I warn you and anyone who is listening , YOU Can Not trust the INTEL XMP Proflies, you can use them as a guideline ,but make your own settings .

These are the highest stable rates I have gotten with the current setup:

Bump your mem volts to 1.6 and see if that improves stability.

And Bob's right.  The XMP is useless.  I tried both settings and both of them crashed.

I've only run one render on this system.  It was done with Poser Pro 2010, because it has 64 bit capability.  It was made up of 460 different renders at 1920x1080 and took almost 6 hours to finish.  I kept RealTemp up the whole time and the processor stayed in the high 70's without ever hitting the 80c mark.  I really don't want to go over 80c. When I get the new RAM, I'll try the same render again with default settings and whatever the highest stable overclock is, determined by the Extreme Tunning Utility in automatic mode.  I'll always use the default settings of the bios unless I'm doing something that can use extra power.  When I run the two different renders, I'll decide if overclocking is useful in times of rendering and if the temp goes too high. I think using the default settings most of the time will extend the life of the system, and with the profiles that the Extreme Tunning Utility lets you save, it's easy to switch.

Don't forget that Intel designed the XMP data extension to the SPD data specifications in memory modules.  The data that is stored in an XMP profile is what the memory manufacture states are the correct settings for a given speed for that memory module (their product), and what the memory setting will be set to if an XMP profile is selected in the BIOS.

I've used XMP profiles in memory on two different PC platforms, and they have worked fine for me, with few if any issues.  In my i7-930 CPU PC, which has a specified 800/1066 MHz memory speed capability, I have it's memory's XMP profile enabled, and it runs at 1600 MHz, 9 9 9 24, at 1.5V.  That is not on the mother board you are discussing in this thread.  That XMP profile also has my QPI link running at 6.4 GT/s.  The standard SPD profile for these modules run at the standard 1066 MHz speed, which is used if the memory speed is set to Auto in the BIOS.

With memory of speeds beyond 1600 MHz, the BCLK and Multiplier setting affect the CPU and QPI speed to the extent that other adjustments need to be made (ie lowering multiplier) or stability and OCing problems appear. IETU is nice in that is shows those affects in the settings while you adjust things, or "fixes" other setting to keep things sane.  In my experience, XMP profiles have fine.

Parsec can you explain why , when I used the Intel XMP , ( reading the SPD from the Ram )  , Set the voltage to 1.9 volts insteed of the 1.5 volts which the Ram was rated for ??

The XMP profile set the memory voltage to 1.9V?!  That is not good of course, given this platform, and explaining it will take some checking.  First, I would run CPU-Z on the PC with those memory modules in it, and see what the SPD tab displays, like so:

This memory has the XMP data voltage at 1.5V.  If the memory that you are using has the voltage set to 1.9V, the manufacture is responsible for that, and it may be an error.  If the XMP profile voltage is not set to 1.9V, then whatever program (BIOS?) is reading and setting the XMP data is likely the culprit.  Testing that would be simple, unless there is some other strange interaction between things that causes the voltage to be set to 1.9V. Any idea what the XMP data for that memory looks like?

Unless Gskil put the wrong SPD chip on both modules ??   When I tried them in my other sys they came up as 1.5 volts . Might be a BIOS Prob??

I ran cpu-z over my kingston ram and it showed 1.5 volts for the lower speed settings and 1.65 for the upper.  I'm still waiting for the Corsair to get here.

When you get your coursair Ram call them and find what is the MAX Vlots that they can safely be run at . The DDR3-1333 spec said 1.5 But when I called them they said 1.65 , right now I have them at 1.62 , thats how I am able to run them at 1400Mhz at 39C. stable No errors.

Thanks Bob,

I'll give that a try.

So you're saying the on the DX58SO2, when enabling the XMP profile in the BIOS, sets the memory voltage to 1.9V, while the memory's SPD data shows 1.5V for the XMP data?  That seems like a bug.

Parsec NO I had that experince on two other boards DP45SG and DX48BT2 , The XMP set the ram volts wrong on both boards , What I am saying is you can use it as a guideline but trust your own brain not the machine . If you could trust machines 100% None of would have JOBS.