hundreds of "decoder.exe"

[beatsystem]

Hi,

Since the change in scanning programming, I'm continually left with hundreds of "decoder.exe" processes sitting on my machine taking up space (sometimes 1GB or more). If Similarity crashes (as it often does), they're stuck there - I have to use a taskkill command to get rid of them. Any idea why they're not closing automatically during the program's normal run?

Also when I restart, and scan using the cache, I get yet more decoder.exe files! Surely if Similarity is using the cache, it shouldn't be decoding anything?

This is all on premium version for audio only, I don't use it for pics.

Thanks,

Beats

[beatsystem]

An FYI update on the above, it seemed to be caused by Quicktime Decoder - disabling that in the settings stops the flood of "decoders".

[Admin]

Id "decoder.exe" crashed internally or hang up, Similarity tries to restart it and kill previous one, it not always succeed if example you have configured crashes to uploade to Microsoft sites. Do you have crashes on main process "Similarity.exe" ? or only on decoders ?

[beatsystem]

Well, like I said, deselecting Quicktime decoder solved it.

A quick look in the log shows tons of things like this:

2013-06-11 00:00:19   Decoder hang up on file: U:\MP3\Ripped CDs\Glass, Philip\Philip Glass - 1973-5 - CD.jpg, decoder: Apple QuickTime Framework
2013-06-11 00:00:19   System error [2]: The system cannot find the file specified. 
2013-06-11 00:00:19   CProcessLink::InnerSend:174 failed [-100]

Note that I DO NOT have image comparison selected in Settings, but this explains the decoder problem, and also explains why I see decoders when running from the cache, as these files haven't been "cached", so the program will try to decode them every time and fail!!!

Despite the above I'm still having problems getting a complete scan with the latest version.I get continual crashes which seem to be out of memory. I have 4GB on a 32-bit machine, but some error causes the Similarity.exe to swell to 1.7+ GB.

Now I've solved the decoder problem, I'll start to see what can be done to solve the other problem.

[Admin]

You have encounter "out of memory" problem, 32bit Windows only gives 2Gb of memory to program, 1.7Gb shows you have near this limit (Windows doesn't show shared memory for dll as program use, but they consumes memory).
1) You can use 64bit version, but you need 64-bit version of Windows
2) You can increase with some "hacks" memory limit for Similarity to 3Gb by editing boot.ini
3) You can disable unnecesery algorithms (disable old "content" based algorithm, use only precise) this gives 2 times more memory for other alogrithm.

[hsei]

To my experience 32-bit-similarity becomes unstable with 60K to 90K files.
This seems to have been solved in the 64-bit version.
If you can't move to 64 bit, you can split your search e.g. into A-L and M-Z, This misses duplicates between the two groups, but most duplicates in the tagging scheme "artist - title" are within the same starting letter. Files of type "unknown artist - unknown title" can be checked as separate group against every of the "letter groups".