Skip to content
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

Add jetstream_model_load_time metric #154

Open
wants to merge 5 commits into
base: main
Choose a base branch
from
Open

Add jetstream_model_load_time metric #154

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
5 commits
Select commit Hold shift + click to select a range
fa42cdb
first commit
Bslabe123 Dec 2, 2024
d86b69b
fmt
Bslabe123 Dec 2, 2024
7746665
fmt
Bslabe123 Dec 2, 2024
62c5fdf
missing labels
Bslabe123 Dec 2, 2024
203a6fd
typo
Bslabe123 Dec 2, 2024
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
9 changes: 8 additions & 1 deletion jetstream/core/metrics/prometheus.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -75,7 +75,11 @@ def __new__(cls):
documentation="The percentage of decode slots currently being used",
labelnames=["id", "idx"],
)

_model_load_time = Gauge(
name="jetstream_model_load_time",
documentation="Total time taken to load the model",
labelnames=["id"],
)
_server_startup_latency = Gauge(
name="jetstream_server_startup_latency",
documentation="Total time taken to start the Jetstream server",
Expand Down Expand Up @@ -232,6 +236,9 @@ def get_slots_used_percentage_metric(self, idx: int):
def get_server_startup_latency_metric(self):
return self._server_startup_latency.labels(id=self._id)

def get_model_load_time_metric(self):
return self._model_load_time.labels(id=self._id)

def get_time_to_first_token(self):
return self._time_to_first_token.labels(id=self._id)

Expand Down
5 changes: 5 additions & 0 deletions jetstream/core/server_lib.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -113,10 +113,15 @@ def create_driver(
An orchestrator driver.
"""
engines = config_lib.get_engines(config, devices=devices)
model_load_start_time = time.time()
prefill_params = [pe.load_params() for pe in engines.prefill_engines]
generate_params = [ge.load_params() for ge in engines.generate_engines]
shared_params = [ie.load_params() for ie in engines.interleaved_engines]
logging.info("Loaded all weights.")
if metrics_collector:
metrics_collector.get_model_load_time_metric().set(
time.time() - model_load_start_time
)
interleaved_mode = (
len(config.prefill_slices) + len(config.generate_slices) == 0
)
Expand Down