Does this mean my ram is no good?

[OILFIELDTRASH]

Never had any issues with ram. Looks like I failed memtest? Is that what I'm seeing on screenshot? Thanks for any help.

 

[Emulex]

i would move the ram around or take out one stick at a time. but yeah that is bad.

bad motherboard
bad cpu
bad ram
bad bios settings
bad psu

take your pick if it repeats the same error in the same spot/pattern then yeah bad stick - if its intermittent (different pattern/location) probably bad ram but investigate other items

 

[RebateMonger]

Yeah, something's wrong. You should be able to past Memtest86+ with zero errors with multiple passes.

Occasionally, memory will test "bad" but the user won't see any symptoms. Years ago, I had a system that seemed to be OK. But when I tried to upgrade it to XP, the hard drive got scrambled. Later, I sold the RAM to somebody who tried to install XP on his computer and couldn't. He ran Memtest and found a bad memory module.

 

[OILFIELDTRASH]

Well it kept showing 4gb ram only 2 gb usable in system. So I ran memtest and sure enough it's bad. I just picked some more corsair dominator ram from fry's. 6gb for $189 today only. I'm through with cheap ram from here on out. It's ocz ram btw. I'm only buying corsair dominator from now on. Got 6 gb's in my computer and it's worked flawlessly and gonna put the other 6 in my wife's.

 

[SimMike2]

Sometimes if you let the motherboard set the timings automatically with the SPD chip on the ram, it might set it for too much performance. Tweaking your BIOS settings slow down the ram might help, for instance with DDR21066, try running at 800 or even 667. This is in case you can't return it for warranty replacement. I've been able to salvage borderline ram in the past following this strategy.

 

[konakona]

on a related note, what would you suspect with the system being unstable at the exact rated timings / voltage for the ram, but stability improves a whole lot (while still not 100&#37 at lower speed / looser timings? No memtest errors at that either.. would the mobo be next to blame?

 

[Emulex]

its too complex. what if your video card is doing turbo-cache and using ram? that might induce extra stress. putting a pci video card in place and seeing if the problem goes away might isolate that problem - but then would you point fingers at video card or mobo?

[why i buy prebuild machines, preferably with ecc] - the amount of time and frustration (weekends) far outweighed any savings by DIY'ing it even at a measly $25/hr

 

[OILFIELDTRASH]

Put the corsair ram 6gb and it passed a few passes of memtest no error. It was that cheap ram after all. I might try looser timings later but for now I'm using the dominator ram. Thanks

 

[jvroig]

I'm through with cheap ram from here on out

To be fair, getting a dud in electronic gadgets is inevitable. What you experienced might happen to you again if you keep buying cheap RAM all the time, and it might also happen again even if you keep buying high-end and very expensive RAM.

My (anecdotal) experience, for example, is that just sticking with the cheapest RAM available has not netted me any lemons so far. In fact, the lemon I encountered was from a rather expensive stick of DDR SODIMM (I had no other options as DDR SODIMM was hard to find already, as it was already being phased out by DDR2), which resulted in errors in memtest scrolling through the screen.

I'm not saying I should have just looked for cheap RAM as they are more reliable. I'm saying most RAM, regardless of price, are practically equally reliable, and eventually you'll get a few duds no matter your buying preference. It is the nature of electronic gadgets.

Another anecdotal experience had me using the cheapest DDR 667 RAM I could find for my build 3 years ago, 2 x 1 GB Kingston sticks. Never bothered to overclock it, until recently when I swapped parts to a lesser rig after a huge upgrade. Although it was not rated for 800, it would go all the way to 790+ MHz without touching the default voltage and not lowering the timings, on a cheap MSI board.

 

[Ertaz]

That CAS7 timing is pretty tight. Does the sticker on the ram confirm it's rated for that?

 

[Idontcare]

I'm through with cheap ram from here on out.

Once you get above the true gray-market variety of questionably too-cheap-to-be-true dimms the fail rate of just about any dimms are going to be roughly equivalent across the vendors. What you pay for at that point is warranty and rma service.

I had top of the line mushkin redlines and they went tits-up on me. Under warranty so they got replaced no issues, but still at the time I could not have spent more money for the ram or quality (the micron D9's).

Nowadays I just get the AData or Supertalent stuff and call it good. Just had a 2GB stick of Adata die on me, oh well, cheap as dirt.

 

[dbcooper1]

Make sure the board is setting the voltage to what the OCZ RAM is rated at that speed; I've had boards not set it right and that can cause a problem like that too.

 

[Emulex]

the definition of cheap has changed.

to me cheap ram is anything rated at default DDR3 oem voltage levels. and anything that fails when run at such voltage levels for the rated SPD speed.

Keep in mind there is cheap ram, and cheap boards as well. you can use a 2 or 4 (or 8?) layer board to route your wiring and there are noise considerations for UDIMM.

RDIMM are cool because they provide total ECC (both address and data) so its easier to spot/fix - the buffer also takes a huge load off the bus - but it ain't cheap. $100+ per 4gb stick * times 6 for some server ram is insane.

 

[Rubycon]

RDIMM are cool because they provide total ECC (both address and data) so its easier to spot/fix - the buffer also takes a huge load off the bus - but it ain't cheap. $100+ per 4gb stick * times 6 for some server ram is insane.

Where are you finding registered DDR3 for $100 / 4GB stick? I'm sure my broker whose looking for 500+ pieces would be interested.

Oh and yes to the OP make sure BIOS timings are set to manual and match manufacture's rating and ram is not overclocked. Also ensure voltage is correct then re-run the test(s). If it fails pull it and try again with another piece.