linux-kernel - User interactivity measurements.

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [day] [month] [year] [list]

Date:	Sat, 25 Sep 2010 12:02:29 +0100
From:	James Courtier-Dutton <james.dutton@...il.com>
To:	LKML Mailing List <linux-kernel@...r.kernel.org>
Subject: User interactivity measurements.

Hi,

I was trying to work out how I would measure how good the user interactivity is.
I thought that I could use the keyboard repeat rate to do this.
The repeat rate should be constant, so if I measure it, any large
variations indicate latency problems.
The attach program measures the repeat rate that is observed by the application.
To use the program, just keep the ENTER key down until the program exists.
Compile it with "gcc k.c"
Run it with "./a.out >output" and then examine the output in file "output"

Example output under xfce4
0.050032, 0
0.049969, 0
0.050017, 0
0.049970, 0
0.050034, 0
0.049970, 0
0.050038, 0
0.049962, 0
0.049934, 0
0.049987, 0
0.049980, 0
0.060135, 1
0.049878, 0
0.050150, 0
0.049934, 0

As you can see, only one value that is far away from the normal of
0.050000 seconds.
The 1 in the last column tries to identify problem values.

Example output from the same computer, but under gnome and compiz.
0.049984, 1
0.040191, 1
0.029801, 0
0.040048, 1
0.040022, 1
0.029998, 0
0.049945, 1
0.030110, 0
0.049925, 1
0.050020, 1
0.039993, 1
0.039969, 1

As you can see, the variability under gnome and compiz is far worse
than xfce4 with the normal value being about 0.030000 seconds. In
fact, it is so variable, that it is hard to tell what the normal value
should be!!!  There appear to be values clumped around 0.4, and around
0.3 and around 0.5.

Can anyone spot any problems with this approach to measuring the user
interactivity and perceived latency.
For me, the latency was far worse, to the eye, on gnome/compiz than
xfce4 and these results seem to prove it.

Kind Regards

James

View attachment "k.c" of type "text/x-csrc" (1009 bytes)