djbambi
Well, I've been following the "memory-leak 2/CPU..." thread for a while, and would like to participate by letting the numbers speak.
I decided to open a new subject because frankly the other one was getting confusing towards page 3 :)
I have a P3 notebook running at 700Mhz , 512 Mb RAM, tidy Windows 2000 Pro.
1) Before Launching VDJ 2.04:
commit charge 86 Mb
2) After Launching VDJ 2.04:
commit charge 121 Mb
virtualdj.exe process => mem usage 27Mb VM size 29Mb
3) After loading a 8 minute track in deck A:
commit charge 208 Mb
virtualdj.exe process => mem usage 114Mb VM size 116Mb
(note: those numbers do not change when I play the track and use loops/effects)
4) After loading a 7:50 minute track in deck B:
commit charge 292 Mb
virtualdj.exe process => mem usage 199Mb VM size 200Mb
(note: those numbers do not change when I play both tracks with loops/multi-effects)
5) After playing A+B simultaneously, then stopping both:
commit charge 292 Mb
virtualdj.exe process => mem usage 4Mb (yes, four!) VM size 200Mb
note: the mem usage starts climbing slowly again if I play one of the tracks again...
6) After loading a 2:54 minute track in deck A:
commit charge 237 Mb
virtualdj.exe process => mem usage 39Mb VM size 145Mb
7) After loading same 8 minute track as step 3 in deck A again:
commit charge 292 Mb
virtualdj.exe process => mem usage 94Mb VM size 200Mb
8) Close virtual DJ
commit charge 86 Mb
My personal conclusion is:
- VDJ is behaving just fine memory-wise
- releasing memory exactly as it should (50+ Mb decrease in step 6)
- reclaiming it in a consistent way (step 7 = step 4)
- performs occasional garbage collecting (step 5) which I don't quite understand since it seems to be freeing up memory for tracks that are currently loaded so potentially not obsolete
I have never run into any issues or crashes, and even after hours of VDJ use my peak Commit Charge as never exceeded 400Mb. Also, my CPU usage only hits 100% at track loads, then stays down the rest of the time. Note that I never analyze tracks during a set, all my tracks are analyzed in advance during long, boring analyze/cue-set/cbg-set/bpm-correct sessions...
Now it's obvious that some people have problems with their use of VDJ, and I don't doubt their words. Maybe it's related to a particular setting? If so, only one way to find out.
Here are my settings, that have always worked just fine:
GENERAL:
--------
Multi-users: Disabled
Security: None
AutoUpdate: Disabled
Tooltip: Enabled
Max Load: 15mn (I never have tracks longer than that)
Scrolling: Inverted
Disc: 33rpm
Pitch Range: -8% to +8%
FAME:
-----
Bpm Engine: any music
Auto BPM: Disabled
Auto Gain: Disabled (I mp3gain my tracks)
Auto Pitch-Reset: Disabled
Force Fade: Disabled
Fade Length: 4s
Crossfader: Full
Limiter: Disabled
Time Stretch: Disable
Disengage BeatLock: Manual
To Andre, and others that seem to encounter memory leak issues:
- could you replay the above steps 1 to 8 with your current config and see if you have different result than me?
- if so, write down your original settings and switch to mine, then perform steps 1 to 8 again and see if it changes anything
Maybe this would shed some light and give something to the dev team to work with.
Regards,
Nicolas
I decided to open a new subject because frankly the other one was getting confusing towards page 3 :)
I have a P3 notebook running at 700Mhz , 512 Mb RAM, tidy Windows 2000 Pro.
1) Before Launching VDJ 2.04:
commit charge 86 Mb
2) After Launching VDJ 2.04:
commit charge 121 Mb
virtualdj.exe process => mem usage 27Mb VM size 29Mb
3) After loading a 8 minute track in deck A:
commit charge 208 Mb
virtualdj.exe process => mem usage 114Mb VM size 116Mb
(note: those numbers do not change when I play the track and use loops/effects)
4) After loading a 7:50 minute track in deck B:
commit charge 292 Mb
virtualdj.exe process => mem usage 199Mb VM size 200Mb
(note: those numbers do not change when I play both tracks with loops/multi-effects)
5) After playing A+B simultaneously, then stopping both:
commit charge 292 Mb
virtualdj.exe process => mem usage 4Mb (yes, four!) VM size 200Mb
note: the mem usage starts climbing slowly again if I play one of the tracks again...
6) After loading a 2:54 minute track in deck A:
commit charge 237 Mb
virtualdj.exe process => mem usage 39Mb VM size 145Mb
7) After loading same 8 minute track as step 3 in deck A again:
commit charge 292 Mb
virtualdj.exe process => mem usage 94Mb VM size 200Mb
8) Close virtual DJ
commit charge 86 Mb
My personal conclusion is:
- VDJ is behaving just fine memory-wise
- releasing memory exactly as it should (50+ Mb decrease in step 6)
- reclaiming it in a consistent way (step 7 = step 4)
- performs occasional garbage collecting (step 5) which I don't quite understand since it seems to be freeing up memory for tracks that are currently loaded so potentially not obsolete
I have never run into any issues or crashes, and even after hours of VDJ use my peak Commit Charge as never exceeded 400Mb. Also, my CPU usage only hits 100% at track loads, then stays down the rest of the time. Note that I never analyze tracks during a set, all my tracks are analyzed in advance during long, boring analyze/cue-set/cbg-set/bpm-correct sessions...
Now it's obvious that some people have problems with their use of VDJ, and I don't doubt their words. Maybe it's related to a particular setting? If so, only one way to find out.
Here are my settings, that have always worked just fine:
GENERAL:
--------
Multi-users: Disabled
Security: None
AutoUpdate: Disabled
Tooltip: Enabled
Max Load: 15mn (I never have tracks longer than that)
Scrolling: Inverted
Disc: 33rpm
Pitch Range: -8% to +8%
FAME:
-----
Bpm Engine: any music
Auto BPM: Disabled
Auto Gain: Disabled (I mp3gain my tracks)
Auto Pitch-Reset: Disabled
Force Fade: Disabled
Fade Length: 4s
Crossfader: Full
Limiter: Disabled
Time Stretch: Disable
Disengage BeatLock: Manual
To Andre, and others that seem to encounter memory leak issues:
- could you replay the above steps 1 to 8 with your current config and see if you have different result than me?
- if so, write down your original settings and switch to mine, then perform steps 1 to 8 again and see if it changes anything
Maybe this would shed some light and give something to the dev team to work with.
Regards,
Nicolas
Posted Sun 24 Oct 04 @ 10:12 pm
Dach
While I appreciate your scientific analysis of the problem, I don't think your sample was "large" enough. See if it maintains that same consistency after 3-4 hours, which is more of a real-world situation. I recently did a 6-hour stress test on my system, and it performed flawlessly. The problem is, there was a good deal of automixing.
HOWEVER, if I use the EQ (or even click on the EQ button to look at it), the program freezes (i.e., no movement, and the music stops completely) for 2 full seconds, then the memory is all messed up after that. Songs will stutter while fading out, and no memory will be freed until I restart the program. So, my workaround is to have an external EQ.
So, while it's good to give the dev team as much info as possible, I think the memory problem is there, but it's really inconsistent.
Däch
http://www.closedcaskets.com/
http://www.clubapparition.info/
HOWEVER, if I use the EQ (or even click on the EQ button to look at it), the program freezes (i.e., no movement, and the music stops completely) for 2 full seconds, then the memory is all messed up after that. Songs will stutter while fading out, and no memory will be freed until I restart the program. So, my workaround is to have an external EQ.
So, while it's good to give the dev team as much info as possible, I think the memory problem is there, but it's really inconsistent.
Däch
http://www.closedcaskets.com/
http://www.clubapparition.info/
Posted Sun 24 Oct 04 @ 10:49 pm
djzero
which skin are you using?
Posted Sun 24 Oct 04 @ 11:08 pm
djbambi
djzero> I'm using the mixstation skin.
Dach> test case is small indeed, but I figured if ANdre can recreate his memory leak with it, that'd be the best starting point. If not, then we'll have to go further... I just know that a dev team cannot do anything with a non-reproduceable test case. If it's too abstract, e.g. "it just crashes after four hours", they'll try to recreate, like me won't be able to, and they'll blame it on murphy's law and there really isn't anything else they can do...
Note that I don't use the VDJ EQs anymore: I'm back on my mixer since I got the XP10s. But back when I had the DJ Console, I didn't have any issues with tweeking the EQs constantly like a madman. It's true that everyonce in a while the waveform scrolling got a bit jumpy (it's never ever fluid anyway).
Maybe EQ use is one starting point, if Dach's problems are experienced by others...
Dach> test case is small indeed, but I figured if ANdre can recreate his memory leak with it, that'd be the best starting point. If not, then we'll have to go further... I just know that a dev team cannot do anything with a non-reproduceable test case. If it's too abstract, e.g. "it just crashes after four hours", they'll try to recreate, like me won't be able to, and they'll blame it on murphy's law and there really isn't anything else they can do...
Note that I don't use the VDJ EQs anymore: I'm back on my mixer since I got the XP10s. But back when I had the DJ Console, I didn't have any issues with tweeking the EQs constantly like a madman. It's true that everyonce in a while the waveform scrolling got a bit jumpy (it's never ever fluid anyway).
Maybe EQ use is one starting point, if Dach's problems are experienced by others...
Posted Mon 25 Oct 04 @ 7:18 am
DJ Andre Landry
Thanx for the info...wil try the test on Thrusday night but like i said ive gone past this and just close the program.
Its nice to see support from this forum!If anyone still is trying to figure out if they should spend the money on VDJ.I think the answer seems quite obvious.
Too bad there are no professional dj's that are using Vdj in my area.Montreal,Quebec,Canada.
I seem to be the only one most dj's here are still doing the cd thing?????Eventually they will figure it out...lol
Andre
Its nice to see support from this forum!If anyone still is trying to figure out if they should spend the money on VDJ.I think the answer seems quite obvious.
Too bad there are no professional dj's that are using Vdj in my area.Montreal,Quebec,Canada.
I seem to be the only one most dj's here are still doing the cd thing?????Eventually they will figure it out...lol
Andre
Posted Mon 25 Oct 04 @ 12:22 pm
DJ Andre Landry
Well DjBambi: I tried your test over the weekend and you are right, I have also come to the conclusion that the problem becomes inconsistent, worked vdj tonight from 8pm to 4am at the club,the halloween party.I ran the s it out of vdj tonight.It played like a charm.I have also spent my time opening task and checking the memory usage.It does come back sometimes in a kind of clean sweep.I had to shut down once at about midnight i loaded a 11min song that was not analyzed....ohoh..i was scratching using web cam..vdj froze and looped i punched in cd player shut vdj down and reopened.
So i guess i wiil let this topic go...i just dont have enough proof for the Dev team to have something to work with.
My computer behaviour hes changed since setting the priority to high,controlling temp,minimum of processes,lot of memory and ram.........
Thanks to all for the advice!
Best Regards
Andre
So i guess i wiil let this topic go...i just dont have enough proof for the Dev team to have something to work with.
My computer behaviour hes changed since setting the priority to high,controlling temp,minimum of processes,lot of memory and ram.........
Thanks to all for the advice!
Best Regards
Andre
Posted Sun 31 Oct 04 @ 10:05 am