Download gui/pages/data_network.py from a-lems/power: direct link, hf CLI and curl.
- Browser
- Download file 7.75 kB
-
https://huggingface.co/spaces/a-lems/power/resolve/main/gui/pages/data_network.py
- Command line
-
hf download hf://spaces/a-lems/power/gui/pages/data_network.py
-
curl -L -o data_network.py https://huggingface.co/spaces/a-lems/power/resolve/main/gui/pages/data_network.py
7.75 kB
| """ | |
| gui/pages/data_network.py — ⌁ Network & API | |
| Energy cost of network latency — DNS, API, compute time. | |
| """ | |
| import streamlit as st | |
| import pandas as pd | |
| import plotly.graph_objects as go | |
| from gui.config import PL, WF_COLORS | |
| from gui.pages._dm_helpers import get_runs, no_data_banner | |
| ACCENT = "#a78bfa" | |
| def render(ctx: dict) -> None: | |
| df = get_runs(ctx) | |
| if df.empty or "api_latency_ms" not in df.columns: | |
| no_data_banner("No network latency data available yet.", ACCENT) | |
| return | |
| df = df[df["api_latency_ms"].notna() & (df["api_latency_ms"] > 0)].copy() | |
| if df.empty: | |
| no_data_banner("No runs with API latency data yet.", ACCENT) | |
| return | |
| has_energy = "energy_j" in df.columns | |
| has_dns = "dns_latency_ms" in df.columns | |
| has_compute = "compute_time_ms" in df.columns | |
| if "duration_ns" in df.columns: | |
| df["duration_ms"] = df["duration_ns"] / 1e6 | |
| else: | |
| df["duration_ms"] = 0 | |
| lin = df[df["workflow_type"] == "linear"] | |
| age = df[df["workflow_type"] == "agentic"] | |
| avg_api_lin = lin["api_latency_ms"].mean() if not lin.empty else 0 | |
| avg_api_age = age["api_latency_ms"].mean() if not age.empty else 0 | |
| avg_dns = df["dns_latency_ms"].mean() if has_dns else 0 | |
| avg_compute = df["compute_time_ms"].mean() if has_compute else 0 | |
| # ── Header ──────────────────────────────────────────────────────────────── | |
| st.markdown( | |
| f"<div style='padding:14px 20px;" | |
| f"background:linear-gradient(135deg,{ACCENT}14,{ACCENT}06);" | |
| f"border:1px solid {ACCENT}33;border-radius:12px;margin-bottom:20px;'>" | |
| f"<div style='font-size:11px;font-weight:700;color:{ACCENT};" | |
| f"text-transform:uppercase;letter-spacing:.1em;margin-bottom:10px;'>" | |
| f"Network & API — {len(df)} runs</div>" | |
| f"<div style='display:grid;grid-template-columns:repeat(4,1fr);gap:12px;'>" | |
| + "".join([ | |
| f"<div><div style='font-size:18px;font-weight:700;color:{c};" | |
| f"font-family:IBM Plex Mono,monospace;line-height:1;'>{v}</div>" | |
| f"<div style='font-size:9px;color:#94a3b8;text-transform:uppercase;" | |
| f"letter-spacing:.08em;margin-top:3px;'>{l}</div></div>" | |
| for v, l, c in [ | |
| (f"{avg_api_lin:.0f}ms", "API latency linear", WF_COLORS["linear"]), | |
| (f"{avg_api_age:.0f}ms", "API latency agentic", WF_COLORS["agentic"]), | |
| (f"{avg_dns:.1f}ms", "Avg DNS latency", "#60a5fa"), | |
| (f"{avg_compute:.0f}ms", "Avg compute time", "#f59e0b"), | |
| ] | |
| ]) | |
| + "</div></div>", | |
| unsafe_allow_html=True) | |
| col1, col2 = st.columns(2) | |
| with col1: | |
| st.markdown( | |
| f"<div style='font-size:11px;font-weight:600;color:{ACCENT};" | |
| f"text-transform:uppercase;letter-spacing:.1em;margin-bottom:8px;'>" | |
| f"API latency distribution</div>", unsafe_allow_html=True) | |
| fig = go.Figure() | |
| for wf, clr in WF_COLORS.items(): | |
| sub = df[df["workflow_type"] == wf]["api_latency_ms"].dropna() | |
| if sub.empty: continue | |
| fig.add_trace(go.Histogram( | |
| x=sub, name=wf, marker_color=clr, opacity=0.7, nbinsx=40)) | |
| fig.update_layout(**PL, height=260, barmode="overlay", | |
| xaxis_title="API latency (ms)", | |
| yaxis_title="Run count") | |
| st.plotly_chart(fig, use_container_width=True, key="dm_net_hist") | |
| with col2: | |
| st.markdown( | |
| f"<div style='font-size:11px;font-weight:600;color:{ACCENT};" | |
| f"text-transform:uppercase;letter-spacing:.1em;margin-bottom:8px;'>" | |
| f"API latency vs Energy</div>", unsafe_allow_html=True) | |
| fig2 = go.Figure() | |
| if has_energy: | |
| for wf, clr in WF_COLORS.items(): | |
| sub = df[df["workflow_type"] == wf] | |
| sub = sub[sub["energy_j"].notna() & (sub["energy_j"] > 0)] | |
| if sub.empty: continue | |
| fig2.add_trace(go.Scatter( | |
| x=sub["api_latency_ms"], y=sub["energy_j"], | |
| mode="markers", name=wf, | |
| marker=dict(color=clr, size=5, opacity=0.6))) | |
| fig2.update_layout(**PL, height=260, | |
| xaxis_title="API latency (ms)", | |
| yaxis_title="Energy (J)") | |
| st.plotly_chart(fig2, use_container_width=True, key="dm_net_scatter") | |
| # ── Latency breakdown ───────────────────────────────────────────────────── | |
| st.markdown( | |
| f"<div style='font-size:11px;font-weight:600;color:{ACCENT};" | |
| f"text-transform:uppercase;letter-spacing:.1em;margin:16px 0 8px;'>" | |
| f"Latency breakdown — DNS vs API vs Compute</div>", | |
| unsafe_allow_html=True) | |
| agg_dict = {"api": ("api_latency_ms", "mean")} | |
| if has_dns: agg_dict["dns"] = ("dns_latency_ms", "mean") | |
| if has_compute: agg_dict["cmp"] = ("compute_time_ms", "mean") | |
| breakdown = df.groupby("workflow_type").agg(**agg_dict).reset_index() | |
| fig3 = go.Figure() | |
| for col_name, label, clr in [ | |
| ("dns", "DNS", "#60a5fa"), | |
| ("api", "API wait","#f59e0b"), | |
| ("cmp", "Compute", "#22c55e"), | |
| ]: | |
| if col_name not in breakdown.columns: continue | |
| fig3.add_trace(go.Bar( | |
| x=breakdown["workflow_type"], y=breakdown[col_name], | |
| name=label, marker_color=clr, marker_line_width=0)) | |
| fig3.update_layout(**PL, height=240, barmode="stack", | |
| yaxis_title="Avg ms") | |
| st.plotly_chart(fig3, use_container_width=True, key="dm_net_breakdown") | |
| # ── Percentiles ─────────────────────────────────────────────────────────── | |
| st.markdown( | |
| f"<div style='font-size:11px;font-weight:600;color:{ACCENT};" | |
| f"text-transform:uppercase;letter-spacing:.1em;margin:16px 0 8px;'>" | |
| f"API latency percentiles</div>", unsafe_allow_html=True) | |
| rows = [] | |
| for wf in df["workflow_type"].unique(): | |
| sub = df[df["workflow_type"] == wf]["api_latency_ms"].dropna() | |
| if sub.empty: continue | |
| rows.append({ | |
| "Workflow": wf, | |
| "P50": round(sub.quantile(0.50), 1), | |
| "P75": round(sub.quantile(0.75), 1), | |
| "P90": round(sub.quantile(0.90), 1), | |
| "P95": round(sub.quantile(0.95), 1), | |
| "P99": round(sub.quantile(0.99), 1), | |
| "Max": round(sub.max(), 1), | |
| }) | |
| if rows: | |
| st.dataframe(pd.DataFrame(rows), use_container_width=True) | |
| # ── Insight ─────────────────────────────────────────────────────────────── | |
| latency_overhead = avg_api_age - avg_api_lin | |
| st.markdown( | |
| f"<div style='margin-top:12px;padding:10px 14px;" | |
| f"background:#1a0e40;border-left:3px solid {ACCENT};" | |
| f"border-radius:0 8px 8px 0;font-size:11px;" | |
| f"color:#c4b5fd;font-family:IBM Plex Mono,monospace;line-height:1.7;'>" | |
| f"Agentic runs have <b>{latency_overhead:+.0f}ms</b> API latency vs linear. " | |
| f"During API wait the CPU idles but RAPL still records package power — " | |
| f"reducing API latency is one of the highest-leverage ways to cut agentic energy." | |
| f"</div>", unsafe_allow_html=True) | |