Skip to content

server: batch checkpoints to support kvcache context truncation #19970

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

Closed
aagit wants to merge 1 commit into from
Closed

server: batch checkpoints to support kvcache context truncation #19970

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
49 changes: 26 additions & 23 deletions tools/server/server-context.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2521,7 +2521,8 @@ struct server_context_impl {
}

// entire prompt has been processed
if (slot.prompt.n_tokens() == slot.task->n_tokens()) {
const bool finish = slot.prompt.n_tokens() == slot.task->n_tokens();
if (finish) {
slot.state = SLOT_STATE_DONE_PROMPT;

GGML_ASSERT(batch.n_tokens > 0);
Expand All @@ -2533,44 +2534,46 @@ struct server_context_impl {
slot.i_batch = batch.n_tokens - 1;

slot.init_sampler();
}

const auto pos_min = llama_memory_seq_pos_min(llama_get_memory(ctx), slot.id);
const auto pos_max = llama_memory_seq_pos_max(llama_get_memory(ctx), slot.id);
const auto pos_min = llama_memory_seq_pos_min(llama_get_memory(ctx), slot.id);
const auto pos_max = llama_memory_seq_pos_max(llama_get_memory(ctx), slot.id);

// no need for empty or small checkpoints
do_checkpoint = do_checkpoint && (pos_min >= 0 && pos_max >= 64);
// no need for empty or small checkpoints
do_checkpoint = do_checkpoint && (pos_min >= 0 && pos_max >= 64);

// no need to create checkpoints that are too close together
do_checkpoint = do_checkpoint && (slot.prompt.checkpoints.empty() || pos_max > slot.prompt.checkpoints.back().pos_max + 64);
// no need to create checkpoints that are too close together
do_checkpoint = do_checkpoint && (slot.prompt.checkpoints.empty() || pos_max > slot.prompt.checkpoints.back().pos_max + 64);

// note: we create the checkpoint before calling llama_decode(), so the current batch is not
// yet processed and therefore it is not part of the checkpoint.
if (do_checkpoint) {
while (slot.prompt.checkpoints.size() >= (size_t) params_base.n_ctx_checkpoints) {
// make room for the new checkpoint, if needed
const auto & cur = slot.prompt.checkpoints.front();
// note: we create the checkpoint before calling llama_decode(), so the current batch is not
// yet processed and therefore it is not part of the checkpoint.
if (do_checkpoint) {
while (slot.prompt.checkpoints.size() >= (size_t) params_base.n_ctx_checkpoints) {
// make room for the new checkpoint, if needed
const auto & cur = slot.prompt.checkpoints.front();

SLT_WRN(slot, "erasing old context checkpoint (pos_min = %d, pos_max = %d, n_tokens = %" PRId64 ", size = %.3f MiB)\n",
cur.pos_min, cur.pos_max, cur.n_tokens, (float) cur.data.size() / 1024 / 1024);
SLT_WRN(slot, "erasing old context checkpoint (pos_min = %d, pos_max = %d, n_tokens = %" PRId64 ", size = %.3f MiB)\n",
cur.pos_min, cur.pos_max, cur.n_tokens, (float) cur.data.size() / 1024 / 1024);

slot.prompt.checkpoints.erase(slot.prompt.checkpoints.begin());
}
slot.prompt.checkpoints.erase(slot.prompt.checkpoints.begin());
}

const size_t checkpoint_size = llama_state_seq_get_size_ext(ctx, slot.id, LLAMA_STATE_SEQ_FLAGS_PARTIAL_ONLY);
const size_t checkpoint_size = llama_state_seq_get_size_ext(ctx, slot.id, LLAMA_STATE_SEQ_FLAGS_PARTIAL_ONLY);

auto & cur = slot.prompt.checkpoints.emplace_back(server_prompt_checkpoint{
auto & cur = slot.prompt.checkpoints.emplace_back(server_prompt_checkpoint{
/*.pos_min = */ pos_min,
/*.pos_max = */ pos_max,
/*.n_tokens = */ slot.prompt.n_tokens() - batch.n_tokens,
/*.data = */ std::vector<uint8_t>(checkpoint_size),
});

llama_state_seq_get_data_ext(ctx, cur.data.data(), checkpoint_size, slot.id, LLAMA_STATE_SEQ_FLAGS_PARTIAL_ONLY);
llama_state_seq_get_data_ext(ctx, cur.data.data(), checkpoint_size, slot.id, LLAMA_STATE_SEQ_FLAGS_PARTIAL_ONLY);

SLT_WRN(slot, "created context checkpoint %d of %d (pos_min = %d, pos_max = %d, n_tokens = %" PRId64 ", size = %.3f MiB)\n",
(int) slot.prompt.checkpoints.size(), params_base.n_ctx_checkpoints, cur.pos_min, cur.pos_max, cur.n_tokens, (float) cur.data.size() / 1024 / 1024);
}
SLT_WRN(slot, "created context checkpoint %d of %d (pos_min = %d, pos_max = %d, n_tokens = %" PRId64 ", size = %.3f MiB)\n",
(int) slot.prompt.checkpoints.size(), params_base.n_ctx_checkpoints, cur.pos_min, cur.pos_max, cur.n_tokens, (float) cur.data.size() / 1024 / 1024);
}

if (finish) {
SLT_INF(slot, "prompt processing done, n_tokens = %d, batch.n_tokens = %d\n", slot.prompt.n_tokens(), batch.n_tokens);
} else {
SLT_INF(slot, "prompt processing progress, n_tokens = %d, batch.n_tokens = %d, progress = %f\n", slot.prompt.n_tokens(), batch.n_tokens, (float) slot.prompt.n_tokens() / slot.task->n_tokens());
Expand Down