Skip to content

UPSTREAM PR #22129: Tensor-parallel: Fix delayed AllReduce on Gemma-4 MoE #1361

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

Open
loci-dev wants to merge 1 commit into
base: main
Choose a base branch
from
Open

UPSTREAM PR #22129: Tensor-parallel: Fix delayed AllReduce on Gemma-4 MoE #1361

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
30 changes: 26 additions & 4 deletions ggml/src/ggml-backend-meta.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1661,6 +1661,24 @@ static enum ggml_status ggml_backend_meta_graph_compute(ggml_backend_t backend,

ggml_tensor * node = cgraph->nodes[id];
int32_t n_used = ggml_node_get_use_count(cgraph, id);

// Skip MIRRORED nodes that don't consume node
auto skip_unrelated = [&]() {
while (id + 1 < cgraph->n_nodes) {
ggml_tensor * next = cgraph->nodes[id+1];
bool uses_node = false;
for (int s = 0; s < GGML_MAX_SRC; s++) {
if (next->src[s] == node) { uses_node = true; break; }
}
if (uses_node) break;
if (ggml_backend_meta_get_split_state(next, false).axis != GGML_BACKEND_SPLIT_AXIS_MIRRORED) {
break;
}
id++;
}
};

skip_unrelated();
if (id + 1 >= cgraph->n_nodes) {
return idr;
}
Expand All @@ -1675,17 +1693,21 @@ static enum ggml_status ggml_backend_meta_graph_compute(ggml_backend_t backend,
n_used = ggml_node_get_use_count(cgraph, id);
}
}
if (id + 1 >= cgraph->n_nodes) {
return idr;
}
{
// Chain of MULs with MIRRORED src[1]
while (true) {
skip_unrelated();
if (id + 1 >= cgraph->n_nodes) {
return idr;
}
ggml_tensor * next = cgraph->nodes[id+1];
if (next->op == GGML_OP_MUL && next->src[0] == node &&
ggml_backend_meta_get_split_state(next->src[1], false).axis == GGML_BACKEND_SPLIT_AXIS_MIRRORED) {
node = next;
id++;
idr = id;
n_used = ggml_node_get_use_count(cgraph, id);
} else {
break;
}
}

Expand Down
Loading