Skip to content

llama : update worst-case graph for unified cache #17379

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
wants to merge 2 commits into
base: sl/realloc-error
Choose a base branch
from
+13 −12
Open

llama : update worst-case graph for unified cache #17379

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
e483804
llama : update worst-case graph for unified cache
ggerganov Nov 19, 2025
6cdda87
ci : disable op offload in some tests
ggerganov Nov 20, 2025
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
14 changes: 7 additions & 7 deletions ci/run.sh
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -428,10 +428,10 @@ function gg_run_qwen3_0_6b {

(time ./bin/llama-imatrix --model ${model_f16} -f ${wiki_test} -ngl 99 -c 1024 -b 512 --chunks 2 ) 2>&1 | tee -a $OUT/${ci}-imatrix.log

(time ./bin/llama-save-load-state --model ${model_q4_0} -ngl 10 -c 1024 -fa off ) 2>&1 | tee -a $OUT/${ci}-save-load-state.log
(time ./bin/llama-save-load-state --model ${model_q4_0} -ngl 10 -c 1024 -fa on ) 2>&1 | tee -a $OUT/${ci}-save-load-state.log
(time ./bin/llama-save-load-state --model ${model_q4_0} -ngl 99 -c 1024 -fa off ) 2>&1 | tee -a $OUT/${ci}-save-load-state.log
(time ./bin/llama-save-load-state --model ${model_q4_0} -ngl 99 -c 1024 -fa on ) 2>&1 | tee -a $OUT/${ci}-save-load-state.log
(time ./bin/llama-save-load-state --model ${model_q4_0} -ngl 10 -c 1024 -fa off --no-op-offload) 2>&1 | tee -a $OUT/${ci}-save-load-state.log
(time ./bin/llama-save-load-state --model ${model_q4_0} -ngl 10 -c 1024 -fa on --no-op-offload) 2>&1 | tee -a $OUT/${ci}-save-load-state.log
(time ./bin/llama-save-load-state --model ${model_q4_0} -ngl 99 -c 1024 -fa off ) 2>&1 | tee -a $OUT/${ci}-save-load-state.log
(time ./bin/llama-save-load-state --model ${model_q4_0} -ngl 99 -c 1024 -fa on ) 2>&1 | tee -a $OUT/${ci}-save-load-state.log

function check_ppl {
qnt="$1"
Expand Down Expand Up @@ -523,8 +523,8 @@ function gg_run_embd_bge_small {

./bin/llama-quantize ${model_f16} ${model_q8_0} q8_0

(time ./bin/llama-embedding --model ${model_f16} -p "I believe the meaning of life is" -ngl 99 -c 0 ) 2>&1 | tee -a $OUT/${ci}-tg-f16.log
(time ./bin/llama-embedding --model ${model_q8_0} -p "I believe the meaning of life is" -ngl 99 -c 0 ) 2>&1 | tee -a $OUT/${ci}-tg-q8_0.log
(time ./bin/llama-embedding --model ${model_f16} -p "I believe the meaning of life is" -ngl 99 -c 0 --no-op-offload) 2>&1 | tee -a $OUT/${ci}-tg-f16.log
(time ./bin/llama-embedding --model ${model_q8_0} -p "I believe the meaning of life is" -ngl 99 -c 0 --no-op-offload) 2>&1 | tee -a $OUT/${ci}-tg-q8_0.log

set +e
}
Expand Down Expand Up @@ -564,7 +564,7 @@ function gg_run_rerank_tiny {
model_f16="${path_models}/ggml-model-f16.gguf"

# for this model, the SEP token is "</s>"
(time ./bin/llama-embedding --model ${model_f16} -p "what is panda?\thi\nwhat is panda?\tit's a bear\nwhat is panda?\tThe giant panda (Ailuropoda melanoleuca), sometimes called a panda bear or simply panda, is a bear species endemic to China." -ngl 99 -c 0 --pooling rank --embd-normalize -1 --verbose-prompt) 2>&1 | tee -a $OUT/${ci}-rk-f16.log
(time ./bin/llama-embedding --model ${model_f16} -p "what is panda?\thi\nwhat is panda?\tit's a bear\nwhat is panda?\tThe giant panda (Ailuropoda melanoleuca), sometimes called a panda bear or simply panda, is a bear species endemic to China." -ngl 99 -c 0 --pooling rank --embd-normalize -1 --no-op-offload --verbose-prompt) 2>&1 | tee -a $OUT/${ci}-rk-f16.log

# sample output
# rerank score 0: 0.029
Expand Down
7 changes: 4 additions & 3 deletions examples/embedding/embedding.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -104,12 +104,16 @@ int main(int argc, char ** argv) {

params.embedding = true;

// get max number of sequences per batch
const int n_seq_max = llama_max_parallel_sequences();

// if the number of prompts that would be encoded is known in advance, it's more efficient to specify the
// --parallel argument accordingly. for convenience, if not specified, we fallback to unified KV cache
// in order to support any number of prompts
if (params.n_parallel == 1) {
LOG_INF("%s: n_parallel == 1 -> unified KV cache is enabled\n", __func__);
params.kv_unified = true;
params.n_parallel = n_seq_max;
}

// utilize the full context
Expand All @@ -123,9 +127,6 @@ int main(int argc, char ** argv) {
params.n_ubatch = params.n_batch;
}

// get max number of sequences per batch
const int n_seq_max = llama_max_parallel_sequences();

llama_backend_init();
llama_numa_init(params.numa);

Expand Down
4 changes: 2 additions & 2 deletions src/llama-context.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -299,7 +299,7 @@ llama_context::llama_context(

cross.v_embd.clear();

const uint32_t n_seqs = cparams.kv_unified ? 1 : cparams.n_seq_max;
const uint32_t n_seqs = cparams.n_seq_max;
const uint32_t n_tokens = std::min(cparams.n_ctx, cparams.n_ubatch);

// avoid reserving graphs with zero outputs - assume one output per sequence
Expand Down Expand Up @@ -542,7 +542,7 @@ bool llama_context::memory_update(bool optimize) {
throw std::runtime_error("failed to initialize memory context");
}

const uint32_t n_seqs = cparams.kv_unified ? 1 : cparams.n_seq_max;
Copy link
Member Author

@ggerganov ggerganov Nov 19, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I can't recall why this cparams.kv_unified check was added in #14363. It seems unnecessary now and removing it gives a better estimate of the worst-case graph.

const uint32_t n_seqs = cparams.n_seq_max;
const uint32_t n_tokens = std::min(cparams.n_ctx, cparams.n_ubatch);

auto * gf = graph_reserve(n_tokens, n_seqs, n_tokens, mctx.get());
Expand Down
Loading