← all runs

Cohere-100M — 2026-09-04

r8i.8xlarge · peak recall 0.9972 @ 24 QPS

QPS at recall

Recallmkt-defaults
0.90502
0.95236
0.9955

Latency at recall

mkt-defaults
@ recall 0.90
p501.96 ms
p952.51 ms
p992.76 ms
@ recall 0.95
p504.18 ms
p955.20 ms
p995.65 ms
@ recall 0.99
p5018.02 ms
p9521.91 ms
p9923.33 ms

All values are taken at exactly the stated recall: each configuration's own sweep is log-interpolated between its two surrounding measured points (measured values are used verbatim on an exact hit) and the best configuration at that recall is shown — values are never interpolated across different configurations. “—” means no single sweep spans that recall; nothing is extrapolated. Recall is distance-based (ann-benchmarks, tie-aware). Insert/update/delete workloads are not measured.

Plots

Pareto svg  ·  Distance-recall svg  ·  Per-config svg
paretodistance-recall

Host

Instance typer8i.8xlarge
Hostnameip-172-31-14-122
CPUIntel(R) Xeon(R) 6975P-C (32 vCPU)
Memory248 GB
OSUbuntu 26.04.1 LTS
Kernel7.0.0-1011-aws

Versions

rekall0.1.0 (90330ec / main)
PostgreSQLPostgreSQL 18.6 (Ubuntu 18.6-1.pgdg26.04+2) on x86_64-pc-linux-gnu, compiled by gcc (Ubuntu 15.2.0-16ubuntu1) 15.2.0, 64-bit
meerkat0.1.0 (fea1bca)
pg_prewarm1.2
vector0.8.6

Search

k10
runs3
warmup500
num_queries10000
S3 archives3://rekall-bench/runs/cohere-100m/2026-09-04/r8i.8xlarge-093817/
Instance settings (observed)
instance type          r8i.8xlarge
vCPU (observed)        32
memory (observed)      247.7 GiB
cpu model              Intel(R) Xeon(R) 6975P-C
kernel                 7.0.0-1011-aws
data directory         /var/lib/postgresql/18/main
mount point            /
filesystem             ext4
media                  unknown
device                 root
provisioned IOPS       — not an EBS volume
PostgreSQL settings
checkpoint_completion_target = 0.9
cluster_name = 18/main
data_directory = /var/lib/postgresql/18/main
effective_cache_size = 192 GB
effective_io_concurrency = 256
huge_pages = try
jit = off
maintenance_io_concurrency = 256
maintenance_work_mem = 102 GB
max_connections = 100
max_parallel_maintenance_workers = 16
max_parallel_workers = 32
max_wal_size = 32 GB
random_page_cost = 1.1
server_version = 18.6 (Ubuntu 18.6-1.pgdg26.04+2)
shared_buffers = 64 GB
wal_level = replica
work_mem = 1 GB