Added option to write sensor log directly to disk - what does it mean?

What kind of implications/benefit or under what condition should I on/off this option?
I am using hwinfo logging together with another software that reads the csv log periodically while hwinfo continues to log. Should I enable or disable this option?
 
By default Windows uses file buffering, which is a sort of write-back cache for data written to files/disk. As first the data is kept in memory and then later it's written to disk (delayed writes). This significantly improves performance, but if a system crash occurs, data that were not written to disk yet can be lost.
Enabling this option in HWiNFO will instruct Windows to bypass file buffers and write data directly to disk. This can be especially useful if one needs to diagnose a system crash and there is a need to have latest data (prior to crash) written in the log file. Disadvantage of this is option is slower disk performance (depending on amount of data written).
 
By default Windows uses file buffering, which is a sort of write-back cache for data written to files/disk. As first the data is kept in memory and then later it's written to disk (delayed writes). This significantly improves performance, but if a system crash occurs, data that were not written to disk yet can be lost.
Enabling this option in HWiNFO will instruct Windows to bypass file buffers and write data directly to disk. This can be especially useful if one needs to diagnose a system crash and there is a need to have latest data (prior to crash) written in the log file. Disadvantage of this is option is slower disk performance (depending on amount of data written).
Martin, I'm currently trying to figure out the reason for the PC randomly rebooting. The option to record without buffering is very useful in this case. The only thing missing is the ability to cyclically rewrite the log file. Does the program have such an opportunity? The first time I worked with the program, I did not find it.
Enabling sensor recording in the hope of catching a reboot is not a very effective approach. Because a reboot can happen in an hour or a week. And cyclic recording (from 1 minute, for example) would avoid creating huge log files..
I am using a translator, so I apologize in advance for possible distortion of the meaning.
 
Back
Top