I need to know if an individual bad RAM module can be detected out the 8 modules installed. And/or is there a diagnostic test that can be run to determine that? It's on a 2008 Mac Pro, (32 GB RAM) Thank you!
Mac Pro, OS X Yosemite (10.10.4)
Your Mac Pro tower with Xeon Processor uses Error Correcting Code memory, which has 8 additional "syndrome" bits included with each 64 data bits in a word. When read back, Hardware in the Xeon Processor can detect single-bit errors and correct them in one stretched memory cycle, and continue processing without issue.
Double-bit errors can be detected, but most of these cannot be corrected. If an uncorrectable double-bit error occurs, your Mac is designed to halt on a kernel panic, machine check, to avoid poisoning your data.
Corrected errors are eventually tabulated by a background process, and can be displayed as a STATIC report, with error counts for each module.
graphic from anandtech.com
to get fresh data, choose Refresh from its menus or invoke the report again.
It sounds radical, but if you are not getting kernel panics and not seeing errors on this report, you are simply not having memory errors -- not even correctable errors. Running memory tests overnight or all weekend is not necessary. You can go about your normal work and keep checking that report for accumulating errors. Hardware is checking each and every read from memory.
also, the error correction is used during the power-On Self Test, and if a memory error (correctable or not) occurs during those few seconds of testing, the module slot is declared "Empty" and the system will not attempt to use it. The next power-On will test it again, so a module may come and go over time, but it is still Bad.
Your Mac Pro tower with Xeon Processor uses Error Correcting Code memory, which has 8 additional "syndrome" bits included with each 64 data bits in a word. When read back, Hardware in the Xeon Processor can detect single-bit errors and correct them in one stretched memory cycle, and continue processing without issue.
Double-bit errors can be detected, but most of these cannot be corrected. If an uncorrectable double-bit error occurs, your Mac is designed to halt on a kernel panic, machine check, to avoid poisoning your data.
Corrected errors are eventually tabulated by a background process, and can be displayed as a STATIC report, with error counts for each module.
graphic from anandtech.com
to get fresh data, choose Refresh from its menus or invoke the report again.
It sounds radical, but if you are not getting kernel panics and not seeing errors on this report, you are simply not having memory errors -- not even correctable errors. Running memory tests overnight or all weekend is not necessary. You can go about your normal work and keep checking that report for accumulating errors. Hardware is checking each and every read from memory.
also, the error correction is used during the power-On Self Test, and if a memory error (correctable or not) occurs during those few seconds of testing, the module slot is declared "Empty" and the system will not attempt to use it. The next power-On will test it again, so a module may come and go over time, but it is still Bad.
Thank you Grant Bennet-Alder for a very timely response. Glad folks like you are here to assist us!
Can bad RAM be detected?
