Re: [PD] Troubleshooting help CPU Overload
Luigi Rensinghoff wrote: >> >> use a profiler (e.g. valgrind, shark,...) > > What is that ??? to keep in line with roman: http://en.wikipedia.org/wiki/Profiler_%28computer_science%29 fmgasd.r IOhannes ___ Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
Re: [PD] Troubleshooting help CPU Overload
On Tue, 2008-06-17 at 02:02 +0200, Luigi Rensinghoff wrote: > Could you explain that a little more ??? denormals, memory > allocation ?? denormals: http://en.wikipedia.org/wiki/Denormal computing denormal numbers is more expensive than normal numbers. this effect is dependent on cpu type (and architecture?). memory allocation: http://en.wikipedia.org/wiki/Memory_allocation roman ___ Der frühe Vogel fängt den Wurm. Hier gelangen Sie zum neuen Yahoo! Mail: http://mail.yahoo.de ___ Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
Re: [PD] Troubleshooting help CPU Overload
Actually the problem disappeared... Just in case it appears again i have some more questions... Am 13.06.2008 um 13:50 schrieb IOhannes m zmoelnig: Luigi Rensinghoff wrote: Hi List... I have a patch that drives the cpu-load up Wha do you think is the most common source for that ?? streams of messages (i often find that people just send tons of statechanges triggering all kind of superfluous calculations, while they really are just interested in the "settled" state) vline~ naive search implementations, denormals, memory allocation,... Could you explain that a little more ??? denormals, memory allocation ?? How should i pin down the malicious circuit/external or whatevr it is.. hmm... try to find out when problems occur (do you have constant load? spikes? is the CPU load growing while the patch is running?) what does this tell you regarding your program logic? try to disable parts of the patch (by deleting it). does the problem go away? try to narrow down the culprit. use [print] for debugging rather than the numberbox (nbx will not tell you if a number changes 1 times per tick) use a profiler (e.g. valgrind, shark,...) What is that ??? It seems not to be related to Audio-Processing even with audio turned off, [vline~] has a known bug (fixed by claude, but i don't know whether this has already made it into Pd) that would start eating CPU if you keep sending it messages. there might be other objects that have similar problems (that is audio-objects that start eating CPU without dsp-on) it could have been that actuallyi realized it was behaving very strange and replaced it fgmasdr IOhannes ___ Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/ listinfo/pd-list >---< Luigi Rensinghoff [EMAIL PROTECTED] skype:gigischinke ichat:gigicarlo ___ Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
Re: [PD] Troubleshooting help CPU Overload
Luigi Rensinghoff wrote: > Hi List... > > I have a patch that drives the cpu-load up > > Wha do you think is the most common source for that ?? streams of messages (i often find that people just send tons of statechanges triggering all kind of superfluous calculations, while they really are just interested in the "settled" state) vline~ naive search implementations, denormals, memory allocation,... > How should i pin down the malicious circuit/external or whatevr it is.. hmm... try to find out when problems occur (do you have constant load? spikes? is the CPU load growing while the patch is running?) what does this tell you regarding your program logic? try to disable parts of the patch (by deleting it). does the problem go away? try to narrow down the culprit. use [print] for debugging rather than the numberbox (nbx will not tell you if a number changes 1 times per tick) use a profiler (e.g. valgrind, shark,...) > > It seems not to be related to Audio-Processing > even with audio turned off, [vline~] has a known bug (fixed by claude, but i don't know whether this has already made it into Pd) that would start eating CPU if you keep sending it messages. there might be other objects that have similar problems (that is audio-objects that start eating CPU without dsp-on) fgmasdr IOhannes ___ Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
[PD] Troubleshooting help CPU Overload
Hi List... I have a patch that drives the cpu-load up Wha do you think is the most common source for that ?? How should i pin down the malicious circuit/external or whatevr it is.. It seems not to be related to Audio-Processing Thanks Luigi >---< Luigi Rensinghoff [EMAIL PROTECTED] skype:gigischinke ichat:gigicarlo ___ Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
