← zurück zur Übersicht

2026-06-16T12-28-26Z_weaviate-S0-tune-hnsw-low-r2

weaviateStufe S0 · 2.00 GBtopkStatus: ok

Metriken

Throughput
446.4 QPS
Latenz ⌀
2.21 ms
Latenz p50
2.19 ms
Latenz p95
2.50 ms
Latenz p99
2.71 ms
Recall@1
69.10%
Recall@10
66.42%
Recall@100
53.20%
Precision@10
66.42%
NDCG@10
75.39%

Rohdaten

Konfiguration

Config-Name
weaviate-S0-tune-hnsw-low
DB-Version
Image
Index-Typ
hnsw
Index-Params
{"M":8,"ef_construction":64,"ef":32}
Dataset
S0 · 550.000 Vektoren · dim 1024 · Variante A · 2.00 GB
Queries
1.000 (Concurrency 1)

Lauf

Gestartet
16.06.2026, 14:28:26
Beendet
16.06.2026, 14:28:57
Dauer
7231s
Index-Bauzeit
86.21s
Index-Größe
3362.3 MB
CPU ⌀ / Peak
0.43 / 0.43 cores
RAM ⌀ / Peak
3701 / 3701 MB
Disk read (Paging)
0 MB
Disk write
0 MB
Pod-RAM-Limit
8 GiB
Warmup-Queries
1.000
K8s
v1.31.4+k3s1
Nodes
4

Notizen

measured: in-cluster
decoupled: true
ingested_at: 2026-06-16T12:27:44Z
ingest_config: weaviate-S0-tune-hnsw-low
insert_time_s: 198.22
n_vectors_actual: 550000
has_metadata: true
mem_limit_gb: 8
pre_run_reset: rollout-restart
timeout_s: 180
gomemlimit_mib: 7000
query_mem_gb: 8
n_queries_executed: 1000
n_warmup: 1000
gt_file: ground_truth_ids.npy
server_latency_ms: {"source":"weaviate_prometheus","metric":"queries_durations_ms","class_name":"Bench","query_type":"get_graphql","count":1000,"count_all_types":1000,"mean_ms":1.247,"windowed":true,"p50_ms":5,"p95_ms":9.5,"p99_ms":9.9}
mode: measure-only
repeat_group: 2026-06-16T12-27-44Z_weaviate-S0-tune-hnsw-low
repeat_index: 2
repeat_total: 6