huggingface/text-generation-inference
5
0
Fork 0
mirror of https://github.com/huggingface/text-generation-inference.git synced 2025-06-19 15:52:08 +00:00
Code Issues Packages Projects Releases Wiki Activity

Remove the workaround for HPU distributed.

Browse Source 
Signed-off-by: yuanwu <yuan.wu@intel.com>
This commit is contained in:
yuanwu 2025-06-11 06:15:11 +00:00
parent e202b5f98f
commit 5f26a72876
2 changed files with 0 additions and 13 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
launcher/src/env_runtime.rs
View File

@ -27,10 +27,6 @@ impl Env {
docker_label: option_env!("DOCKER_LABEL").unwrap_or("N/A"),
}
}
pub fn should_start_a_single_hpu_shard(&self) -> bool {
self.hpu_env != "N/A" && std::env::var("ATTENTION").as_deref() != Ok("paged")
}
}
impl fmt::Display for Env {

9
launcher/src/main.rs
View File

@ -1590,11 +1590,6 @@ fn spawn_shards(
) -> Result<(), LauncherError> {
// Start shard processes
for rank in 0..num_shard {
if rank != 0 && env_runtime::Env::new().should_start_a_single_hpu_shard() {
tracing::info!("Running on HPU, the launcher will not do any sharding as actual sharding is done in the server");
break;
}
let model_id = args.model_id.clone();
let revision = args.revision.clone();
let uds_path = args.shard_uds_path.clone();
@ -1670,10 +1665,6 @@ fn spawn_shards(
if shard_ready == num_shard {
break;
}
if env_runtime::Env::new().should_start_a_single_hpu_shard() {
tracing::info!("HPU detected, shard is ready");
break;
}
}
Err(TryRecvError::Empty) => {
sleep(Duration::from_millis(100));