[Resolved] 2gb+ L2 Cache Corruption?
Posted: Tue Oct 28, 2014 1:07 am
I have just recently started using L2 Cache to augment the overall caching routine. Currently I have 2 drives under a single cache being a 4gb Boot Drive (Compact Flash over Sata) and a 320gb storage drive over Sata. The L2 Cache consists of a 4gb Physical Ramdisk with known working ram over the Sata bus.
During testing I have noticed that once the L2 cache fills up to at close to 2gb, I will start getting phantom errors during the windows session. These issues would be 'corrupt' and 'unrecoverable file' errors that would be exposed as a long list upon a chkdsk during testing. If I were to restart, I would then end up with a BSOD of 'process1_intialization_failed' upon startup. The only way to remedy this would be to either disconnect the L2 Cache medium or disable L2 if I can before shutdown which defeats the purpose. Furthermore, if I continue with the session trying to access the parts of the drive now producing fake file errors and then commit the writes to that drive afterward those files and directories will, in fact, become corrupt in actuality. Otherwise if I do not continue and remove the l2 cache and restart, the file system/corruption errors that were otherwise exposed as a generously long list no longer exist and the scan comes up without a single error.
To further add on, if the L2 Cache is around oh.. 1.75gb or so filled and I issue a restart - the next session will instead issue a boatload of memory access errors and 'insufficient memory' allocation errors despite having 6gb+ free. These errors will happen while accessing most applications during the resulting session thereafter.
The issue never occurs long as I keep the L2 cache at around 1gb or so maximum. The OS the L2 Cache is being used on is Windows Server 2003 Enterprise x86. So my question is if this is a known issue under a 32bit system? Is it because L2 handles itself as a separate form of memory allocation addressing and thus pushing memory boundaries under a 32-bit system?
Full system specs can be found in the following text file: http://rwinds.no-ip.org/~archive/SpecList_Scarab.txt
During testing I have noticed that once the L2 cache fills up to at close to 2gb, I will start getting phantom errors during the windows session. These issues would be 'corrupt' and 'unrecoverable file' errors that would be exposed as a long list upon a chkdsk during testing. If I were to restart, I would then end up with a BSOD of 'process1_intialization_failed' upon startup. The only way to remedy this would be to either disconnect the L2 Cache medium or disable L2 if I can before shutdown which defeats the purpose. Furthermore, if I continue with the session trying to access the parts of the drive now producing fake file errors and then commit the writes to that drive afterward those files and directories will, in fact, become corrupt in actuality. Otherwise if I do not continue and remove the l2 cache and restart, the file system/corruption errors that were otherwise exposed as a generously long list no longer exist and the scan comes up without a single error.
To further add on, if the L2 Cache is around oh.. 1.75gb or so filled and I issue a restart - the next session will instead issue a boatload of memory access errors and 'insufficient memory' allocation errors despite having 6gb+ free. These errors will happen while accessing most applications during the resulting session thereafter.
The issue never occurs long as I keep the L2 cache at around 1gb or so maximum. The OS the L2 Cache is being used on is Windows Server 2003 Enterprise x86. So my question is if this is a known issue under a 32bit system? Is it because L2 handles itself as a separate form of memory allocation addressing and thus pushing memory boundaries under a 32-bit system?
Full system specs can be found in the following text file: http://rwinds.no-ip.org/~archive/SpecList_Scarab.txt