Understanding the Output

Understanding the Output

In short

A YABS report has up to six blocks: a banner, basic system information, network information, the fio disk table, the iperf3 network tables (IPv4 and/or IPv6), and the Geekbench scores, followed by a one-line runtime footer. This page walks through a complete, current-format report line by line and explains what each value is and where it comes from. For what the values mean and whether they are any good, continue to Interpreting Results.

The sample below is a composite of a real 2025 budget-KVM result and a run of the current script; values are realistic but the server is fictional.

The full report

# ## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ## #
#              Yet-Another-Bench-Script              #
#                     v2026-10-31                    #
# https://github.com/masonr/yet-another-bench-script #
# ## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ## #

Thu Feb 13 07:01:48 AM EST 2025

Basic System Information:
---------------------------------
Uptime     : 0 days, 0 hours, 21 minutes
Processor  : AMD Ryzen 9 7950X3D 16-Core Processor
CPU cores  : 2 @ 4191.912 MHz
AES-NI     : ✔ Enabled
VM-x/AMD-V : ✔ Enabled
RAM        : 3.8 GiB
Swap       : 128.0 MiB
Disk       : 98.2 GiB
Distro     : Debian GNU/Linux 12 (bookworm)
Kernel     : 6.1.0-9-amd64
VM Type    : KVM
IPv4/IPv6  : ✔ Online / ❌ Offline

IPv4 Network Information:
---------------------------------
ISP        : Example Networks GmbH
ASN        : AS64496 Example Networks
Host       : Example Cloud
Location   : Frankfurt am Main, Hesse (HE)
Country    : Germany

fio Disk Speed Tests (Mixed R/W 50/50) (Partition /dev/vda1):
---------------------------------
Block Size | 4k            (IOPS) | 64k           (IOPS)
  ------   | ---            ----  | ----           ----
Read       | 45.43 MB/s   (11.3k) | 461.52 MB/s   (7.2k)
Write      | 45.52 MB/s   (11.3k) | 463.95 MB/s   (7.2k)
Total      | 90.95 MB/s   (22.7k) | 925.48 MB/s  (14.4k)
           |                      |
Block Size | 512k          (IOPS) | 1m            (IOPS)
  ------   | ---            ----  | ----           ----
Read       | 840.34 MB/s   (1.6k) | 853.47 MB/s    (833)
Write      | 884.99 MB/s   (1.7k) | 910.31 MB/s    (888)
Total      | 1.72 GB/s     (3.3k) | 1.76 GB/s     (1.7k)

iperf3 Network Speed Tests (IPv4):
---------------------------------
Provider        | Location (Link)           | Send Speed      | Recv Speed      | Ping
-----           | -----                     | ----            | ----            | ----
Clouvider       | London, UK (10G)          | 1.95 Gbits/sec  | 1.43 Gbits/sec  | 10.7 ms
Eranium         | Amsterdam, NL (100G)      | 2.04 Gbits/sec  | 1.81 Gbits/sec  | 3.52 ms
Uztelecom       | Tashkent, UZ (10G)        | 646 Mbits/sec   | 984 Mbits/sec   | 91.9 ms
Leaseweb        | Singapore, SG (10G)       | 316 Mbits/sec   | 1.05 Gbits/sec  | 177 ms
Clouvider       | Los Angeles, CA, US (10G) | 399 Mbits/sec   | 495 Mbits/sec   | 142 ms
Leaseweb        | NYC, NY, US (10G)         | 620 Mbits/sec   | 1.35 Gbits/sec  | 77.9 ms
Edgoo           | Sao Paulo, BR (1G)        | busy            | 304 Mbits/sec   | 213 ms

Geekbench 6 Benchmark Test:
---------------------------------
Test            | Value
                |
Single Core     | 1702
Multi Core      | 3039
Full Test       | https://browser.geekbench.com/v6/cpu/12345678

YABS completed in 9 min 12 sec

Block by block

Banner and date

#                     v2026-10-31                    #
Thu Feb 13 07:01:48 AM EST 2025

• The version is a date (vYYYY-MM-DD) — the day the script was last changed. When comparing results, the banner tells you which era of the script (and therefore which defaults and iperf servers) produced them.
• The date line is simply the system's date in its local time zone. It is the single most important piece of context in the report: a benchmark is a snapshot.
• On RISC-V you will also see RISC-V compatibility is considered *experimental* here; a missing C locale produces a warning about parsing.

Basic System Information

Line	Example	What it is	Source
Uptime	0 days, 0 hours, 21 minutes	How long the OS has been up. 21 minutes = a freshly created VM.	uptime, parsed with awk
Processor	AMD Ryzen 9 7950X3D 16-Core Processor	The CPU model string the hypervisor exposes. Some providers show generic names (QEMU Virtual CPU, Common KVM processor). The "16-Core" in the name is the physical chip, not your allocation.	/proc/cpuinfo model name (x86); lscpu on ARM; several fallbacks on RISC-V
CPU cores	2 @ 4191.912 MHz	vCPUs allocated to you, and the current clock of one of them at the moment the script started. Idle servers show low MHz; that is normal.	count of model name lines; last cpu MHz
AES-NI	✔ Enabled	Whether hardware AES instructions are exposed to the guest (matters for TLS, SSH, VPNs, encrypted disks).	aes flag in /proc/cpuinfo (Zkn* in the RISC-V ISA string)
VM-x/AMD-V	✔ Enabled	Whether hardware virtualisation extensions are exposed (needed to run VMs inside this server). Disabled on most VPS plans; always enabled on bare metal.	vmx/svm flags (h extension on RISC-V)
RAM / Swap	3.8 GiB / 128.0 MiB	Total memory and swap in binary units (GiB = 1024³). 3.8 GiB is what a "4 GB" plan looks like after the kernel reserves its share.	free
Disk	98.2 GiB	Sum of all mounted local filesystems (ext2/3/4, btrfs, xfs, vfat, exfat, ntfs, simfs, swap), each device counted once. Network mounts are excluded.	/proc/mounts + stat -f, fallback df --total
Distro	Debian GNU/Linux 12 (bookworm)	The OS release.	/etc/os-release PRETTY_NAME
Kernel	6.1.0-9-amd64	Kernel version. Container platforms (LXC/OpenVZ) show the host's kernel.	uname -r
VM Type	KVM	Virtualisation technology: KVM, XEN, VMWARE, MICROSOFT, LXC, OPENVZ, DOCKER, WSL, NONE (bare metal), or UNKNOWN.	systemd-detect-virt, upper-cased
IPv4/IPv6	✔ Online / ❌ Offline	Whether each protocol reached the internet during the start-up check. Decides which iperf tables run and whether Geekbench can be downloaded.	ping to ipv4/ipv6.google.com, or curl/wget to icanhazip.com

Details, per-architecture differences and edge cases: System Information.

Network Information

IPv4 Network Information:
ISP        : Example Networks GmbH
ASN        : AS64496 Example Networks
Host       : Example Cloud
Location   : Frankfurt am Main, Hesse (HE)
Country    : Germany

Who owns the IP address the server uses to reach the internet and where it is registered. Note that this is the geolocation database's view: IP ranges announced from one city can be located elsewhere, and "Host" (the organisation field) may name the provider's upstream rather than the brand you bought from. The protocol in the heading (IPv4 or IPv6) is whichever the host preferred when contacting ip6.me. The lookup itself uses ip-api.com; because that API is IPv4-only, IPv6-only servers show no Network Information block. Skip the whole block with -n (useful when sharing results publicly).

fio Disk Speed Tests

fio Disk Speed Tests (Mixed R/W 50/50) (Partition /dev/vda1):
Block Size | 4k            (IOPS) | 64k           (IOPS)
Read       | 45.43 MB/s   (11.3k) | 461.52 MB/s   (7.2k)
Write      | 45.52 MB/s   (11.3k) | 463.95 MB/s   (7.2k)
Total      | 90.95 MB/s   (22.7k) | 925.48 MB/s  (14.4k)

• Header. Mixed R/W 50/50 reminds you that every number comes from a test issuing reads and writes simultaneously in equal measure. Partition names the block device holding the directory you ran the script from — the only disk tested.
• Layout. Two tables of two block sizes each: 4k and 64k, then 512k and 1m. Each cell is throughput (IOPS).
• Read / Write / Total. Read-side and write-side figures from the same 30-second run, and their sum. Because the mix is 50/50, Read and Write are nearly identical by design.
• Units. Throughput is in decimal MB/s or GB/s (1 MB = 1,000,000 bytes), converted from fio's KiB/s. IOPS are abbreviated with k for thousands (11.3k = 11,300 operations per second; 833 is just 833).
• If this block is titled dd Sequential Disk Speed Tests instead, fio could not be downloaded or failed to run, and YABS fell back to three dd passes. Those are sequential, single-threaded numbers (and the read side is cache-inflated) — not comparable to fio results.
• Warnings that may precede it: Less than 2GB of space available. Skipping disk test..., or the ZFS free-space warning (which means the numbers that follow are unreliable).

How the test works: Tests: Disk (fio). What the numbers mean: Interpreting Results § Disk.

iperf3 Network Speed Tests

Provider        | Location (Link)           | Send Speed      | Recv Speed      | Ping
Clouvider       | London, UK (10G)          | 1.95 Gbits/sec  | 1.43 Gbits/sec  | 10.7 ms
Edgoo           | Sao Paulo, BR (1G)        | busy            | 304 Mbits/sec   | 213 ms

• One table per protocol. The IPv4 table appears if the server has IPv4; an IPv6 table follows if it has IPv6. The sample server has no IPv6, so there is only one.
• Provider / Location (Link). Who donated the remote iperf3 server, where it is, and the speed of its port. The remote port is an upper bound on what you can see at that row.
• Send Speed. Throughput from your server to the remote — your upload — measured as the remote received it.
• Recv Speed. Throughput from the remote to your server — your download.
• Ping. One ICMP round trip to that server. It mostly tells you how far away it is.
• Units. Bits per second (Gbits/sec, Mbits/sec). Divide by 8 for bytes: 1.95 Gbit/s ≈ 244 MB/s.
• busy. No usable result after three attempts in that direction (server full, refused, or returned 0.00). It is about the remote server at that moment, not about you.
• With -r only three locations are tested; with -p your own list replaces the defaults; with -i this block is absent.

How the test works: Tests: Network (iperf3). What the numbers mean: Interpreting Results § Network.

Geekbench Benchmark Test

Geekbench 6 Benchmark Test:
Single Core     | 1702
Multi Core      | 3039
Full Test       | https://browser.geekbench.com/v6/cpu/12345678

• Version. Geekbench 6 by default; -4, -5, -7 select other versions and -9 runs 4 and 5. Scores from different versions are on different scales and must never be compared.
• Single Core. Performance of one thread. Multi Core. Aggregate across all vCPUs.
• Full Test. A public page on Geekbench's site with every sub-test, the detected hardware, and comparisons. Anyone with the link can view it. The claim URL (to attach the result to your Geekbench account) is not printed; it is appended to a file named geekbench_claim.url in the directory you ran from.
• If this block is missing you will see one of the explanatory messages instead — for example about CentOS 7 and glibc, about IPv4 being required to download Geekbench, about low memory, or about tar not being installed. See Troubleshooting and FAQ.

How the test works: Tests: CPU (Geekbench). What the numbers mean: Interpreting Results § CPU.

Footer

YABS completed in 9 min 12 sec

Wall-clock time from the banner to the end of clean-up. Typical full runs take 5–15 minutes; Geekbench dominates on slow CPUs, and iperf retries against busy servers add up (each failed attempt costs up to 15 s plus a 2 s pause).

If you passed -j, the JSON document is printed after this line; -w writes it to the file you named; -s posts it to the URL(s) you gave. See JSON Output and Integrations.

What is left on disk after a run

• Nothing from the tests themselves: the temporary directory (./<timestamp>/) holding the fio test file, the downloaded binaries and the Geekbench extraction is deleted at the end (and on Ctrl+C).
• geekbench_claim.url (appended to on every Geekbench run) in the current directory.
• Any file you requested with -w.

Reading a report quickly: the 30-second scan

1. Banner date and Uptime — when, and how fresh is the VM?
2. Processor and VM Type — what hardware class and virtualisation are we looking at?
3. fio 4k Total IOPS and 1m Total MB/s — responsive? wide? Do they agree on a storage class?
4. iperf3 nearest rows — does the port deliver what was sold?
5. Geekbench single-core, and multi ÷ single — fast core? real cores?
6. Anything surprising → Interpreting Results, then re-run at another hour.

See also

• Interpreting Results
• Benchmarks Are Snapshots
• Command-Line Flags
• System Information
• Glossary