[koitsu]

First, please read my *entire post* before replying. I believe you found the source of the problem at the very end.

Originally Posted by RAM GUY

The 6400C4 modules will fail when using them together in dual channel on two separate P5N-E SLI motherboards which both are using the 1101 BIOS.

Mostly correct.

1) BIOS version does not matter; I have tried all versions between 07xx and 12xx, the behaviour is the same,
2) The 6400C4 modules I refer to were purchased in a TWINX pack, which means they're supposed to be perfectly matched (according to the advertising, packaging, and reference manual on Corsair's site). They are the same "version" according to the version number on the heat spreader label.

Further technical details:

1) I only buy TWINX pairs, and I only use a single pair (e.g. two sticks); I do not EVER install 4, so this is not an issue of the motherboard screwing with settings due to 4 sticks in use,
2) I do not mix 6400 and 6400C4 modules, nor will I ever.

Originally Posted by RAM GUY

Configuring the modules to their tested settings (4-4-4-12-2T, 2.1 volts, 800MHz), gives you no problems on either system?

I do not "configure the modules to their tested settings" -- the only people who choose custom timing/voltage settings are a) "leet gamer d00dz olzz gotta get 34893243489fps in quake!!!1!!" which I am NOT, and b) individuals who are having stability problems with modules (which in this case, I am/was).

I rely solely on JEDEC and EPP data for all of my timings and negotiated voltages (1.8V for JEDEC, and 2.1V for EPP). I'm using EPP right now, so the voltage as confirmed in CPU-Z is indeed 2.1V. When using JEDEC, the voltage is 1.8V (also confirmed in CPU-Z).

But let me make this clear once more: voltage is not the problem, Command Rate is.

Originally Posted by RAM GUY

Command Rate is a setting for your memory controller on your motherboard, it is not specifically dependent on the type of memory you are using. You mention that the Command Rate is written in the EPP profile as "2T" and when you enable the EEP profile, the system is still setting Command Rate to 1T. There is no way that the memory itself is changing this value in your motherboards BIOS when enabling the EPP, the SPD is static and can not be changed by your motherboard.

Where did I say the SPD profiles were inducing BIOS changes? I didn't. I'm saying: the BIOS, by default, uses whatever values are specified in JEDEC or EPP profiles. The setting in the BIOS is "Auto" (this is the factory default).

The BIOS will print out what timings are negotiated during POST. The same data is available using CPU-Z, which shows you what your current/actively used timings are (see "Memory" tab in CPU-Z).

When using "Auto" with the 6400C4 modules, a Command Rate of 1T is automatically negotiated, which causes the system to randomly crash. If you force Command Rate to 2T in the BIOS, the problem goes away.

If you use 6400 (non-C4) modules with "Auto" selected, the negotiated Command Rate is 2T (using either JEDEC or EPP).

The SPD data is 100% identical between the 6400C4 and the 6400. I've personally confirmed this (which is also what prompted me to blog about it).

Originally Posted by RAM GUY

Your motherboard/BIOS is setting its own values for Command Rate and ignoring the EPP for that setting. In the past, when we have seen this issue corrected with a newer revision of the BIOS, which would tell us that something other than the memory was setting the Command Rate incorrectly. More info on Command Rate can be found HERE

See above; the 6400C4 and 6400 both have the same SPD data. So I repeat: the problem is with the DRAM, and not with the motherboard or the BIOS.

Originally Posted by RAM GUY

The 6400 and the 6400C4 modules that you have are likely using different chips from different manufacturers and each chip has different characteristics. There are many more values in the SPD other than the ones that are listed by CPU-Z, and its unlikely that the SPD is the same on both parts.

Now *that* explanation make sense! :-) That means that certain brands of DRAM are not compatible with the P5N-E and P5N-E SLI.

A question for you: does the EEPROM SPD contain the brand/manufacturer of DRAM anywhere? I doubt it, but I thought I'd ask.

The problem is then ultimately this...

Corsair does not disclose on the packaging, on the heat spread labels, or in a string ANYWHERE on the RAM what DRAM manufacturer was used. The DRAM is covered with the heat spreader, which cannot be removed without voiding warranty or risking damage to the modules.

I fully realise Corsair reserves the right to change DRAM manufacturers whenever they desire, or possibly you have multiple fabs that use different DRAM -- that's fine.

But you need to start printing this information on your products somewhere. I guarantee you by doing so, you will make the tech community's job of debugging what Corsair memory is compatible/incompatible with motherboards significantly easier, and even more importantly, more reliable.

Imagine this:

User #1: My Corsair XXX modules work fine, and they're Hynix
User #2: I have the same modules, but mine are Samsung DRAM, and they work fine
User #3: Me too, also Samsung, things work fine
User #1: Hynix incompatibility maybe? I replaced the RAM I got with the same model but with Samsung on it, and it now works!

Please see about reporting this request to your manager, or whoever makes the decision of what gets printed on the spreader label or inside of the packaging. You could save a lot of support pain doing this. :-)

Thanks!