Skip to content

[Frontend] vllm-rs: port prompt hashing and content prefix memoization (Rubin) #170

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
YibaiMeng merged 1 commit into from
Oct 27, 2026
Merged

[Frontend] vllm-rs: port prompt hashing and content prefix memoization (Rubin) #170

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: 2 additions & 0 deletions rust/Cargo.lock
View file
Open in desktop

Some generated files are not rendered by default. Learn more about how customized files appear on GitHub.

2 changes: 2 additions & 0 deletions rust/src/engine-core-client/Cargo.toml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -17,6 +17,7 @@ enum-as-inner.workspace = true
futures.workspace = true
half.workspace = true
hex.workspace = true
indexmap.workspace = true
itertools.workspace = true
parking_lot.workspace = true
rmp-serde.workspace = true
Expand All @@ -27,6 +28,7 @@ serde_json.workspace = true
serde_repr.workspace = true
serde_tuple.workspace = true
serde_with.workspace = true
sha2.workspace = true
task-local.workspace = true
tempfile = { workspace = true, optional = true }
thiserror.workspace = true
Expand Down
15 changes: 15 additions & 0 deletions rust/src/engine-core-client/src/client.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -268,6 +268,7 @@ pub struct EngineCoreClient {
input_address: String,
output_address: String,
engines: Vec<ConnectedEngine>,
frontend_hasher: Option<Arc<crate::frontend_hashing::FrontendHasher>>,
inner: Arc<ClientInner>,
coordinator: Option<CoordinatorHandle>,
abort_tx: mpsc::UnboundedSender<AbortRequest>,
Expand Down Expand Up @@ -410,7 +411,11 @@ impl EngineCoreClient {
(None, None, None)
};

let frontend_hasher = crate::frontend_hashing::FrontendHasher::from_ready_responses(
engines.iter().map(|engine| &engine.ready_response),
);
Ok(Self {
frontend_hasher,
config,
input_address: connected.input_address,
output_address: connected.output_address,
Expand Down Expand Up @@ -596,6 +601,16 @@ impl EngineCoreClient {
pub async fn call(&self, mut req: EngineCoreRequest) -> Result<EngineCoreOutputStream> {
req.client_index = self.config.client_index;
req.validate()?;
if let Some(hasher) = &self.frontend_hasher {
let hasher = Arc::clone(hasher);
req = self
.runtime
.spawn_blocking(move || {
hasher.attach(&mut req);
req
})
.await?;
}
trace!(
request_id = %req.request_id,
client_index = req.client_index,
Expand Down
Loading