Topic: Memory leak in Replay Gain? (Read 3288 times)

Mirko · « **on:** July 29, 2002, 08:39:10 pm »

Hi,

tonight I was replay gaining about 4500 songs. Better said, I was trying. Because after he did about 2300 I got memory errors (out of memory or similar) and MJ hang. Then I kicked the task out of memory and restartet and MJ processed another couple of hundreds.

I was wondering if anyone has the same problem and if my guess with the memory leak is right.

Besides. 4500 MP3s. Some waves.

Mirko

RemyJ · « **Reply #1 on:** July 29, 2002, 09:47:09 pm »

Mirko,

Check your event log. I had exactly the same problem twice before and I just noticed yesterday that my Ultra/ATA disc controller controller was throwing timeout errors at exactly the instant replay gain analyzation went crazy. It's possible that MJ is not catching or handling IO errors properly.

Remy

Mirko · « **Reply #2 on:** July 29, 2002, 09:55:32 pm »

I will check after work. Thanks.

jmcarney · « **Reply #3 on:** July 29, 2002, 11:26:30 pm »

Mirko,

You're lucky that you got to process 2500. I can only get 100 or so before it starts hemorrhaging memory.

This problem was reported by myself and many others some weeks ago, but Nikolai has not seen fit to acknowledge its existence.

John.

Harry The Hipster · « **Reply #4 on:** July 30, 2002, 04:20:20 am »

Same issue as John. However I've been using mp3gain, which of course adjusts the file itself, and not just the playback settings.
Seems to me I'm better off correlating playback volume within the file rather than relying on the program to do it. Any disadvantages to mp3gain that I'm missing?

HTH

Listening to: 'All the Things You Are' from 'Concord Duo Series, Volume Nine' by 'Bill Charlap' on Media Jukebox

Mirko · « **Reply #5 on:** July 30, 2002, 05:05:57 am »

I have checked the system event log. But the only entry in the questionable time is "Out of virtual memory". About 2:32 after I manually started replay gain.

JimH · « **Reply #6 on:** July 30, 2002, 05:12:29 am »

jmcarney
> This problem was reported by myself and many others some weeks
> ago, but Nikolai has not seen fit to acknowledge its existence.

> John.

John, your language suggest that we owe you an apology. Let me try to explain why we haven't apologized.

First, not all bugs are ours. There are plenty of known bugs in Windows, in drivers, in other applications.

Second, unless we can see it, we can't fix it.

Third, we are ten people with a million things to do. Your priority may not be ours.

Lastly, many reported bugs are not bugs, so, particularly at the end of a development cycle, when we think things are pretty stable, we don't look as carefully as we might.

We will take a look again. In fact, I had asked the guys to do that about an hour ago, before I saw your message.

Nikolay · « **Reply #7 on:** July 30, 2002, 05:54:56 am »

That's right, blame everything on Nikolay.

I skill can't reproduce it. What is your system info? Are you using any special IDE adapters?

Nikolay

KingSparta · « **Reply #8 on:** July 30, 2002, 05:58:39 am »

I Noticed this before, but i think J river Took Care Of This On A Later Build, I Did Not Notice It The Other Day When I Used It.

I Hope Everyone Is Using The Latest Version 8.0.336.

I Think There Was Two People Who Reported This Problem Sometime Ago And They Reported Back That It Was There Video Driver That Was Eating The Memory.

Mirko · « **Reply #9 on:** July 30, 2002, 05:59:18 am »

I can give you mine:

Media Jukebox PLUS 8.0.336
Install Path: C:\Programme\Media Jukebox\

CPU: Intel Pentium 4 2375 MHz MMX
Memory: Total - 523 MB, Free - 246 MB
OS: Microsoft Windows XP Workstation 5.1 (Build 2600)

Internet Explorer: 6.0.2600.0000
ComCtl32.dll: 6.00.2600.0000
Shlwapi.dll: 6.00.2600.0000
Shell32.dll: 6.00.2600.0000
wnaspi32.dll: 4.60 (1021) , ASPI for Win32 (95/NT) DLL, Copyright © 1989-1999 Adaptec, Inc.
Aspi32.sys: 4.60 (1021)

Listening to: 'Don't break the circle' from 'Follow The Blind' by 'Blind Guardian' on Media Jukebox

Nikolay · « **Reply #10 on:** July 30, 2002, 06:05:41 am »

Finally, I can reproduce the bug. We will try to fix it ASAP.

Nikolay

RemyJ · « **Reply #11 on:** July 30, 2002, 06:50:51 am »

Cool. Can you fix my IDE timeouts while you're at it?

Nikolay · « **Reply #12 on:** July 30, 2002, 02:22:22 pm »

Please test this decoder control and let us know if it's any better.

ftp://ftp.jriver.com/pub/downloads/music/custom_mj/decoder.ocx

Download and move it into MJ install directory.

Nikolay

joe|PLS|mama · « **Reply #13 on:** July 30, 2002, 03:49:35 pm »

I have only been having problems analyzing MPC files. Basically, the crash happens after about 10-15 files are analyzed. I installed the new decoder control and began analyzing a list of about 400 MPCs. It crashed after 14 were done. For some reason, it never crashes when I analyze MP3, OGG or APE.

Anyway, here's the error message:

MEDIA JUKEBOX caused an invalid page fault in
module MEDIA JUKEBOX.EXE at 017f:00599c9f.
Registers:
EAX=81a07668 CS=017f EIP=00599c9f EFLGS=00010202
EBX=006c4f58 SS=0187 ESP=0095efbc EBP=0095efe4
ECX=d2976890 DS=0187 ESI=006c4f48 FS=400f
EDX=00000003 ES=0187 EDI=00000001 GS=0000
Bytes at CS:EIP:
8b 07 89 46 0c ff 15 0c f3 5f 00 8b 4d f4 8b c7
Stack dump:
006c4f58 00000040 00000017 0095f028 ffffffff 006c4f48 0095efc0 0095f02c 005fe0c0 ffffffff 0095f038 005bb58b 00000009 0095f028 0000000e 005bb902

Here are my system specs:

Media Jukebox PLUS 8.0.333
Install Path: C:\Program Files\Media Jukebox\

CPU: Intel Pentium III 731 MHz MMX
Memory: Total - 261 MB, Free - 156 MB
Resource Info: System - 56 %, GDI - 71 %, User - 56 %
OS: Microsoft Windows 98 SE

Internet Explorer: 6.0.2600.0000
ComCtl32.dll: 5.50.4704.1100
Shlwapi.dll: 6.00.2600.0000
Shell32.dll: 4.72.3812.600
wnaspi32.dll: 4.60 (1019) , ASPI for Win32 (95/NT) DLL, Copyright © 1989-1998 Adaptec, Inc.
Aspi32.sys: N/A

jmcarney · « **Reply #14 on:** July 30, 2002, 05:42:15 pm »

Jim & Nikolay,

No apology necessary, and as a developer myself I fully understand the problems of developing non-trivial software applications on less than ideal platforms and I certainly don't expect you guys to suddenly drop everything and attend to this particular squeaky wheel

However, my comment was born from frustration. Like many other users, I have a large audio library that I would like to analyse for replay gain and, like many other users, I ran into the memory leak problem while trying to do so. So I duly reported the issue on these forums. So far so good. The frustration comes in because until now, as far as I can tell the response from you guys has been silence. I could accept a "cannot reproduce", but a complete lack of response is just plain frustrating.

Anyway, glad to hear you can reliably reproduce the problem - in my experience, that's 90% of the debugging battle.

You should know that my overall impression of MJ is that it is streets ahead of any other jukebox program I have tried, so any dissatisfaction I express on these boards is relative

Nikolay · « **Reply #15 on:** July 31, 2002, 05:25:27 am »

Did it work?

Nikolay

Nikolay · « **Reply #16 on:** July 31, 2002, 07:12:24 am »

bump.

RemyJ · « **Reply #17 on:** July 31, 2002, 07:24:26 am »

I can't reproduce my driver timeouts with enough reliability to say. I'll continue to test though. Hopefully someone who can more easily reproduce the problem can say.

Remy

Gatobrit · « **Reply #18 on:** July 31, 2002, 07:40:51 am »

Nikolay - Looks like it is now fixed. See below.

I did replay gain analysis on 106 files earlier this morning (running v334) following a visit to Emusic, and I hit the memory leak around 70 files or so. I upgraded to 336 and then updated the decoder.ocx and re-ran the analysis on the same files.

I watched memory usage in Task Manager. With MJ playing in the background memory usage started at around 12Mb. As replay gain got underway it crept up to 15Mb after 15/20 files. After 50 files
memory usage was at 15.6Mb. After 70 files it was 15.8Mb. After 106 files it was 15.8Mb and then it dropped to 13.9Mb after completing.

Interstingly the radio gain and peak level this time through was (sometimes wildly) different from the levels recorded earlier? Is this because of the new decoder? Do I need to reanalyze all my files?

What gives?

Media Jukebox PLUS 8.0.336
Install Path: C:\Program Files\J River\Media Jukebox\

CPU: Intel Pentium 4 1538 MHz MMX
Memory: Total - 261 MB, Free - 107 MB
OS: Microsoft Windows 2000 Workstation 5.0 Service Pack 2 (Build 2195)

Internet Explorer: 5.50.4807.2300
ComCtl32.dll: 5.50.4807.2300
Shlwapi.dll: 5.50.4807.2300
Shell32.dll: 5.00.3502.4718
wnaspi32.dll: 4.70 , ASPI for Win32 (95/NT) DLL, Copyright © 1989-2001 Adaptec, Inc.
Aspi32.sys: 4.70

Listening to: 'X-Ray Hip' from 'Kksf Sampler For Aids Relief 12' by 'Bona Fide' on Media Jukebox

KingSparta · « **Reply #19 on:** July 31, 2002, 08:26:28 am »

Well For Whats It Worth

I Started MJ, Loaded Up About 2800 Files

Then Looked At Memory Used For MJ 106,304K

After 400 Files

Memory Used Is 101,048

When I Hit Stop I Did Noticed I Did Not Get Control Back From MJ After 15 Mins I Closed MJ.

Should I Have Waited Longer?

Windows XP Pro

RemyJ · « **Reply #20 on:** July 31, 2002, 09:30:13 am »

>>When I Hit Stop I Niticed I Did Not Get Control Back From MJ After 15 Mins I Closed MJ

Yeah, I had that happen a few time last night as I was testing. Pressed Stop but none of the other buttons re-enabled and I had to kill MJ from the task manager. Seems to happen about 1 in 4 times.

joe mama · « **Reply #21 on:** July 31, 2002, 11:35:12 am »

Updated MJ to version 8.0.336 and downloaded the new decoder.ocx. Started analying list of 400 MPC files...crashed after completing 14.

MEDIA JUKEBOX caused an invalid page fault in
module MEDIA JUKEBOX.EXE at 017f:00599bcf.
Registers:
EAX=81a0e5e4 CS=017f EIP=00599bcf EFLGS=00010202
EBX=006c4f30 SS=0187 ESP=0095efbc EBP=0095efe4
ECX=d2999760 DS=0187 ESI=006c4f20 FS=400f
EDX=00000003 ES=0187 EDI=00000002 GS=0000
Bytes at CS:EIP:
8b 07 89 46 0c ff 15 0c f3 5f 00 8b 4d f4 8b c7
Stack dump:
006c4f30 00000040 00000017 0095f028 ffffffff 006c4f20 0095efc0 0095f02c 005fdf90 ffffffff 0095f038 005bb4bb 00000009 0095f028 0000000e 005bb832

Media Jukebox PLUS 8.0.336
Install Path: C:\Program Files\Media Jukebox\

CPU: Intel Pentium III 730 MHz MMX
Memory: Total - 261 MB, Free - 154 MB
Resource Info: System - 54 %, GDI - 75 %, User - 54 %
OS: Microsoft Windows 98 SE

Internet Explorer: 6.0.2600.0000
ComCtl32.dll: 5.50.4704.1100
Shlwapi.dll: 6.00.2600.0000
Shell32.dll: 4.72.3812.600
wnaspi32.dll: 4.60 (1019) , ASPI for Win32 (95/NT) DLL, Copyright © 1989-1998 Adaptec, Inc.
Aspi32.sys: N/A

JGourd · « **Reply #22 on:** July 31, 2002, 12:25:27 pm »

With the new OCX the handle count seems constant at 174.

Mirko · « **Reply #23 on:** July 31, 2002, 08:30:37 pm »

Yesterday I installed the new OCX and have successfully replay gained about 1000 mp3s.
It seems that the memory thing is solve, here.

joe mama · « **Reply #24 on:** July 31, 2002, 08:59:28 pm »

Tried analyzing a smaller group of MPCs, only 200. Crashed after 12.

MEDIA JUKEBOX caused an invalid page fault in
module MEDIA JUKEBOX.EXE at 017f:00599bcf.
Registers:
EAX=81a07358 CS=017f EIP=00599bcf EFLGS=00010202
EBX=006c4f30 SS=0187 ESP=0095efbc EBP=0095efe4
ECX=d2a14e30 DS=0187 ESI=006c4f20 FS=402f
EDX=00000003 ES=0187 EDI=00000001 GS=0000
Bytes at CS:EIP:
8b 07 89 46 0c ff 15 0c f3 5f 00 8b 4d f4 8b c7
Stack dump:
006c4f30 00000040 0000001a 0095f028 ffffffff 006c4f20 0095efc0 0095f02c 005fdf90 ffffffff 0095f038 005bb4bb 00000010 0095f028 0000000a 005bb832

Media Jukebox PLUS 8.0.336
Install Path: C:\Program Files\Media Jukebox\

CPU: Intel Pentium III 730 MHz MMX
Memory: Total - 261 MB, Free - 156 MB
Resource Info: System - 58 %, GDI - 75 %, User - 58 %
OS: Microsoft Windows 98 SE

Internet Explorer: 6.0.2600.0000
ComCtl32.dll: 5.50.4704.1100
Shlwapi.dll: 6.00.2600.0000
Shell32.dll: 4.72.3812.600
wnaspi32.dll: 4.60 (1019) , ASPI for Win32 (95/NT) DLL, Copyright © 1989-1998 Adaptec, Inc.
Aspi32.sys: N/A

KingSparta · « **Reply #25 on:** August 01, 2002, 04:57:25 am »

My computer finished all 2,800 mp3s, no lockups (only when I pressed "STOP")

Windows xp did give me a warning I think is was that the virtual memory was low.

when looking the memory the program was using thru the program manager it was lower.

this is using the new ocx

jmcarney · « **Reply #26 on:** August 01, 2002, 06:03:28 am »

So far so good. This morning I analyzed about 70 files. I have had the leak bug on fewer, though it usually doesn't manifest until after analyzing 100|PLS| tracks. Tonight I analyzed about 180 before my wife decided she needed to do her accounts. I'm about to kick off the remaining unanalyzed files in my library (about 1200 tracks) overnight. If that goes without a hitch, I'd be willing to call it fixed

joe mama · « **Reply #27 on:** August 01, 2002, 02:05:18 pm »

Tried analyzing a batch of 20 MPCs, crashed after 6. Never crashes when analyzing MP3, OGG, APE, only MPC.

Rob

jmcarney · « **Reply #28 on:** August 01, 2002, 05:06:34 pm »

Wouldn't you know it, my overnight run failed after analyzing a little over a 100 MP3s.

So, the new decoder.ocx doesn't fix the memory leak bug.

Mirko · « **Reply #29 on:** August 01, 2002, 08:16:50 pm »

Yesterday I replay gained about 500 MP3s. No failure. I guess the new OCX works very good here.

jmcarney · « **Reply #30 on:** August 01, 2002, 08:34:52 pm »

> Yesterday I replay gained about 500 MP3s. No failure. I guess the new OCX works very good here.

Actually, Mirko, it may not be any different. You originally said that you only encountered the problem after analyzing over 2000 tracks. On my system, I get it after 100-200 tracks. My system is much less powerful than yours (PIII 450Mhz, 384Mb RAM), that might make a difference.

Cheers,
John Carney.

shdbcamping · « **Reply #31 on:** August 03, 2002, 12:00:55 am »

I also have a P4 system but have tried to stay away from self tweaking as i have windows XP. I ran my whole 3400 file MP3 library through MJ's replay gain thingy. I had no problem at all. I have RD ram and only 256M of it. Sounds like a Ram and/or Pagefile problem with the systems.
Hope this helps

INTERACT FORUM

Author Topic: Memory leak in Replay Gain? (Read 3288 times)

Harry The Hipster

joe|PLS|mama

joe mama

JGourd

joe mama

joe mama

shdbcamping