Consistent failures in a specific address range - even if you swap/rotate modules

[trjp]

System: 2nd gen Core i3 laptop running DDR3-10800(1333) memory - 2x4Gb (original chips so manufacturer approved)

Issue: PC freezes/crashes regularly (usually within 5 mins of power-on)

I've run a variety of Memtests (using 4.3.7) and it throws errors inside 5 mins across most tests - addresses are always between 1151Mb and 2369Mb

I've swapped the SODIMMs - same range/errors
I tried each SODIMM in each socket 'solo' - same range/errors
I tried the 2x4Gb SODIMMs in a different laptop - no errors
I tried a spare 1Gb SODIMM 'solo' - no errors

I'm assuming all that means the modules are AOK and there's a fault in the laptop itself!?

Note: BIOS doesn't offer anything like Memory Voltage/Latency setup and the system cooling is working as best I can tell but the modules are toasty when I'm swapping them around...

TIA

[David (PassMark)]

Can you post the details of the errors. In particular it would be good to see the number of bits in error for each error.

Yes, it would seem unlikely that both sticks would be faulty in an identical fashion.

Real test to confirm it was bad CPU or motherboard would be to replace the RAM with at least 2GB of known good RAM (if you have any spare).

[trjp]

I've attached 2 screen pics I took during testing

What's interesting is that one is on the 2nd pass and has 5 errors wheras the later test is on the 1st pass and has a lot of errors!
I'm starting to think heat might be an issue here - it would explain why a later test has more errors?

I don't have a thermal camera but my IR thermometer reckons the chips/chipset (which sits right next to the chips) is running at 60+C - in a room which is probably 18Cish - that's quite warm and it gets that warm sitting in the BIOS screen - you don't need to run MEMTEST to warm it up!

At this point I'm assuming the original SODIMMS are 'known good' simply because they test OK (and work OK) in another laptop - all I can dig-up is another 512K chip (which would put me inside the "failure range" with the 1Gb) so I'll try that and see if it fails (and if it does - I think that's game over for the laptop)??

[trjp]

Trying again with the attachments!

[David (PassMark)]

Screen shot show 1 bit in error. Which makes it more likely to be a RAM problem,
CPU or motherboard problems tend to have a larger impact.

[trjp]

I continued to test stuff today - mostly for academic purposes at this point.

What I've noticed is the longer you run the test - the faster errors start to pile-up - so I put the original SODIMMs back into my own laptop and ran MEMTEST for a few hours
The first error appeared around 2h15m and after 3 hours a few more - the test then froze-solid around 4 hours (I guess the faulty RAM affected MEMTEST?)

I reckon the original laptop has cooked it's RAM to the point it no-longer tolerates heat - the design of this laptop puts the chips RIGHT next to the CPU with, effectively, zero cooling so it fails fast - it takes longer in mine (which being a ThinkPad X Series is a bit more solidly put together than this consumer Z series)





Note the plastics around the ram socket are 'rainbow colored' as-if from heat - that CPU 'lock' runs around 65C - the heatsink around 62C - the heatpipe near the fan is only 28C and the air leaving the PC isn't that hot - e.g. it's not a great cooling system (the air coming from my X Series would warm your lunch!)

After all this I've decided to order a pair of 2Gb chips - they're fairly cheap - we'll see what happens (and they're warranted so if they fail, they fail!)

Thanks for the help!!

[trjp]

Update:

2x2Gb SODIMMs arrived today - tested each one individually and then the pair - all for 2 full passes - zero errors encountered!!

Thanks for the feedback - I'd almost certainly have binned the laptop as faulty on the basis that 2 memory chips failing at the same time seemed unlikely - but I guess they're both damaged in some way.

It is still possible the problem is in the laptop and only affects larger memory chips (the originals were 4Gb each) but we'll never know as those chips are too pricey at this point

It's also possible the laptop killed the last SODIMMs and will kill these too - these were stupidly cheap tho so we can just keep replacing em!

Thanks again!