Re: Anyone using Foxconn motherboards

Kp wrote:
I feel it is time to try and she some light on these memory issues.
Most memory issues are either timing or chip compatibility related if
there is no physical fault.

Yes, but we're talking about physical faults here. The vast majority of the faults I've seen with generic memory were cell faults - ie. one or two memory locations repeatedly produced the wrong bit in certain tests.

Then you have the cheaper form of memory know as OEMs
(this is normally product that is is built buy a smaller memory house
and can be their chips or someone else's on any pcb) because this has
the greatest variable in design it also has the greatest chance of
being problematic even though there may be no physical flaw with the D
RAM or the PCB.

This might be true, although I'd be surprised; this sort of memory normally has very conservative SPD timings. However, cell faults are extremely common, and are virtually unknown with more expensive memory. I've *never* had a cell fault on a memory stick with chips marked from a major manufacturer (Micron, Samsung, Hynix, Infineon etc).

I've seen quite a few generic sticks with PCB faults too, but that's less of an issue because people tend to notice when their PC doesn't boot.

We also batch test our memory
with real hardware DIMM checkers these are full on hardware test beds
that cost £1000s and are often fare more reliable than the software
clones that are available free to down load from the web

A couple of years ago, back when I was daft enough to buy generic sticks from Novatech, I bought two DDR sticks (marked Matrix). One had a PCB fault and the other turned up a cell fault after 30 mins. I took these back to the returns counter and they tested both sticks with the hardware DIMM checker. According to them, the cell-fault stick was fine.

Three choices here:
1. The guy at the returns counter lied.
2. Hardware DIMM checkers need more time to check a stick.
3. Hardware DIMM checkers do not work.

Same thing happened when one of the replacement sticks also turned out to have a cell fault.

I am
sure you can imagine being the size that we are if 1000+ sticks
(average batch size) contaminated our production stocks there would be
out cry and we would probably end up on watchdog as some of out down
defunct competitors often did.

In my experience, most people simply don't link an occasional extra crash or bizarre software behaviour with their memory upgrade. Unless people run memtest, they'll never know.

For all I know, Novatech may have cleaned up its act in the last year, or the untested memory business may have collapsed. However, if you're going to deny that there is or has ever been a cell fault problem with generic sticks, I see no reason to trust you or anyone else selling unbranded memory.


--
John Jordan
.

Relevant Pages

  • Panic: Kernel page fault with ath0_com_lock held, r211295
    ... Kernel page fault with the following non-sleepable locks held: ... Physical memory: 2031 MB ... Loaded symbols for /boot/kernel/tmpfs.ko ... data packets ...
    (freebsd-current)
  • Re: Binary data in multiple adjacent files
    ... Mapping it doesn't eat any working set except, possibly, for the ... It's not faulted into physical memory. ... You also get a whole "page fault cluster" faulted in. ... grid sections immediately north, south, northeast, northwest, southeast ...
    (comp.os.vms)
  • Re: [SUSPECTED SPAM] Re: [linux-pm] Proposal for a new algorithm for reading & writing a hib
    ... fault allowed to proceed. ... page faulting code touches memory, ... will know to do just one thing, write the memory to disk. ... Linux loads the suspend module to memory (if it is inside kernel ...
    (Linux-Kernel)
  • Re: Applications "buying" resources
    ... L4's approach for virtual memory is similar to yours: ... While the OS can field the fault, ... resources are disposable into the same entity which manages the ... you also need a way of letting the kernel ...
    (comp.arch.embedded)
  • Re: Anyone using Foxconn motherboards
    ... your memory suffered from a cell fault may I ask what you were using to ... RAM or the PCB. ... Hardware DIMM checkers need more time to check a stick. ...
    (uk.comp.homebuilt)