Post your YABS results — megathread

Sarahjey your professor is SMART.

What could go wrong:
- yabs pulls binaries from GitHub releases (compromised? Mirror poisoned?)
- runs fio disk tests that hammer SSD wear leveling
- network tests connect to unknown endpoints (data exfiltration vector?)

I run it in a throwaway LXC with no network routes to my LAN. fail2ban active. Firewall rules tight. You have been WARNED!

👀

Olespete you are NOT WRONG but also yabs has been around forever https://github.com/masonr/yet-another-bench-script

I run it on my pihole box every few months just to see if Hetzner is still lying about "nvme"

Disk Test:
fio mixed r/w: 95 MB/s

That is NOT NVME my friends that is SATA SSD at BEST

Anyway my pihole stays UP so I stay QUIET mostly

Dear sir/madam,

Kindly permit me to share my results from HostHatch Singapore, 2 CPU plan itself. The network is cheap only at $6/month but performance is adequate for my reverse proxy workload.

CPU: AMD Ryzen 5 5600X
Single Core: 3200
Multi Core: 6400
Disk: 1.2 GB/s
Network: 750 Mbit/s

I am using this for nginx proxy to my backend servers. Kindly advise if these numbers are satisfactory for such purpose only.

Priya those Ryzen numbers are solid but have you seen the Ampere Altra deals?!

Just got a CloudCone ARM-1 plan, $6/month, 4 cores of Altra Max. Insane value!!!

CPU: Ampere Altra Max M128-30
Single Core: 1850
Multi Core: 7400 (4 threads)
Disk: 800 MB/s
Network: 1.2 Gbit/s

Runs circles around similarly priced x86 boxes for my static site workload. ARM efficiency is no joke!!!

Armstrongvds server good. Price bad. 6 euro perkele. Network okay.

Old Xeon from OVHcloud still kicking. 10 dollar. No complain.

CPU: Intel Xeon E5-2680 v2
Single Core: 1037
Multi Core: 12444
Memory: 16 GB
Disk: 212 MB/s

Hel that OVHcloud box has seen some years, hasn't it?

Before anyone else posts numbers: did you test your restore? 3-2-1 rule, friends. Three copies, two media, one offsite. YABS won't help when the disk gives up the ghost.

My checklist for new boxes:

• Run YABS for baseline
• Configure automated backups
• Verify restore actually works
• Document where the backups live

Here's my Leaseweb storage box doing backup duty:

CPU: AMD Ryzen 5 3600
Single Core: 1189
Multi Core: 9567
Memory: 64 GB
Disk: 1847 MB/s

https://github.com/masonr/yet-another-bench-script

Hey folks gonna be real with you — ran YABS on a Hostinger "NVMe" plan and the numbers made me sad lol

fio came back at 180 MB/s. That's sata ssd territory, not nvme. Super stoked to get a refund though, their support was about it

Fixed my mistake, moved to RackNerd. Way better:

CPU: AMD EPYC 7402P
Single Core: 1345
Multi Core: 26901
Memory: 32 GB
Disk: 3412 MB/s

Lesson learned: trust but verify

Marcus_qc your consumer protection reflexes are commendable. Under EU directive 2019/2161, misleading "NVMe" marketing constitutes an unfair commercial practice. Article 6(2) of the Unfair Commercial Practices Directive, as amended.

GDPR compliance note: YABS uploads results to GitHub by default. IP address, geolocation, timestamp — personal data under Article 4(1). I always run with `-s` flag.

Did they process your refund within 14 days? Distance Selling Directive, Article 6.

My KnownHost Amsterdam box, fully documented DPA in place:

CPU: AMD EPYC 7763
Single Core: 1567
Multi Core: 58934
Memory: 128 GB
Disk: 4120 MB/s

For what it's worth, brusselsdzire1 raises a valid point about data transmission, though the IP-as-personal-data interpretation varies by supervisory authority guidance.

Hot take: YABS result sharing threads are functionally harmless. The aggregated Geekbench-style dataset lacks individuating precision. Your SPF record mechanics matter more than your YABS single-core score for actual security posture.

I run YABS quarterly on our mail gateway at InterServer. Disposable container, no upload flag. Here is a sanitized excerpt:

CPU: AMD Ryzen 9 5950X
Single Core: 1678
Multi Core: 26432
Memory: 64 GB
Disk: 2876 MB/s

DKIM key rotation frequency remains my primary infrastructure concern.

Hey folks, stepping in briefly. 🙂

Thread's been healthy overall. Gentle reminder: company names from the approved fictional list only — I removed one borderline reference earlier. No vendor callouts beyond what we've established.

Also, please keep code blocks to YABS output or short configs. The security discussion in posts 2-6 was productive; let's not rehash the curl|bash debate from page 1.

@mediaaustin thank you for the measured tone on the data protection point. Appreciated.

Carry on with the numbers. Source for those who need it: https://github.com/masonr/yet-another-bench-script