← zurück zur Übersicht

2026-06-14T13-33-21Z_pgvector-M-throughput-c8-r1

pgvectorStufe M · 20.00 GBbatchStatus: ok

Metriken

Throughput
12.2 QPS
Latenz ⌀
655.17 ms
Latenz p50
650.34 ms
Latenz p95
818.65 ms
Latenz p99
873.07 ms
Recall@1
65.20%
Recall@10
59.72%
Recall@100
51.71%
Precision@10
59.72%
NDCG@10
70.16%

Rohdaten

Konfiguration

Config-Name
pgvector-M-throughput-c8
DB-Version
Image
Index-Typ
ivfflat
Index-Params
{"lists":4900,"probes":10}
Dataset
M · 5.250.000 Vektoren · dim 1024 · Variante A · 20.00 GB
Queries
2.000 (Concurrency 8)

Lauf

Gestartet
14.06.2026, 15:33:21
Beendet
14.06.2026, 15:36:11
Dauer
7370s
Index-Bauzeit
2381.09s
Index-Größe
41053.9 MB
CPU ⌀ / Peak
0.37 / 0.40 cores
RAM ⌀ / Peak
19180 / 19182 MB
Disk read (Paging)
47987 MB
Disk write
0 MB
Pod-RAM-Limit
20 GiB
Warmup-Queries
1.000
K8s
v1.31.4+k3s1
Nodes
4

Notizen

measured: in-cluster
decoupled: true
ingested_at: 2026-06-14T12:52:14Z
ingest_config: pgvector-M-pressure-20gb
insert_time_s: 2196.77
n_vectors_actual: 5250000
has_metadata: true
mem_limit_gb: 20
pre_run_reset: skipped (decoupled, warmup-discard)
n_queries_executed: 1000
n_warmup: 1000
gt_file: ground_truth_ids.npy
mode: measure-only
repeat_group: 2026-06-14T13-33-21Z_pgvector-M-throughput-c8
repeat_index: 1
repeat_total: 3