Hi List,
I've got used to putting my PC to sleep (aka hibernation) often lately. Now there is this behaviour of Pd that when you leave a patch open and put the computer to sleep, once it wakes up Pd will try to do everything it missed while the computer was sleeping, so the CPU goes 100% for quite a while. I suppose this is by design. What I've just noticed using l2ork is that I had closed my patch before hibernating (in order to avoid the CPU boost when waking up), put the computer to sleep for a few hours, and when i woke it up, surprisingly the 100% CPU boost still happened - with only the main window and console open. This makes me think some things are not destroyed properly when a patch is closed. Any thoughts appreciated...
András
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
On 2013-06-18 22:58, András Murányi wrote:
Hi List,
I've got used to putting my PC to sleep (aka hibernation) often lately. Now there is this behaviour of Pd that when you leave a patch open and put the computer to sleep, once it wakes up Pd will try to do everything it missed while the computer was sleeping, so the CPU goes 100% for quite a while. I suppose this is by design. What I've just noticed using l2ork is that I had closed my patch before hibernating (in order to avoid the CPU boost when waking up), put the computer to sleep for a few hours, and when i woke it up, surprisingly the 100% CPU boost still happened - with only the main window and console open. This makes me think some things are not destroyed properly when a patch is closed. Any thoughts appreciated...
i'm experiencing the same, and have not found the reason, nor did i come up with a nice way how to easily debug this problem without doing time consuming suspend/resume cycles (at least you don't need to *hibernate* the system, a simple *suspend* to RAM triggers the problem as well).
my system (as always) is:
- debian jessie/sid (x86_64)
- pd-vanilla 0.43, 0.44 (most likely; i have only recently started to
use 0.44 on a regular basis)
personally i don't think that this is related to some improper cleanup (as it also happens with minimal patches that do not really do anything "on their own"), but rather that Pd is checking the (empty?) event-queue in tiny increments rather than one big go. or it might be related to an overly simple watchdog logic.
but these are wild guesses.
ghsmadf IOhannes
On Tue, 2013-06-18 at 22:58 +0200, András Murányi wrote:
Hi List,
I've got used to putting my PC to sleep (aka hibernation) often lately. Now there is this behaviour of Pd that when you leave a patch open and put the computer to sleep, once it wakes up Pd will try to do everything it missed while the computer was sleeping, so the CPU goes 100% for quite a while. I suppose this is by design.
What I've just noticed using l2ork is that I had closed my patch before hibernating (in order to avoid the CPU boost when waking up), put the computer to sleep for a few hours, and when i woke it up, surprisingly the 100% CPU boost still happened - with only the main window and console open.
This makes me think some things are not destroyed properly when a patch is closed.
Any thoughts appreciated...
Does it help to toggle DSP off and on, when Pd is in this post-suspend mode? I had the impression it did, but I wasn't sure if it was just a coincident, that CPU usage stopped at the same moment.
Roman
I don't think it is incomplete freeing of patch resources based on what I saw so far. Could it be the watchdog's ping, isn't each of these tied to a specific timestamp? On Jun 19, 2013 3:50 AM, "Roman Haefeli" reduzent@gmail.com wrote:
On Tue, 2013-06-18 at 22:58 +0200, András Murányi wrote:
Hi List,
I've got used to putting my PC to sleep (aka hibernation) often lately. Now there is this behaviour of Pd that when you leave a patch open and put the computer to sleep, once it wakes up Pd will try to do everything it missed while the computer was sleeping, so the CPU goes 100% for quite a while. I suppose this is by design.
What I've just noticed using l2ork is that I had closed my patch before hibernating (in order to avoid the CPU boost when waking up), put the computer to sleep for a few hours, and when i woke it up, surprisingly the 100% CPU boost still happened - with only the main window and console open.
This makes me think some things are not destroyed properly when a patch is closed.
Any thoughts appreciated...
Does it help to toggle DSP off and on, when Pd is in this post-suspend mode? I had the impression it did, but I wasn't sure if it was just a coincident, that CPU usage stopped at the same moment.
Roman
Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
This may be related: after upgrading to Fedora core 17 (linux 3.6.3-1.fc17.x86_64) I found that, after running a patch for hours or days, just shutting DSP off sometimes freezes my machine for somewhere between 1 and about 20 seconds (I think). I had never suspended or hibernated the machine. I don't know how to debug this as the machine is frozen while it's happening :) Anyhow I don't get it on other machines so I'm suspicious it's a particularity with some range of linux kernels.
Miller
On Tue, Jun 18, 2013 at 10:58:41PM +0200, András Murányi wrote:
Hi List,
I've got used to putting my PC to sleep (aka hibernation) often lately. Now there is this behaviour of Pd that when you leave a patch open and put the computer to sleep, once it wakes up Pd will try to do everything it missed while the computer was sleeping, so the CPU goes 100% for quite a while. I suppose this is by design. What I've just noticed using l2ork is that I had closed my patch before hibernating (in order to avoid the CPU boost when waking up), put the computer to sleep for a few hours, and when i woke it up, surprisingly the 100% CPU boost still happened - with only the main window and console open. This makes me think some things are not destroyed properly when a patch is closed. Any thoughts appreciated...
András
Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
-
András Murányi -
IOhannes m zmoelnig -
Ivica Bukvic -
Miller Puckette -
Roman Haefeli