Dataset Viewer
Auto-converted to Parquet Duplicate
uid
stringclasses
10 values
dataset_id
stringclasses
1 value
year
int64
2.02k
2.03k
rack_avg_temp_c
int64
24
32
hotspot_temp_c
int64
30
44
cooling_margin
stringclasses
3 values
fan_speed_headroom
stringclasses
3 values
throttle_events_per_day
stringclasses
3 values
latency_p95_ms
stringclasses
3 values
ops_summary
stringclasses
10 values
question
stringclasses
1 value
ground_truth_label
stringclasses
2 values
ground_truth_rationale
stringclasses
10 values
outcome_signal
stringclasses
2 values
source_citation
stringclasses
1 value
DCTP-001
datacenter-thermal-performance-coherence-risk-v0.1
2,025
26
32
high
high
low
stable
Temps stable and performance steady.
Is thermal to performance coherence intact. Answer coherent or incoherent.
coherent
Cooling margin prevents throttling drift.
stable_ops
Synthetic
DCTP-002
datacenter-thermal-performance-coherence-risk-v0.1
2,025
30
41
low
low
high
spiking
Hotspots rising and throttling frequent.
Is thermal to performance coherence intact. Answer coherent or incoherent.
incoherent
Heat drives throttling and latency spikes.
thermal_risk
Synthetic
DCTP-003
datacenter-thermal-performance-coherence-risk-v0.1
2,025
25
31
medium
high
low
stable
Good fan headroom and low throttling.
Is thermal to performance coherence intact. Answer coherent or incoherent.
coherent
Thermal control holds.
stable_ops
Synthetic
DCTP-004
datacenter-thermal-performance-coherence-risk-v0.1
2,025
29
39
low
medium
medium
drifting
Cooling near limit and latency drifts upward.
Is thermal to performance coherence intact. Answer coherent or incoherent.
incoherent
Margin insufficient under sustained load.
thermal_risk
Synthetic
DCTP-005
datacenter-thermal-performance-coherence-risk-v0.1
2,024
24
30
high
high
low
stable
Cold aisle stable.
Is thermal to performance coherence intact. Answer coherent or incoherent.
coherent
Strong thermal margin.
stable_ops
Synthetic
DCTP-006
datacenter-thermal-performance-coherence-risk-v0.1
2,024
31
43
low
low
high
spiking
Thermal runaway conditions.
Is thermal to performance coherence intact. Answer coherent or incoherent.
incoherent
Hotspots cause repeated throttling.
thermal_risk
Synthetic
DCTP-007
datacenter-thermal-performance-coherence-risk-v0.1
2,024
27
34
medium
medium
low
stable
Temps manageable and p95 steady.
Is thermal to performance coherence intact. Answer coherent or incoherent.
coherent
No drift.
stable_ops
Synthetic
DCTP-008
datacenter-thermal-performance-coherence-risk-v0.1
2,024
30
40
low
medium
high
spiking
Hotspots persist and p95 spikes.
Is thermal to performance coherence intact. Answer coherent or incoherent.
incoherent
Thermal instability.
thermal_risk
Synthetic
DCTP-009
datacenter-thermal-performance-coherence-risk-v0.1
2,023
26
33
high
medium
low
stable
Stable environment.
Is thermal to performance coherence intact. Answer coherent or incoherent.
coherent
Aligned.
stable_ops
Synthetic
DCTP-010
datacenter-thermal-performance-coherence-risk-v0.1
2,023
32
44
low
low
high
spiking
Overheating and throttling.
Is thermal to performance coherence intact. Answer coherent or incoherent.
incoherent
Heat-performance decoupling.
thermal_risk
Synthetic

What this repo is for

Detect when heat and performance start to decouple.

Focus

• hotspot temperature vs throttling • cooling margin vs p95 latency stability • fan headroom vs sustained load • early drift before alarms and shutdowns

Why it matters

Many outages are preceded by thermal drift. Operators see it as “random latency.” This dataset turns it into an early warning signal.

Downloads last month
28