CrestVPS

Geekbench 6 · fio · iperf3 · 40 runs each

Benchmarks

Published in full, including the runs that make us look worse than a competitor. The methodology is at the bottom and the script is a single curl.

01 — Head to head

Single-core, Geekbench 6

Geekbench 6 single-core, median of 40 runs on production nodes.

The comparison rows are deliberately unnamed. They are our own measurements on comparable instance types bought at list price, and publishing a competitor’s name against a number we measured ourselves — on hardware we do not control, at a moment we chose — would be dishonest. The methodology is below; run it against whoever you are considering and compare like for like.

Single-core, Geekbench 6 · higher is better

CrestVPS NITRO3480
Hyperscaler C7a2110
Hyperscaler N2D1690
Budget KVM host1240

Storage latency, p99, milliseconds · lower is better

CrestVPS NITRO0.28 ms
Hyperscaler C7a0.61 ms
Hyperscaler N2D0.74 ms
Budget KVM host1.92 ms

02 — Every size

Multi-core, Geekbench 6

PlanvCPUSingleMultiMulti-core4K IOPSStorageSeq read
PULSE P1221804,120186,0005,100 MB/s
PULSE P2421807,940224,0005,600 MB/s
PULSE P36218011,480268,0006,100 MB/s
PULSE P48218014,900312,0006,400 MB/s
TORQUE T1417209,800402,00011,200 MB/s
TORQUE T28172018,900512,00012,400 MB/s
TORQUE T316172035,600684,00013,800 MB/s
TORQUE T432172068,200902,00014,600 MB/s
NITRO N14348013,100596,00012,900 MB/s
NITRO N28348025,400742,00013,900 MB/s
NITRO N312348036,800884,00014,400 MB/s
NITRO N416348044,1001,020,00014,900 MB/s
FORGE F14226011,900498,00012,100 MB/s
FORGE F28226022,800646,00013,200 MB/s
FORGE F316226043,600812,00014,100 MB/s
FORGE F424226062,4001,140,00014,800 MB/s
VAULT V1418408,60062,0002,400 MB/s
VAULT V26184011,40078,0002,800 MB/s
VAULT V38184014,80094,0003,100 MB/s
VAULT V412184026,900118,0003,600 MB/s
FUSION G1 · RTX 4000 Ada8178018,400620,00013,100 MB/s
FUSION G2 · RTX 4090 24 GB12178027,600780,00013,900 MB/s
FUSION G3 · L40S 48 GB16178036,200940,00014,600 MB/s
FUSION G4 · H100 80 GB NVL24178052,8001,180,00014,900 MB/s

03 — Methodology

Methodology

Forty runs per configuration, spread over seven days, on production nodes carrying real customer load. Median reported, not best. Storage figures come from fio with a 4K random-read profile, queue depth 32, sixteen jobs, on a fully provisioned volume. Network figures are from iperf3 between two of our own PoPs and from RIPE Atlas probes for third-party paths.

Reproduce it yourself

On any CrestVPS instance, run the command below. It prints the same table you see above for your own machine.

# one command, no dependencies $ curl -sL crestvps.com/bench | sh # geekbench 6 …………… single 3480 / multi 25400 # fio randread 4k qd32 … 742 130 IOPS # fio latency p99 ……… 0.28 ms # fio seq read 1M ……… 13 914 MB/s # iperf3 → AMS1 ……… 9.41 Gbit/s # mtr → 1.1.1.1 ………… 0.9 ms, 0.0% loss

The script writes nothing outside /tmp and uninstalls itself when it finishes.

04 — Silicon

Server families

PULSEAMD Ryzen 9 7950X5.7 GHz boost
TORQUEAMD EPYC 9754 Bergamo3.1 GHz all-core
NITROAMD Ryzen 9 9950X5.9 GHz pinned
FORGEIntel Xeon 6544Y4.1 GHz turbo
VAULTAMD EPYC 93543.8 GHz boost
FUSIONEPYC 9654 + NVIDIA acceleratorPCIe Gen5 x16 passthrough