Skip to content

Inline Chakra LLM converter instead of subprocess fork #37

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
YWHyuk merged 1 commit into from
May 19, 2026
+12 −15
Merged

Inline Chakra LLM converter instead of subprocess fork #37

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
27 changes: 12 additions & 15 deletions serving/core/graph_generator.py
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,6 +1,5 @@
import os
import subprocess
from time import time
from chakra.src.converter.llm_converter import LLMConverter
from .request import *
from .logger import get_logger

Expand All @@ -23,18 +22,16 @@ def generate_graph(batch, hardware, num_npus, node_id=0, instance_id=0, npu_offs
workload_dir = f'../../../inputs/workload/{output_name}'
os.makedirs(workload_dir, exist_ok=True)

cmd = f'python -m chakra.src.converter.converter LLM ' \
f'--input ../../../inputs/trace/{file_name}.txt ' \
f'--output ../../../inputs/workload/{output_name}/llm ' \
f'--num-npus {num_npus} ' \
f'--npu-offset {npu_offset}'
input_path = f'../../../inputs/trace/{file_name}.txt'
output_path = f'../../../inputs/workload/{output_name}/llm'

if enable_local_offloading:
cmd += ' --local-offloading'
logger.debug(
"Generating graph: input=%s output=%s num_npus=%d npu_offset=%d local_offloading=%s",
input_path, output_path, num_npus, npu_offset, enable_local_offloading,
extra={"node_id": node_id, "instance_id": instance_id},
)

logger.debug("Generating graph with command: %s", cmd, extra={"node_id": node_id, "instance_id": instance_id})

cmd = cmd.split()
subprocess.run(cmd, text=True)
os.chdir(cwd)
return
try:
LLMConverter(input_path, output_path, num_npus, npu_offset, enable_local_offloading).convert()
finally:
os.chdir(cwd)
Loading