Current linux-bench for STH reviews?

Notice: Page may contain affiliate links for which we may earn a small commission through services like Amazon Affiliates or Skimlinks.


New Member
Apr 28, 2021
Anyone know if there's anything more current than GitHub - STH-Dev/linux-bench: STH Linux benchmarking script Looks like it was updated in 2017, but only to change the copywrite.

I see reviews (like the new nuc 12 pro) that includes benchmarks for a Linux-4.4.2 compile, 7-zip compression, OpenSSL Sign and Verify, MariaDB, and C-ray.

My motivation is I'd like to compare hardware that hasn't been reviewed to hardware reviewed. Maybe even drumming up interest in the forums for future STH reviews. In particular I just have a Nanopi R6s with the CNC machined/passively cooled case. It's in the general category as the various intel based fanless firewalls. It's got 8 cores, 8GB ram, two 2.5G ethernet ports, and can do 3GB/sec to NVMe, if you add one. Of course the question is, how fast is the RK3588 chip compared to the Celeron J4125/N5095/N5105 and Pentium N6005? I'm betting pretty competitive, on a compile benchmark it was *seven* times faster than a RPI4 with 8GB ram.

I can happily get 7-zip, kernel compiles, and the like going, but it's not a good reference point unless the same exact flags/settings are used. Even things as simple as file compression can be impacted by file size, if the filecache is warm, if you writing the output, compression radio/dictionary size, etc. Compiling kernels can change change radically with what functionality is enabled, how many threads are used, etc.

Anyone know how to replicate the benchmarks used in a recent review, like:

If not, maybe we can just pick some standard options so we can compare amongst ourselves. Maybe something like:
  • openssl speed -bytes 16384 sha256 sha512 aes-256-cbc rsa2048
  • 7z b
  • c-ray 2.0: make; time ./c-ray-fast ./sphfract.scn -s 3840x2160 -r 4 -o output.pnm -p
Suggestions for benchmark submissions:
  • Open source
  • Easy to configure (not much more than wget/git clone, tar xvf <file>, and make
  • Runs in less than 5 minutes on a 2015 ish quad core desktop
  • Produces results that can be summarized in less 5 numbers.
  • Can run in 8GB ram or less


Jun 15, 2014

This might suite your use case but it's probably a lot complicated and take more than 5 mins to run if you ran multiple tests.


New Member
Apr 28, 2021
Thanks, I've tinkered with it, but was hoping for some overlap with the STH posted benchmarks for easy comparisons. My goal mainly being to compare cool little router/switches/firewalls using arm chips with the collection of STH reviewed J4125, J6413, N5096, N5105, and N6005.

Geekbench5 isn't open source, but does have numbers for most of the above and the Arm base Nano pi R6S with a RK3588 CPU, so maybe that's good enough.
Last edited: