← all runs
Cohere-1M — 2026-09-08
c8g.medium · peak recall 0.9990 @ 39 QPS
“Meerkat 1M on c8g.medium”
QPS at recall
| Recall | mkt-defaults | mkt-norerank | mkt-rerank20 | mkt-rerank40 | mkt-rerank80 |
|---|
| 0.90 | 80 | — (max r=0.823) | 137 | 91 | 63 |
| 0.95 | 71 | — (max r=0.823) | 115 | 84 | 59 |
| 0.99 | 57 | — (max r=0.823) | — (max r=0.970) | 65 | 52 |
Latency at recall
| mkt-defaults | mkt-norerank | mkt-rerank20 | mkt-rerank40 | mkt-rerank80 |
|---|
| @ recall 0.90 | | | | | |
| p50 | 12.49 ms | — | 6.87 ms | 11.12 ms | 15.80 ms |
| p95 | 19.43 ms | — | 12.40 ms | 16.75 ms | 25.45 ms |
| p99 | 25.57 ms | — | 15.67 ms | 20.04 ms | 28.54 ms |
| @ recall 0.95 | | | | | |
| p50 | 13.92 ms | — | 8.42 ms | 11.69 ms | 17.04 ms |
| p95 | 23.05 ms | — | 14.08 ms | 18.63 ms | 26.89 ms |
| p99 | 31.35 ms | — | 17.53 ms | 22.54 ms | 30.53 ms |
| @ recall 0.99 | | | | | |
| p50 | 17.05 ms | — | — | 15.00 ms | 19.16 ms |
| p95 | 27.91 ms | — | — | 23.64 ms | 29.94 ms |
| p99 | 35.19 ms | — | — | 28.63 ms | 36.03 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 svgHost
| Instance type | c8g.medium |
| Hostname | ip-172-31-1-22 |
| CPU | unknown (1 vCPU) |
| Memory | 2 GB |
| OS | Ubuntu 26.04.1 LTS |
| Kernel | 7.0.0-1012-aws |
Versions
| rekall | 0.1.0 (62ad31c / main) |
| PostgreSQL | PostgreSQL 18.6 (Ubuntu 18.6-1.pgdg26.04+2) on aarch64-unknown-linux-gnu, compiled by gcc (Ubuntu 15.2.0-16ubuntu1) 15.2.0, 64-bit |
| meerkat | 0.1.0 (c884b15) |
| pg_prewarm | 1.2 |
| vector | 0.8.6 |
Search
Instance settings (observed)
instance type c8g.medium
vCPU (observed) 1
memory (observed) 1.8 GiB
cpu model unknown
kernel 7.0.0-1012-aws
data directory /var/lib/postgresql/18/main
mount point /
filesystem ext4
media ebs
device nvme0n1
device size 107 GB
ebs volume vol-03774164746bbcca0
ebs volume type gp3
provisioned IOPS 3000
provisioned throughput 125 MB/s
io scheduler mq-deadline
nr_requests 126
read_ahead_kb 128
PostgreSQL settings
checkpoint_completion_target = 0.9
cluster_name = 18/main
data_directory = /var/lib/postgresql/18/main
effective_cache_size = 1 GB
effective_io_concurrency = 200
huge_pages = try
jit = off
maintenance_io_concurrency = 200
maintenance_work_mem = 384 MB
max_connections = 100
max_parallel_maintenance_workers = 1
max_parallel_workers = 1
max_wal_size = 4 GB
random_page_cost = 1.1
server_version = 18.6 (Ubuntu 18.6-1.pgdg26.04+2)
shared_buffers = 512 MB
wal_level = replica
work_mem = 16 MB
