File size: 13,260 Bytes
b555842 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 | """
gui/pages/query_analysis.py — ◑ Query Analysis
Render function: render(ctx)
ctx keys: ov, runs, tax, lin, age, avg_lin_j, avg_age_j, tax_mult,
plan_ms, exec_ms, synth_ms, plan_pct, exec_pct, synth_pct
"""
import subprocess
import streamlit as st
import pandas as pd
import plotly.express as px
import plotly.graph_objects as go
from plotly.subplots import make_subplots
from gui.config import PROJECT_ROOT, DB_PATH, LIVE_API, WF_COLORS, PL
from gui.db import q, q_safe, q1
from gui.helpers import fl, _human_energy, _human_water, _human_carbon, _gauge_html, _bar_gauge_html
try:
import requests as _req
_REQUESTS_OK = True
except ImportError:
_REQUESTS_OK = False
class _req:
@staticmethod
def get(*a, **kw): raise RuntimeError("requests not installed")
try:
import yaml as _yaml
_YAML_OK = True
except ImportError:
_YAML_OK = False
def render(ctx: dict):
ov = ctx["ov"]
runs = ctx["runs"]
tax = ctx["tax"]
avg_lin_j = ctx["avg_lin_j"]
avg_age_j = ctx["avg_age_j"]
tax_mult = ctx["tax_mult"]
plan_ms = ctx["plan_ms"]
exec_ms = ctx["exec_ms"]
synth_ms = ctx["synth_ms"]
plan_pct = ctx["plan_pct"]
exec_pct = ctx["exec_pct"]
synth_pct = ctx["synth_pct"]
lin = ctx["lin"]
age = ctx["age"]
st.title("Query Type Analysis")
st.caption("Energy · latency · tokens · sustainability — grouped by category and workflow")
# ── Level 1: category × workflow summary ──────────────────────────────────
cat_df, _e1 = q_safe("""
SELECT
COALESCE(tc.category, 'uncategorised') AS category,
r.workflow_type,
COUNT(*) AS runs,
ROUND(AVG(r.total_energy_uj) / 1e6, 4) AS avg_energy_j,
ROUND(AVG(r.dynamic_energy_uj) / 1e6, 4) AS avg_dynamic_j,
ROUND(AVG(r.duration_ns) / 1e9, 3) AS avg_duration_s,
ROUND(AVG(r.total_tokens), 1) AS avg_tokens,
ROUND(AVG(CASE WHEN r.total_tokens > 0
THEN r.total_energy_uj / r.total_tokens END) / 1e3, 4) AS avg_mj_per_token,
ROUND(AVG(r.total_energy_uj / 1e6 /
NULLIF(r.duration_ns / 1e9, 0)), 4) AS avg_j_per_sec,
ROUND(AVG(r.planning_time_ms), 1) AS avg_plan_ms,
ROUND(AVG(r.execution_time_ms), 1) AS avg_exec_ms,
ROUND(AVG(r.synthesis_time_ms), 1) AS avg_synth_ms,
ROUND(AVG(r.carbon_g) * 1000, 4) AS avg_carbon_mg,
ROUND(AVG(r.water_ml), 4) AS avg_water_ml,
ROUND(AVG(es_agg.core_j), 4) AS avg_core_j,
ROUND(AVG(es_agg.uncore_j), 4) AS avg_uncore_j,
ROUND(AVG(es_agg.dram_j), 4) AS avg_dram_j
FROM runs r
JOIN experiments e ON r.exp_id = e.exp_id
LEFT JOIN task_categories tc ON e.task_name = tc.task_id
LEFT JOIN (
SELECT run_id,
(MAX(core_energy_uj) - MIN(core_energy_uj)) / 1e6 AS core_j,
(MAX(uncore_energy_uj) - MIN(uncore_energy_uj)) / 1e6 AS uncore_j,
(MAX(dram_energy_uj) - MIN(dram_energy_uj)) / 1e6 AS dram_j
FROM energy_samples GROUP BY run_id
) es_agg ON r.run_id = es_agg.run_id
GROUP BY COALESCE(tc.category,'uncategorised'), r.workflow_type
ORDER BY category, r.workflow_type
""")
if _e1:
st.error(f"Query error: {_e1}")
elif cat_df.empty:
st.info("No data — run experiments and ensure task_categories table is populated.")
else:
# KPI row
_lin_cat = cat_df[cat_df.workflow_type == "linear"]
_age_cat = cat_df[cat_df.workflow_type == "agentic"]
k1,k2,k3,k4,k5 = st.columns(5)
k1.metric("Categories", cat_df.category.nunique())
k2.metric("Total runs", int(cat_df.runs.sum()))
k3.metric("Best mJ/token",
f"{cat_df.avg_mj_per_token.min():.4f}" if cat_df.avg_mj_per_token.notna().any() else "—")
k4.metric("Avg linear J", f"{_lin_cat.avg_energy_j.mean():.3f}J" if not _lin_cat.empty else "—")
k5.metric("Avg agentic J",f"{_age_cat.avg_energy_j.mean():.3f}J" if not _age_cat.empty else "—")
st.divider()
c1, c2 = st.columns(2)
with c1:
st.markdown("**Energy per query by category**")
fig = px.bar(cat_df.dropna(subset=["avg_energy_j"]),
x="category", y="avg_energy_j", color="workflow_type",
barmode="group", color_discrete_map=WF_COLORS,
labels={"avg_energy_j":"Avg Energy (J)","category":"Category"})
st.plotly_chart(fl(fig), use_container_width=True)
with c2:
st.markdown("**Energy per token (mJ)**")
fig2 = px.bar(cat_df.dropna(subset=["avg_mj_per_token"]),
x="category", y="avg_mj_per_token", color="workflow_type",
barmode="group", color_discrete_map=WF_COLORS,
labels={"avg_mj_per_token":"mJ / token","category":"Category"})
st.plotly_chart(fl(fig2), use_container_width=True)
c3, c4 = st.columns(2)
with c3:
st.markdown("**Phase time breakdown (agentic)**")
_ap = cat_df[cat_df.workflow_type == "agentic"].copy()
if not _ap.empty:
_ph = _ap[["category","avg_plan_ms","avg_exec_ms","avg_synth_ms"]].melt(
id_vars="category", var_name="phase", value_name="ms")
_ph["phase"] = _ph["phase"].map({
"avg_plan_ms":"Planning","avg_exec_ms":"Execution","avg_synth_ms":"Synthesis"})
fig3 = px.bar(_ph.dropna(), x="category", y="ms", color="phase",
barmode="stack",
color_discrete_map={"Planning":"#f59e0b","Execution":"#3b82f6","Synthesis":"#a78bfa"},
labels={"ms":"ms","category":"Category"})
st.plotly_chart(fl(fig3), use_container_width=True)
else:
st.info("No agentic data.")
with c4:
st.markdown("**Hardware domain breakdown (linear)**")
_hl = cat_df[cat_df.workflow_type == "linear"].copy()
if not _hl.empty:
_hm = _hl[["category","avg_core_j","avg_uncore_j","avg_dram_j"]].melt(
id_vars="category", var_name="domain", value_name="j")
_hm["domain"] = _hm["domain"].map(
{"avg_core_j":"Core","avg_uncore_j":"Uncore","avg_dram_j":"DRAM"})
fig4 = px.bar(_hm.dropna(), x="category", y="j", color="domain",
barmode="stack",
color_discrete_map={"Core":"#3b82f6","Uncore":"#38bdf8","DRAM":"#a78bfa"},
labels={"j":"Joules","category":"Category"})
st.plotly_chart(fl(fig4), use_container_width=True)
else:
st.info("No linear data.")
st.divider()
# ── Level 2: per-task ─────────────────────────────────────────────────
st.markdown("### Level 2 — Per-task detail")
_sel_cat = st.selectbox("Filter category",
["all"] + sorted(cat_df.category.dropna().unique().tolist()),
key="qa_cat")
_cat_where = f"WHERE tc.category = '{_sel_cat}'" if _sel_cat != "all" else ""
task_df, _e2 = q_safe(f"""
SELECT e.task_name,
COALESCE(tc.category,'uncategorised') AS category,
r.workflow_type,
COUNT(*) AS runs,
ROUND(AVG(r.total_energy_uj)/1e6, 4) AS avg_energy_j,
ROUND(AVG(r.duration_ns)/1e9, 3) AS avg_duration_s,
ROUND(AVG(r.total_tokens), 1) AS avg_tokens,
ROUND(AVG(CASE WHEN r.total_tokens > 0
THEN r.total_energy_uj/r.total_tokens END)/1e3, 4) AS avg_mj_per_token,
ROUND(AVG(r.carbon_g)*1000, 4) AS avg_carbon_mg,
ROUND(AVG(r.water_ml), 4) AS avg_water_ml,
ROUND(AVG(r.llm_calls), 1) AS avg_llm_calls,
ROUND(AVG(r.tool_calls), 1) AS avg_tool_calls
FROM runs r
JOIN experiments e ON r.exp_id = e.exp_id
LEFT JOIN task_categories tc ON e.task_name = tc.task_id
{_cat_where}
GROUP BY e.task_name, COALESCE(tc.category,'uncategorised'), r.workflow_type
ORDER BY avg_energy_j DESC
""")
if _e2:
st.error(_e2)
elif not task_df.empty:
fig5 = px.bar(task_df.dropna(subset=["avg_energy_j"]),
x="task_name", y="avg_energy_j", color="workflow_type",
barmode="group", color_discrete_map=WF_COLORS,
hover_data=["category","avg_tokens","avg_mj_per_token"],
labels={"avg_energy_j":"Avg Energy (J)","task_name":"Task"})
fig5.update_xaxes(tickangle=30)
st.plotly_chart(fl(fig5), use_container_width=True)
_sc = [c for c in ["task_name","category","workflow_type","runs","avg_energy_j",
"avg_duration_s","avg_tokens","avg_mj_per_token",
"avg_carbon_mg","avg_water_ml","avg_llm_calls","avg_tool_calls"]
if c in task_df.columns]
st.dataframe(task_df[_sc], use_container_width=True, hide_index=True)
st.divider()
# ── Level 3: human-scale insights ────────────────────────────────────
st.markdown("### Level 3 — Human-scale energy insights")
st.caption("Translating joules into things you can feel")
if not runs.empty and "task_name" in runs.columns:
_t_opts = sorted(runs.task_name.dropna().unique().tolist())
_sel_t = st.selectbox("Select task to interpret", _t_opts, key="qa_human_task")
_tr = runs[runs.task_name == _sel_t]
for _wf, _border in [("linear","#22c55e"),("agentic","#ef4444")]:
_wr = _tr[_tr.workflow_type == _wf]
if _wr.empty: continue
_ej = float(_wr.energy_j.mean())
_wml = float(_wr.water_ml.mean()) if "water_ml" in _wr.columns and _wr.water_ml.notna().any() else 0
_cmg = float(_wr.carbon_g.mean()*1000) if "carbon_g" in _wr.columns and _wr.carbon_g.notna().any() else 0
_tok = float(_wr.total_tokens.mean()) if "total_tokens" in _wr.columns and _wr.total_tokens.notna().any() else 0
_dur = float(_wr.duration_ms.mean()/1000) if "duration_ms" in _wr.columns and _wr.duration_ms.notna().any() else 0
_ins = _human_energy(_ej)
_ins_html = "".join(
f"<div style='margin:2px 0;font-size:10px;color:#b8c8d8;'>{ic} {desc}</div>"
for ic, desc in _ins)
st.markdown(f"""
<div style="background:#0f1520;border:1px solid #1e2d45;border-radius:8px;
padding:14px 18px;margin-bottom:8px;border-left:3px solid {_border};">
<div style="font-size:12px;font-weight:600;color:#e8f0f8;margin-bottom:6px;">
{_wf.upper()} · {_sel_t}
<span style="font-family:monospace;color:{_border};margin-left:12px;">
{_ej:.4f}J</span>
{f'<span style="font-size:9px;color:#3d5570;margin-left:8px;">{_tok:.0f} tokens · {_dur:.1f}s</span>' if _tok > 0 else ''}
</div>
{_ins_html}
<div style="margin-top:8px;font-size:10px;color:#7090b0;border-top:1px solid #1e2d45;padding-top:6px;">
💧 {_human_water(_wml)} · 🌱 {_human_carbon(_cmg)}
</div>
</div>""", unsafe_allow_html=True)
# ══════════════════════════════════════════════════════════════════════════════
# PAGE: RUN REPLAY (formerly Live Monitor)
# Renamed — "live" only makes sense during execution.
# This page lets you inspect any completed run as an interactive timeline.
# ══════════════════════════════════════════════════════════════════════════════
|