Skip to content

feat(demo): stream_client.py warmup mode #179

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
solderzzc merged 1 commit into from
Sep 25, 2026
Merged

feat(demo): stream_client.py warmup mode #179

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
10 changes: 7 additions & 3 deletions scripts/demo/stream_client.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -5,7 +5,7 @@
the server's physical footprint and system swap, then prints a stats line. The
figures come from this run, not from the benchmark files.

usage: stream_client.py short
usage: stream_client.py warmup | short
stream_client.py long <n_filler_lines>
"""
import json, re, subprocess, sys, threading, time, urllib.request
Expand Down Expand Up @@ -36,7 +36,11 @@ def swap_gb():

def main():
kind = sys.argv[1]
if kind == "short":
if kind == "warmup":
# The first request after load pays one-off Metal pipeline/compile costs.
prompt = "Count from 1 to 8, comma-separated."
max_tokens = 24
elif kind == "short":
prompt = "Write a Swift function `fib(_ n: Int) -> Int` that returns the n-th Fibonacci number iteratively. Code only, with a one-line doc comment."
max_tokens = 160
else:
Expand Down Expand Up @@ -111,7 +115,7 @@ def ticker():
sys.stdout.write(RESET + "\n\n")
decode = (n_tok - 1) / (t_end - t_first) if n_tok > 1 else 0
ttft = t_first - t0
pf = f" prompt {n_prompt:,} tok · prefill {n_prompt / ttft:.0f} tok/s ·" if n_prompt and kind != "short" else ""
pf = f" prompt {n_prompt:,} tok · prefill {n_prompt / ttft:.0f} tok/s ·" if n_prompt and kind == "long" else ""
print(f"{CYAN}{BOLD} {n_tok} tokens · decode {decode:.1f} tok/s · TTFT {ttft:.1f}s ·{pf.rstrip(' ·')}{RESET}")
print(f"{CYAN}{BOLD} peak SwiftLM {peak_fp[0]:.1f} GB · swap +{max(0, peak_swap[0]):.1f} GB{RESET}")

Expand Down
Loading