struct ncclSharedNetComms* netComms[NCCL_MAX_NETDEVS];
struct ncclSharedNetComms* sharedNetComms[NCCL_MAX_NETDEVS];
struct ncclProxyArgs* active;
struct ncclProxyArgs* pool;
struct ncclProxyPool* pools;
Expand Down
16 changes: 8 additions & 8 deletions src/transport/net.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -681,10 +681,10 @@ static ncclResult_t sendProxyConnect(struct ncclProxyConnection* connection, str

if (resources->maxRecvs > 1 && ncclParamNetSharedComms()) {
// Connect or reuse connection for a netdev/remote rank.
if (progressState->netComms[resources->netDev] == NULL) {
NCCLCHECK(ncclCalloc(progressState->netComms + resources->netDev, proxyState->tpnRanks));
if (progressState->sharedNetComms[resources->netDev] == NULL) {
NCCLCHECK(ncclCalloc(progressState->sharedNetComms + resources->netDev, proxyState->tpnRanks));
}
struct ncclSharedNetComms* comms = progressState->netComms[resources->netDev] + resources->tpRemoteRank;
struct ncclSharedNetComms* comms = progressState->sharedNetComms[resources->netDev] + resources->tpRemoteRank;
if (comms->sendComm[resources->channelId] == NULL) ret = proxyState->ncclNet->connect(resources->netDev, req->handle, comms->sendComm + resources->channelId, &resources->netDeviceHandle);
resources->netSendComm = comms->sendComm[resources->channelId];
if (comms->sendComm[resources->channelId]) comms->sendRefCount[resources->channelId]++;
Expand Down Expand Up @@ -831,10 +831,10 @@ static ncclResult_t recvProxyConnect(struct ncclProxyConnection* connection, str

if (resources->maxRecvs > 1 && ncclParamNetSharedComms()) {
// Connect or reuse connection for a netdev/remote rank.
if (progressState->netComms[resources->netDev] == NULL) {
NCCLCHECK(ncclCalloc(progressState->netComms + resources->netDev, proxyState->tpnRanks));
if (progressState->sharedNetComms[resources->netDev] == NULL) {
NCCLCHECK(ncclCalloc(progressState->sharedNetComms + resources->netDev, proxyState->tpnRanks));
}
struct ncclSharedNetComms* comms = progressState->netComms[resources->netDev] + resources->tpRemoteProxyRank;
struct ncclSharedNetComms* comms = progressState->sharedNetComms[resources->netDev] + resources->tpRemoteProxyRank;
if (comms->recvComm[resources->channelId] == NULL) ret = proxyState->ncclNet->accept(resources->netListenComm, comms->recvComm+resources->channelId, &resources->netDeviceHandle);
resources->netRecvComm = comms->recvComm[resources->channelId];
if (comms->recvComm[resources->channelId]) comms->recvRefCount[resources->channelId]++;
Expand Down Expand Up @@ -981,7 +981,7 @@ static ncclResult_t sendProxyFree(struct ncclProxyConnection* connection, struct
if (resources->shared) {
NCCLCHECK(sharedNetBuffersDestroy(proxyState, resources->tpLocalRank, 0, connection));
if (resources->maxRecvs > 1 && ncclParamNetSharedComms()) {
struct ncclSharedNetComms* comms = proxyState->progressState.netComms[resources->netDev]+resources->tpRemoteRank;
struct ncclSharedNetComms* comms = proxyState->progressState.sharedNetComms[resources->netDev]+resources->tpRemoteRank;
comms->sendRefCount[resources->channelId]--;
if (comms->sendRefCount[resources->channelId] == 0) NCCLCHECK(proxyState->ncclNet->closeSend(comms->sendComm[resources->channelId]));
} else {
Expand Down Expand Up @@ -1022,7 +1022,7 @@ static ncclResult_t recvProxyFree(struct ncclProxyConnection* connection, struct
if (resources->shared) {
NCCLCHECK(sharedNetBuffersDestroy(proxyState, resources->tpLocalRank, 1, connection));
if (resources->maxRecvs > 1 && ncclParamNetSharedComms()) {
struct ncclSharedNetComms* comms = proxyState->progressState.netComms[resources->netDev] + resources->tpRemoteProxyRank;
struct ncclSharedNetComms* comms = proxyState->progressState.sharedNetComms[resources->netDev] + resources->tpRemoteProxyRank;
comms->recvRefCount[resources->channelId]--;
if (comms->recvRefCount[resources->channelId] == 0) NCCLCHECK(proxyState->ncclNet->closeRecv(comms->recvComm[resources->channelId]));
} else {
Expand Down