Athlon 64 RAM Testing

russreagan

Distinguished
Feb 21, 2005
1
0
18,510
Having an occasional crash or reboot of the system, not running any main programs (few in back-ground). Suspect a hardware problem but having difficulty in tracing. Just built this system on ASUS K8V SE DX, Athlon 64 3400, 2-512 Ultra (ULT 30215) DDR400, ATI Radeon 9550-128, 2-WD 200Gig drives on a RAID 0 Array, XP Home.

I suspect the RAM, but have had little luck with testing. I’ve used Microsoft’s RAM testing download with results of 292,000 errors consistently, which I consider to be invalid. I have also tried MEMTEST86 with similar, large error, results. I’ve torture tested with Pime95 several times at 2 hrs., with no problems and once for 6 hrs. with 1 error. Is there any downloads or RAM test programs that will test this system’s RAM? Any other suggestions welcome.
 

BrentUnitedMem

Distinguished
Oct 8, 2004
693
0
18,980
Sounds like the memory.

Ideally, you want to test your system with a different memory module. If not, try running your system with only one memory module.

Ultra brand has known issues:
<A HREF="http://www.hardwareanalysis.com/content/topic/38126/" target="_new">memory thread</A>

I'd stay away from OEM brands, or third party "major" branded modules. Buy original modules, or buy good name brands like Kingston, Corsair, Crucial.. etc..

____________________________________________________
"Memory with lifetime warranty? So, whose lifetime is that?"
<A HREF="http://www.brentcrowley.com/" target="_new">homepage</A>
 

skipshot4

Distinguished
Dec 9, 2004
17
0
18,510
I have a similar problem. I've tested with MEMTEST86 with just one DIMM and everything is peachy. In fact I tested each DIMM and DIMM Slot individually and I get no errors. The problems comes up when I put both DIMMS into any of the three DIMM slots. In this situation I get a ton of errors on Test #5 (Block move, 64 moves) and #7 (Random number sequence). Any ideas as to why I'm getting these and how to solve the issue?

AMD Athlon 64 2800+, SAPPHIRE ATI RADEON 9800 PRO, Corsair Value Select 184 Pin 512MBx2 DDR PC-3200, Hitachi 80GB 7200RPM SATA HD, EPoX "EP-8KDA3J" nForce3-250Gb Motherboard, Windows XP HOME w/ SP2
 
G

Guest

Guest
Wusy or Brent could answer this, but I thought I read somewhere the nforce chipsets were picky about running value select DIMMS in dual support mode. Not positive though :frown:

<font color=green> Woohoo!! I am officially a <b> Member </b>!! </font color=green>
 

BrentUnitedMem

Distinguished
Oct 8, 2004
693
0
18,980
Unfortunatlely, with memory it's probably more true than with other brands:

"You pay for what you get"

There is a reason as to why certain memory brands are so cheap. When we compare the diffeence between two kinds of chips or modules, the biggest difference is the number of test parameters the chips or modules have to pass to be ranked in a particular class. UTT, for example, is normally tested with maybe a few dozen parameters, whereas major chips pass with many more test parameters. However, some UTT chips tested with more parameters can be more valuable than major chips (Windbond UTT -BH5).

General rule: The fewer the test parameters, the lower the cost, the poorer the quality overall.

Samsung and many retail compaies set very high standard for their memory tests, and so, they produce good quality memory. And of course there are many other factors such as how much money is invested into equipment and high tech lazers.. etc. Since Samsung owns close to 30% of the global DRAM market, and they are the second largest investor of RAMBUS, they can afford the best equippment and clean rooms in developing semiconductor chips; as a result, their chips are the most sough after chips in the market today. (Corsair, and Kingston for example do not make their own chips)

Companies will sell the memory that passes more standards at a higher price. For the ones that do not pass, perhaps they are sold as value RAM or as other cheap memory solutions. Of course this category of memory has more problems; problems with the memory controller, for example.

Most likely the reason you see a problem with two modules as opposed to one has something to do with the following:

When we are talking about data integrity, many things come into play. The integrity of a single module is greater than the integrity of two modules. The reason is that data traveling between chips on a single PCP is more reliable/efficient than data traveling between two sets of chips on two different PCBs. Also the addition of more physical connections decreases data integrity too (ie.. physical pin connections). And also two modules require more voltage too, and thus introduce more noise on the data lines. Without sufficient voltage, you may not even be able to use two modules together.

____________________________________________________
"Memory with lifetime warranty? So, whose lifetime is that?"
<A HREF="http://www.brentcrowley.com/" target="_new">homepage</A>