Skip to content

Cleaner log for adjusted splits #1494

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

Merged
ikawrakow merged 8 commits into from
Mar 24, 2026
Merged

Cleaner log for adjusted splits #1494

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
8 commits
Select commit Hold shift + click to select a range
06c9555
sweep-bench: add more skipped patterns to --minilog
Nexesenex Mar 22, 2026
11c53e2
cleaner log for adjusted splits
Nexesenex Mar 23, 2026
8dd844b
Add totalization for adjusted splits
Nexesenex Mar 23, 2026
a75ab35
Clean up semicolons
Nexesenex Mar 23, 2026
ea14fd0
Addition for totalizer ^^
Nexesenex Mar 23, 2026
78d1e89
Change accordingly to review
Nexesenex Mar 23, 2026
4a15f94
Forgotten leftover removed
Nexesenex Mar 23, 2026
59a8313
'total' instead of 'totalized'
Nexesenex Mar 23, 2026
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: 2 additions & 0 deletions examples/sweep-bench/sweep-bench.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -37,6 +37,8 @@ static void llama_selective_log_callback(ggml_log_level level, const char * text
"Layer ",
"llm_load_tensors:",
"==========================",
"merging up/gate in layer",
"repacking up/gate experts weight in layer",
};
for (const char * pat : skip_patterns) {
if (strstr(text, pat) != nullptr) {
Expand Down
29 changes: 25 additions & 4 deletions src/llama-load-tensors.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -3961,6 +3961,10 @@ bool create_tensors_helper::create_tensors() {
}
}
}
std::vector<float> gpu_split_count;
if (model.max_gpu > 0 && model.max_gpu < int(model.splits.size())) {
gpu_split_count.resize(model.splits.size(), 0.0f);
}
for (int il = 0; il < n_layer; ++il) {
int gqa_ratio = hparams.n_head(il) / hparams.n_head_kv(il);
if (ggml_backend_buft_is_host(model.buft_layer[il].buft_matrix)) {
Expand All @@ -3970,11 +3974,17 @@ bool create_tensors_helper::create_tensors() {
if (model.max_gpu > 0 && model.max_gpu < int(model.splits.size()) && il % adjust_step == 0) {
cur_splits = model.splits;
adjust_split(cur_splits, mem_used, model.max_gpu);
LLAMA_LOG_INFO("Adjusted split at layer %2d:", il);
LLAMA_LOG_INFO("Adjusted split at layer %2d: ", il);
float last_split = 0;
for (auto & p : cur_splits) {
LLAMA_LOG_INFO(" %g", p - last_split);
last_split = p;
for (int i = 0; i < (int)cur_splits.size(); ++i) {
if (i > 0) {
LLAMA_LOG_INFO(" ; ");
}
LLAMA_LOG_INFO("GPU%d: %4g", i, cur_splits[i] - last_split);
if (i < int(gpu_split_count.size())) {
gpu_split_count[i] += cur_splits[i] - last_split;
}
last_split = cur_splits[i];
}
LLAMA_LOG_INFO("\n");
}
Expand Down Expand Up @@ -4243,6 +4253,17 @@ bool create_tensors_helper::create_tensors() {
}
}

if (!gpu_split_count.empty()) {
LLAMA_LOG_INFO("Adjusted splits (total) : ");
for (int i = 0; i < (int)gpu_split_count.size(); ++i) {
if (i > 0) {
LLAMA_LOG_INFO(" ; ");
}
LLAMA_LOG_INFO("GPU%d: %4g", i, gpu_split_count[i]);
}
LLAMA_LOG_INFO("\n");
}

if (model.output) {
if (auto it = split_tensors.find(model.output); it != split_tensors.end()) {
if (ggml_backend_buft_is_host(model.buft_output.buft_matrix)) {
Expand Down
4 changes: 2 additions & 2 deletions src/llama.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -4799,7 +4799,7 @@ static void llama_repack_up_gate_exps(llama_context & lctx) {
if (nbytes > aux_buffer_gate.size()) {
aux_buffer_gate.resize(nbytes);
}
printf("%s: repacking up/gate experts weight in layer %d\n", __func__, il);
LLAMA_LOG_INFO("%s: repacking up/gate experts weight in layer %d\n", __func__, il);
ggml_backend_tensor_get(l.ffn_up_exps, aux_buffer_up.data(), 0, nbytes);
ggml_backend_tensor_get(l.ffn_gate_exps, aux_buffer_gate.data(), 0, nbytes);
if (aux_buffer_up_gate.size() < 2*nbytes) {
Expand All @@ -4825,7 +4825,7 @@ static void llama_repack_up_gate_exps(llama_context & lctx) {
if (nbytes > aux_buffer_gate.size()) {
aux_buffer_gate.resize(nbytes);
}
printf("%s: repacking up/gate experts bias in layer %d\n", __func__, il);
LLAMA_LOG_INFO("%s: repacking up/gate experts bias in layer %d\n", __func__, il);
ggml_backend_tensor_get(l.ffn_up_exps_b, aux_buffer_up.data(), 0, nbytes);
ggml_backend_tensor_get(l.ffn_gate_exps_b, aux_buffer_gate.data(), 0, nbytes);
if (aux_buffer_up_gate.size() < 2*nbytes) {
Expand Down