- May 19, 2011
- 19,954
- 14,261
- 136
I've just diagnosed a RAM problem (to my current satisfaction at least) in a customer's computer.
My impression was that with S3 sleep mode that everything in the system goes off except what's necessary to maintain the contents of RAM, and basically the RAM is left running in exactly the same way as if the system is entirely running in terms of electricity usage. Is this correct?
The situation:
ASRock H81M-DGS R2.0 P1.70 board, Haswell gen CPU, two memory slots, two identical memory modules (Teamgroup 2x4GB DDR3-1600). The customer was allegedly told by the builder that their computer wasn't compatible with S3 sleep mode. The board of course says it is. The symptom was that every time the computer came out of sleep mode, it either BSOD'd or otherwise failed (e.g. black screen, no system response) and the BSOD was almost invariably a "this is probably RAM" type like MEMORY_MANAGEMENT, CACHE_MANAGER etc.
Yesterday I swapped out the two original modules for one I had spare, and the system resumed absolutely fine. I then swapped that module out for one of the originals, still OK resuming from sleep mode, then I swapped out that module for the remaining original module. I'm fairly sure at that point the machine BSOD'd on startup, but then it acted fine including resuming from sleep mode. Huh.
From previous experience I'm aware that it's one thing to do a sleep transition and wake the computer up within seconds, it's another thing for a computer to last say thirty minutes or longer in sleep mode (and from that experience I found a faulty PSU).
Today I restarted testing the modules this time starting with the known-good spare module of mine, sleeping for 10 minutes, wake, then 20, wake, then 30, wake. No problems.
I swapped out that module for one of the original modules and repeated the procedure, but it failed at the 10 minute mark.
I swapped that module out for the other original module, repeated procedure, no problems throughout.
I swapped the suspect original module back in, failed again at the ten minute mark.
I used the same PSU and the same module slot for testing throughout.
Admittedly the customer isn't particularly concerned about sleep mode (especially given that I've just upgraded the system to boot from SSD), but if my understanding of how sleep mode is basically correct then despite the 14 passes with memtest86, this module still has some fault that could resurface in some other scenario surely?
My impression was that with S3 sleep mode that everything in the system goes off except what's necessary to maintain the contents of RAM, and basically the RAM is left running in exactly the same way as if the system is entirely running in terms of electricity usage. Is this correct?
The situation:
ASRock H81M-DGS R2.0 P1.70 board, Haswell gen CPU, two memory slots, two identical memory modules (Teamgroup 2x4GB DDR3-1600). The customer was allegedly told by the builder that their computer wasn't compatible with S3 sleep mode. The board of course says it is. The symptom was that every time the computer came out of sleep mode, it either BSOD'd or otherwise failed (e.g. black screen, no system response) and the BSOD was almost invariably a "this is probably RAM" type like MEMORY_MANAGEMENT, CACHE_MANAGER etc.
Yesterday I swapped out the two original modules for one I had spare, and the system resumed absolutely fine. I then swapped that module out for one of the originals, still OK resuming from sleep mode, then I swapped out that module for the remaining original module. I'm fairly sure at that point the machine BSOD'd on startup, but then it acted fine including resuming from sleep mode. Huh.
From previous experience I'm aware that it's one thing to do a sleep transition and wake the computer up within seconds, it's another thing for a computer to last say thirty minutes or longer in sleep mode (and from that experience I found a faulty PSU).
Today I restarted testing the modules this time starting with the known-good spare module of mine, sleeping for 10 minutes, wake, then 20, wake, then 30, wake. No problems.
I swapped out that module for one of the original modules and repeated the procedure, but it failed at the 10 minute mark.
I swapped that module out for the other original module, repeated procedure, no problems throughout.
I swapped the suspect original module back in, failed again at the ten minute mark.
I used the same PSU and the same module slot for testing throughout.
Admittedly the customer isn't particularly concerned about sleep mode (especially given that I've just upgraded the system to boot from SSD), but if my understanding of how sleep mode is basically correct then despite the 14 passes with memtest86, this module still has some fault that could resurface in some other scenario surely?
Facebook
Twitter