Should I remove these leftovers from the temporary folder before I run the test for the next kernel?
bash-scripts benchie.QiWb blender.zip firefox60.tar pkg stress-ng
benchie_2020-08-27-1600.log blender ffmpeg-d3b963c pi src
The script should take care of necessary cleanups.
Alright, then I leave these.
Nope . It’s hard disk , no compression .
Here it goes:
Machine: Type: Laptop System: Dell product: Vostro 5471 v: N/A serial: <filter> Chassis: type: 10 serial: <filter>
Mobo: Dell model: 08X87Y v: A00 serial: <filter> UEFI: Dell v: 1.15.0 date: 07/03/2020
CPU: Info: Quad Core model: Intel Core i5-8250U bits: 64 type: MT MCP arch: Kaby Lake rev: A L2 cache: 6144 KiB
flags: avx avx2 lm nx pae sse sse2 sse3 sse4_1 sse4_2 ssse3 vmx bogomips: 28800
Speed: 3400 MHz min/max: 400/3400 MHz Core speeds (MHz): 1: 3400 2: 3401 3: 3400 4: 3401 5: 3400 6: 3401 7: 3400
8: 3400
Graphics: Device-1: Intel UHD Graphics 620 vendor: Dell driver: i915 v: kernel bus ID: 00:02.0 chip ID: 8086:5917
Device-2: Microdia Integrated_Webcam_HD type: USB driver: uvcvideo bus ID: 1-5:4 chip ID: 0c45:6a06
Display: x11 server: X.Org 1.20.9 driver: intel unloaded: fbdev,modesetting,vesa resolution: 1920x1080~60Hz
s-dpi: 96
cpufreq >> perormance
Kernel: 5.8.4-zen1-1-zen x86_64 bits
systemd-analyze
Startup finished in 1.595s (kernel) + 3.374s (userspace) = 4,969s
graphical.target reached after 2.777s in userspace
/usr/bin/mini-benchmarker.sh
======= ========================== ======
====| / |==== MINI BENCHMARKER ====| ___ ))====
====| |/| |======= by torvic9 =======| ___ =====
====||==||========= v1.1 =========|//====
- stress-ng cpu: 104.77
- stress-ng memory: 63.17
- perf sched: 73.019
- perf memcpy: 11.13
- calculating 60m digits of pi: 52.56
- argon2 hashing: 19.42
- ffmpeg compilation: 167.76
- xz compression: 207.73
- Blender render: 640.34
Total time in seconds:
1339.899
Total score (lower is better) [multi = 2.0014]:
148.37
Kernel: 5.7.18.a-1-hardened
systemd-analyze
Startup finished in 3.501s (kernel) + 3.554s (userspace) = 7.055s
graphical.target reached after 3.313s in userspace
/usr/bin/mini-benchmarker.sh
======= ========================== ======
====| / |==== MINI BENCHMARKER ====| ___ ))====
====| |/| |======= by torvic9 =======| ___ =====
====||==||========= v1.1 =========|//====
- stress-ng cpu: 107.02
- stress-ng memory: 63.59
- perf sched: 75.106
- perf memcpy: 6.95
- calculating 60m digits of pi: 53.27
- argon2 hashing: 20.23
- ffmpeg compilation: 170.78
- xz compression: 216.19
- Blender render: 645.70
Total time in seconds:
1358.836
Total score (lower is better) [multi = 2.0014]:
148.70
No clear winner as it looks. However linux-zen boots up quite faster than linux-hardened on this machine.
Chill out guys!
Wait a minute ! Who is the winner ???
Switched to zen 5.8.3 (using stable branch here), it seems faster for me than 5.8.3 stable.
So far no compatibility issues.
Thanks, that is very interesting.
I am surprised by the memcpy bench result.
Have to investigate why hardened is faster there.
However to be completely fair we would have to compare 5.8 kernels.
Yes, that was surprising. linux-hardened is almost 62% faster there.
Right!
Anyways, thanks for the script. It was a fun exercise.
By the way, how is the total score calculated?
The 5.8 kernel uses schedutil as the default CPU governor (at least for Intel chips, I’m not sure about Ryzen), versus 5.7 and below’s powersave. That might make comparisons between the two a bit difficult.
The score is actually just a useless value.
I wanted to provide a “normalized” value because the total time can vary widely.
It takes into account the number of CPUs, and applies a square root to the actual time of each bench in order to “compress” the result.
The script asks the user if he wants to switch to performance gov.
In both cases, the cpufreq was set to performance.
In this occasion and on this machine, linux-zen took 18,937 seconds less than linux-hardened to complete the tasks. That is almost 1,39% faster.
So meditation pays off!
So I am staying with zen 
linux-zen and linux are now 5.8.6 and linux-hardened is 5.8.7
Main kernel is 5.8.7-arch1-1 - maybe mirror update speed?
But how did the hardened get to 5.8.7 this fast ?
It lagged one version behind
or Did they skip 5.8.6 ??
linux-zen-5.8.7 just trickled down here