Memory and CPU Footprint of Runbox Web Interface Variants

Runbox Team invited customers to a community to talk about the not finished new Runbox 7 interface.

This motivated me to just checked memory and CPU consumption on my two-core cpu PC with 2 GB RAM and present the results hereby and in https://community.runbox.com/t/memory-and-cpu-footprint/684 Runbox .

Test instructions

Open Browser with timing analysis (Linux command “time firefox”)
Click away start page
Invoke Runbox 6/7 login page; adjust zoom to 70% (Runbox 7) or 100% (RMM 6)
Log in to Runbox
Let it run for 15 seconds
Scroll down for about 100 messages
Let it run for 15 seconds
Check Memory footprint (Linux command “top”; freeze top view)
Click Log out
Close browser before Log out page has loaded.

Results RMM6

Raw

    real    1m17,342s
    user    0m31,203s
    sys     0m3,980s

    305604 firefox
    179268 Web Content
     77468 Web Content

Explanation

I needed 1 m 17 s to test RMM6. In this while, the browser firefox took 35 seconds of 1 CPU to load the pages and scroll up and down. Firefox used 305604+179268+77468 = 562340 kB of memory at the end of the test.

Observation: When browser was open and there was no scrolling/loading then all CPUs were shown idle in my cpu indicator.

Results Runbox 7

Raw

    real    1m22,657s
    user    1m12,300s
    sys     0m11,543s
 
    256284 Web Content
    295556 firefox
     77644 Web Content

Explanation

I needed 1 m 23 s to test Runbox 7. In this while, the browser firefox took 1 m 24 s of 1 CPU to load the pages and scroll up and down. Firefox used 256284+295556+77644 = 629484 kB of memory at the end of the test.

Observation: When browser was open there was no time when all CPUs were shown idle in my cpu indicator.

Discussion

This test as shown in “Test instructions” is not perfect but gives a coarse impression on what’s happening. It can be assumed that the higher usage of ressources of new Runbox version in this first coarse test, would, in it's tendency, be demonstrated by most other test instructions which could be invented and which aim some real-usage-behaviour.

Conclusion

This test affirms the general trend that new software versions over the years take more and more resources thus render old PCs, if still working, obsolete because they cann’t anymore fulfil the tasks which they could ten years ago.

I recently saw a presentation on some scientific approach to measure efficiency of software, it was at “Bits&Bäume” conference: see event description.