Is there a hardware benchmark tool for Debian???

Here you can discuss every aspect of Debian. Note: not for support requests!
Post Reply
User avatar
Posts: 13446
Joined: 2014-06-01 17:46
Location: /dev/chair

Re: Is there a hardware benchmark tool for Debian???

#1 Post by Head_on_a_Stick »

Black Lives Matter

Debian buster-backports ISO image: for new hardware support

Posts: 967
Joined: 2016-01-07 12:25

Re: Is there a hardware benchmark tool for Debian???

#2 Post by Dai_trying »

There is also hardinfo in the repositories, it gives a few benchmarks but is probably more for system information but might be of use.

User avatar
Posts: 1019
Joined: 2015-11-29 18:23
Location: Lithuania, Vilnius

Re: Is there a hardware benchmark tool for Debian???

#3 Post by None1975 »

You cant try lmbench. it is in non-free area. Also try simple search. Here is example.
OS: Debian 10.3 Buster / WM: xmonad
Debian Wiki | DontBreakDebian, My config files on github

Posts: 1191
Joined: 2018-06-20 15:16
Has thanked: 2 times
Been thanked: 6 times

Re: Is there a hardware benchmark tool for Debian???

#4 Post by CwF »

+1 for Hardinfo. It has improved and expanded lately and is perfect for your plan to step up through generations. Keep in mind benchmark software is highly 'generational', especially video. But you can see how cpus have changed to. I've been using it to see the effects of virtualization and get a sense of how the numbers translate to experience. Day to day running might be a sufficient benchmark really, but numbers are fun.

I have a very very old benchmark from Byte, 1995, that can go to 20 threads in cpu, memory, and disk test. Hardinfo alo includes more reference machines in a compare database and it goes back pretty far, interesting. But that ancient benchmark 'nbench' shows me what feels true; old software runs faster on old hardware! It says my most modern cpu's (cpuid 206x, 306x xeons) have the integer performance of the late 'tualatins' with a 3X floating point improvement, ie 'generational'. Fun stuff. Hardinfo then can show how the 'co-processors' or subsystems, or whatever we call these new additions to cpu's show the improved performance of code removed from the old 'integer' based common core and moved to specialized circuitry, like cryptohash.

Another useful purpose is assessing how software influences cpu clock. Often, virtualized software does not trigger an uptick in clock and will poke along at minimum clock at 100% utilization. These simple benchmarks help me find those cases that need manual host adjustments.

gtkperf might also be useful for video card compares.

Post Reply