Re: Memory errors on new memory in new system
On 11/27/2012 11:29 PM, Marc Shapiro wrote:
> Last weekend I put together a new box to replace the one that has been
> locking up repeatedly. The components are:
> Gigabyte 970A-DS3 MB
> 8GB (2x4GB) Patriot G3 RAM
> AMD FX-4100 Quad Core CPU
> ASUS DVD/CD Writer
> MSI R5450 Graphics Card
> Seagate 1TB HD
> I don't know what else to check, or what hardware is actually bad. Any and
> all suggestions cheerfully accepted.
You didn't state the DIMM speed. Are these 1333 or 1600 sticks? What
is SPD "auto" running them at? If they're 1600 modules try running them
at 1333 with appropriate timings and modifying the CPU clock multiplier
Or you may want to try backing the timings off a bit. If your Patriot
modules are DDR3-1333 with 10.5-7-7-7 timings try 12-8-8-8.
I had a Biostar socket a A mobo with the dual channel DDR400 memory
controller in the nVidia Northbridge and two sticks of Geil DDR400 that
routinely threw memory errors until I backed the timings off SPD by 1
clock. Typically these memory errors are caused by a few "marginal"
transistors in some cells that simply won't function correctly at SPD
timings. I won't go as far as saying this is "common" with consumer
DIMMs, but it's far more frequent than with server modules.
BTW, "Patriot" may not be generic but it's not a tier one player like
Crucial, Hynix, Kingston, Micron, Samsung, Wintec, etc. Over the long
haul companies like Patriot, Geil, G.Skill, etc, tend to have higher in
the field defects and failures than the top players. I try to stick
with DIMMs from Crucial and Micron as the entire product from silicon to
final module is manufactured under "one roof". These companies have
been making DRAM chips and modules decades longer than the new kids and
really know the business inside and out. Performance, quality, and QC
are very high, and warranty service excellent. In summary IMHO they
simply have better overall product.