Slashdot Mirror
Toshiba Recalls Notebook RAM
TheSync writes "The Register is reporting that Toshiba is recalling notebook RAM blaiming third-party DIMMs. This follows on HP's bad notebook RAM in June. Which raises the question, is there a vast sea of bad DIMMs out there?"
Yes, and it's found using Pricewatch, Google, and other product search engines. Not all cheap RAM is bad, but you're a lot more likely to get something crappy if you go with the lowest bidder. Those prices are low for a lot of reasons, including support, warranty, and quality. I've bought my share of low-priced memory through Pricewatch, and I've also had to return several of them. Never buy memory that doesn't have a lifetime warranty.
I'm sure Toshiba and Dell didn't buy their memory through Pricewatch (that'd be a hell of an order) but they probably sacrifice in the same way to get their internal costs down. Note that you'll pay a nice premium for ordering memory upgrades through the notebook manufacturer.
Except for a bad 32mb SIMM I had in 1997, I've never ever had ram go bad on me, but then again I always use Crucial - I've seen some of the prices for ram you can find on Pricewatch and all, but remember 'if it sounds to good to be a deal, it probably isnt'.
Besides, with everything else then can go wrong with PCs these days, I like to be reasonably assured my ram is fine.
Looking for hardware (Currently need: Large Etch-a-Sketch) Have one? See my journal!
I've had RAM which could pass all day long on a so-called memory tester, put it into a PC and the thing couldn't even finish POST.
I used to use gcc linux kernel compile to thrash-test memory - start enough of 'em so it just starts to swap and let it run in a loop overnight. If no signal-11's in the morning it'll probably survive anything else.
try { do() || do_not(); } catch (JediException err) { yoda(err); }
An easily repeatable, predictable applicatoin behavoir problem like that doesn't sound much like a RAM problem. RAM issues tend to give you more sporadic errors, either memmory errors themselves or wierd bugs caused by the wrong values being pulled from memory for all sorts of things.
... seems to me like there's better explanations.
You sure its not a conflict between winamp and some resorce on your computer? Maybe it doesn't like your sound card drivers, or the visualizations engine hicups with your graphics card driver, or its expecting a different version of some library, or i dunno
I have built a number of desktops in the past few years for myself and others, and have returned a lot of RAM. Premium RAM seems to have lower failure rates, but I have returned some damn expensive RAM too. I never had problems with older RAM. Are the quality control issues different now?
DDR is a very high speed multipoint parallel interface with very little tolerance in the drivers and recievers. Designing DDR motherboards and DIMM modules is difficult. The capacitive load of the DDR bus varies depending on how many DIMMs are loaded and the DIMM architeture (#chips on DIMM). DDR drivers dont vary the buffer strength based on loading so you will have too much overshoot with one DIMM with 5 chips (x16) and too overdamped with 4 DIMMS with 36 chips (stacked x4) on each DIMM. This is why most motherboards are more relable when all DDR slots are full.
Motherboard manufactures must qualify each DIMM combination separately. You should always use the DIMM modules recommended by the motherboard manufacturer. This is a problem that will only get worse.
I learned something interesting just yesterday about making DIMMs. There are companies out there that specialize in recovering failed DRAM chips. They buy them as factory rejects for pennies, and use some trickery to mask off the bad bits and re-use the recovered DRAM as a smaller density. I know Micron buys lots of this stuff for their value line. Maybe the patching isn't as good as we'd hope?
Eh, blame it on alpha particles. Those buggers are causing all sorts of problems with bit-flips in memory cells. Buy ECC!
RAMBUS is a point to point buffered interface. Each signal only sees one load with dedicated TX and RX signals. The interface is actually a daisy-chained loop, rather than multipoint bus. That is why you need you need the blank DIMM to pass the TX to the RX signals. Since the loading is always the same, it's much more consistent and you will se less vendor to vendor variance.
Bad ram like that is used in things non-crucial applications like digital answering machines, where the memory error will just cause a harmless blip during playback. But who knows what "big brand name memory company" had in mind.