Bugzilla – Bug 246
Uneven motion of playback and recording cursor
Last modified: 2018-08-20 11:45:40 UTC
Instead of moving smoothly, playback cursor has a regular pattern of about three steps per second on Ubuntu. On Win 7 the pattern of steps alternates in a two second cycle, the first second in the cycle having one step, and the final second having two steps. This cycle is sometimes disrupted by unpredictable steps here and there (these appear to correlate with minor CPU spikes). CPU use for me generally remains well below 35%. Recording "stereo mix" from web browser shows the recording cursor behaving the same way. Does not appear to be an issue of inappropriate video drivers or inadequate resources. Mac behaviour not known yet.
(In reply to comment #0) Do you mean, e.g., on Windows that it jumps from 1.0s to 2.0s, then 2.5s, then 3.0s? I'm sure not seeing that on my Win XP machine, and the motion is visually the same in 1.2.6 and HEAD, even in Debug mode where CPU gets to ~60% on playback.
(In reply to comment #0) On Ubuntu 10.10 I'm seeing the playback cursor move at a rate of about 3 steps per second when using ALSA/Pulse. Personally I find this acceptable. When using Jack it becomes very much smoother (I'd estimate about 10 steps per second).
(In reply to comment #1) > Do you mean, e.g., on Windows that it jumps from 1.0s to 2.0s, then 2.5s, > then 3.0s? I'm sure not seeing that on my Win XP machine, and the motion is > visually the same in 1.2.6 and HEAD, even in Debug mode where CPU gets to > ~60% on playback. I see the same on Win 7 and XP with Unicode Release. It is harder to describe on Windows than Ubuntu. The steps themselves on Windows are the same distance as on Ubuntu (one-third of a second), but between the regular steps and the randomly occurring steps the cursor seems to move smoothly. So ignoring the random steps, playing from zero the cursor starts smoothly, then when it reaches 0.3s, it jumps to 0.6s. It continues smoothly to 1.3s, jumps to 1.6s, jumps to 1.9s, smoothly to 2.3s, jumps to 2.6s, smoothly to 3.3s, jumps to 3.6s, jumps to 3.9s.... The pattern is definitely there but especially on Windows it is hard to detect unless you are concentrating on the cursor instead of the waveform.
On Mac, Bruno reports something that looks like the three steps per second motion seen on Ubuntu. I tried an ASIO Unicode Release build from several months ago on Win 7. I saw something more like Steve sees on Jack (a slightly erratic "seven or eight steps per second" motion). When I try DirectSound (on XP, where it's not emulated) I see behaviour like ASIO, but four or five steps per second. So like Steve suggests in the Forum, behaviour seems to reflect the efficiency of the audio transport system and probably isn't directly related to screen refresh.
This bug becomes significantly worse for me on Ubuntu (but not on Windows) as soon as I open a label track, even if there are no labels. Playback and recording cursor movement still about three steps per second on Ubuntu, but some of the steps appear to hang so that the next step is omitted and there is a jump of two steps (2/3rds second). Display of vertical line in waveform is very poor when there is a label track. On the "hangs" above, the line either momentarily disappears, or presents itself as three close-together black lines instead of red or green. However this does not appear to be any worse as a result of fixes for bug 148 or bug 255, comparing 1.3.12 with HEAD. Promoted to P4 on grounds of deterioration when a label track is open.
Testing on latest alphas on Win and Mac this is no longer an issue
