View Issue Details
|ID||Project||Category||View Status||Date Submitted||Last Update|
|0006250||MediaMonkey (current)||Codec||public||2010-01-14 17:24||2011-10-09 02:06|
|Priority||urgent||Severity||minor||Reproducibility||have not tried|
|Target Version||4.0||Fixed in Version||4.0|
|Summary||0006250: Recompile latest version of lame|
|Description||Because of intentional defects in earlier intel compilers, the release of MM 4.0 should include a recompiled version of lame using fixed intel compiler. At the moment, lame is at 3.98.2|
For details see:
|Tags||No tags attached.|
|Fixed in build||1434|
||Re-assigning to Michal, since he's gonna work on similar topics.|
||Fixed in build 1332. Lame upgraded to version 3.98.4.|
It seems that the version of lame included with MM 4.0 is still problematic on some machines/environments.
Also, there's another debug log of failed volume analysis (replay gain) posted to:
Edit: According to the user, the replay gain errors occur independent of what version of LAME is used (but MM3 doesn't generate such errors).
Note: the user experiencing these problems is also experiencing problems with Volume analysis. More details:
HW: cpu is a 32bit amd athlon xp2400 , 2ghz singlecore, (only sse1 capable)
The sample files that trigger the error in his environment, work fine for me. What's strange is that if the same file is moved to a different location, the scan proceeds without a problem.
Comment from the user: to me it seems that the file is still locked and another thread is scheduled.(analyze is complete,db is updated,tag is not updated).
Note: file monitor/auto-organize aren't enabled
As discussed offline, we don't use the optimal configuration for the compiler. Will be modified and hopefully will fix the issue.
Options will be based on: http://software.intel.com/en-us/articles/performance-tools-for-software-developers-intel-compiler-options-for-sse-generation-and-processor-specific-optimizations/
Based on some speed testing we decided to include version optimized for SSSE3 instruction set. The resulting version is not only compatible with any Pentium compatible processor, but is also around 2% faster that the Lame in MM3 and about 4% faster than the Lame in previous MM4 builds.
||Fixed in build 1434.|