Odd issues: TWIN2X2048-6400

Well, it took a week of trying different things, but I think I have some issues with my latest TWIN2X2048-6400 purchase. I bought a 2nd set of TWIN2X2048-6400 for my new 'family room' PC build (other set works flawlessly on my dedicated gaming system).

System Specs for reference:

(Family Room PC)

AeroCool ZeroDBA 620W PSU

Asus P5L-VM 1394 (i945G) motherboard

Corsair TWIN2X2048-6400 RAM

C2D E6300 CPU

Zalman 9500LED HSF

On-board video GMA950

Hitachi Deskstar 160GB ATA100 HDD on Single ATA connector

Asus 16X SATA DVDROM on Master 1 SATA channel

Have set the TWIN2X2048-6400 to both bySPD and set manually to the spec of 5-5-5-15. No performance tweaks or overclocking.

Nature of issues:

After OS install, updates, DX update, etc. went to do my 'normal' battery of stress testing to make sure everything was stable.

(1) Prime95 'yellowed' out (error) on auto, CPU0, CPU1. Tried a few times with the same results. Rechecked BIOS settings, set RAM manually (see above). Same results.

(1a) Encountered a few off program errors "file will not load" for example on a few minor programs and IE7 seems to crash out randomly.

(2) Ran CPUBURN by itself (in error checking mode) for 6 hours without error. Hmmm.

(3) Ran CPUBURN and Quake2 Demoloop (yeah its old, but it works good for multi-threading with other testing programs). CPUBURN ran without error, Q2 demoloop crashed after 1 hour, then half hour, then 1.5 hours. More testing required.

(4) Double checked all of the above with Intel TAT and CORE Temp to ensure processor running in spec for temps. Check, never above 40C on either core, fully loaded.

(5) Got out my handy Memtest86 boot CD. Here's where it gets a bit odd:

(a) Runs the STD battery over night without any errors at all.

(b) 'freezes' if I try to run PROBE or BIOS-ALL memory runs at 43% on both tests. Tried 3X each and froze at same spot. Curious.

(6) DL'd and installed windows MEMTEST program. (http://hcidesign.com/memtest/) Ran two instances to get full memory tested and recieved MULTIPLE errors over several runs.

(7) Started pulling and re-arranging RAM sticks, one stick, both slots, switch, repeat. Same issues encountered above for all tests. One stick did seem to generate less errors in the Windows MEMTEST than the other though. P95 still had issues, Memtest86 still ran fine without errors under STD, but froze at 43% under PROBE and BIOS-ALL.

(8) Put sticks in my P5B-E based gaming system (which normally has its own set of TWIN2X2048-6400). Memtest86 began showing errors in STD almost immediately. However, PROBE actually ran past the 43% mark, but BIOS-ALL froze almost immediately. I did not boot the OS on that system since I did not want to corrupt anything.

(9) Bought a matched pair of 'cheap' XXXXXX PC-667 2X1GB kit from the 'egg. Received it today, and installed on the P5L-VM system. Memtest86 runs without issue, PROBE seems to run fine, BIOS-ALL seems to just not want to run on anything.

(10) After running memtest86 for most of the day I booted into the OS and ran P95 for a few hours with no issues. Ran windows MEMTEST for two full cycles without errors while running Q2 demoloop in the background with no crashes.

(11) All of this took about a week to run through here and there. Getting the 'cheap' ram was the final step in trying to isolate where the problem lay. At this point it is pointing at my newest set of TWIN2X2048-6400, though my results with memtest86 are somewhat confusing. What I can say for certain is the issues I encountered are now non-existant with the cheap XXXXXX ram installed.

So, Ramguy, any ideas, or does this look like I've pretty much tested what I can test at this point? I tried to cover all the bases before posting this.

Thanks, Ram Guy.

The P5L-VM 1394 does not have a vDimm adjustment parameter. At least not in the last official BIOS update, 0602. Per the manual, there are also no listed jumpers for vDimm adjustments. vCore is adjustable under the AI OC drop down, but no vDimm.

Speed can be adjusted to 667 and I have set it manually with no change.

The default tests for memtest86 run without errors on the P5L-VM on the default tests. I think it completed 12 full cycles overnight without errors.

I am wondering at this point if the problem is the lack of voltage adjustment. As the Adata RAM runs at 1.8v default, and the TWIN2X2048-6400 is supposed to run at 1.9v default (2.0v per your suggestion) this could be problematic. My other set of TWIN2X2048-6400 runs at default vDimm of 1.8v on my P5B-E without issues, however. Perhaps that set is the exception, rather than the rule. As all of the issues clear up when the Adata ram is installed, it may point towards a voltage issue.

I'll triple check on the vDimm settings, however, I've scoured the BIOS previously (and repeatedly to my dissapointment) with no adjustments. It's not intended as an enthusiast's board, so vDimm adjustment may not have been a priority on this board for Asus.

Thanks again, I'll update once I get home and get a chance to tinker.

Double, double checked the BIOS and board tonight. Yep, no vDimm adjustment on the P5L-VM 1394 board.

I just tried the new set in my P5B-E that the other set has run fine in at default voltage. OUCH. OS crash, OS consistency check, yikes. That was at 1.9V. Set it to 2.1 and got into the OS fine. I'm going to run the sticks overnight at that vDimm on the P5B-E board and see what comes up on memtest86 default tests.

If they run fine at the 2.1v vDimm in the P5B-E I might just swap out the older set of TWIN2X2048-6400 into the P5L-VM 1394 and see if they run fine at whatever the default vCore is on that board.

Just one question, Ram Guy, are there any long term ramifications of running the ram set at 2.1v? I read the FAQ on the "P5B-X" and noted you suggested setting the vDimm to 2.1v. No warranty issues with running the ram at this voltage long term is there?

Double, double checked the BIOS and board tonight. Yep, no vDimm adjustment on the P5L-VM 1394 board.

 

I just tried the new set in my P5B-E that the other set has run fine in at default voltage. OUCH. OS crash, OS consistency check, yikes. That was at 1.9V. Set it to 2.1 and got into the OS fine. I'm going to run the sticks overnight at that vDimm on the P5B-E board and see what comes up on memtest86 default tests.

 

If they run fine at the 2.1v vDimm in the P5B-E I might just swap out the older set of TWIN2X2048-6400 into the P5L-VM 1394 and see if they run fine at whatever the default vCore is on that board.

 

Just one question, Ram Guy, are there any long term ramifications of running the ram set at 2.1v? I read the FAQ on the "P5B-X" and noted you suggested setting the vDimm to 2.1v. No warranty issues with running the ram at this voltage long term is there?

Okay, after fiddling about a bit, I ended up running the 'new' set of TWIN2X2048-6400 overnight in memtest86 on my P5B-E system (pulling my old set out) at 2.0v vDimm (since my P5L-VM 1394 has no vDimm adjustment). No errors were reported in 9 passes.

Prior to kicking that off last night I was able to run P95 in the background for multiple iterations whilst playing STALKER for about an hour with no issues. (See notes above in my prior response, 1.9v resulted in OS crash and OS consitency check/file rebuild...)

I am going to posit at this point that the 'new' set of TWIN2X2048-6400 is going to be a 'no go' in the P5L-VM 1394 since there are no BIOS or jumper settings with which to adjust vDimm and it appears the 'new' set requires 2.0v to run stable.

I suppose I can always use this set in my P5B-E and try the 'old' set in the P5L-VM 1394. I just want to confirm that running long term at 2.0v or 2.1v on the TWIN2X2048-6400 is 'okay', and not going to cause issues in the long run?

I re-ran the modules again last night making the precharge change as listed (I had previously changed voltage to 2.0v for the prior night's run).

Again, through just shy of 10 passes there were no errors on the P5B-E. I think with the voltage adjustment these are working just fine for the P5B-E, which really nails down for me that the issue relates to the lack of vDIMM (RAM voltage) adjustments on the P5L-VM 1394 which the set was intended for use in.

I would suggest that you update your database to reflect that the TWIN2X2048-6400 is NOT compatible with the Asus P5L-VM 1394 as that motherboard does not have RAM voltage adjustments in the current BIOS or as jumpers. Lacking the ability to adjust RAM voltage (upwards) makes this RAM unusable in the P5L-VM 1394. It would probably save you future headaches in support and save frustration on the part of your customers.

I have no issues using this ram in my P5B-E instead (not what it was intended for since I already have a set of TWIN2X2048-6400 in that system), but if the P5L-VM 1394 was my only option, I would probably be a bit irritated.

All of that aside, thanks once again for your support, help in figuring out where the 'true' issues lay (voltage), and for making a great product. Corsair is still A+ in my book. Great support is part of why I go with Corsair. Thanks, Ram Guy.

Thanks, Ram Guy. The board with no voltage adjustments is the P5L-VM i945G Asus motherboard. The P5B-VM is a i965G board and is not the board in question (I don't own that board).

The P5B-E is a board I have in my dedicated gaming system that made for easy testing of the RAM that was having issues in the P5L-VM. The P5L-VM 1394 is the board that I just got and will not run the TWIN2X2048-6400 due to lack of voltage adjusments.

http://tools.corsairmemory.com/configurator/product_results.aspx?id=13270#other_modules

The TWIN2X2048-6400 is listed about half way down the list of 'Complete List of Modules Compatible with your Product' for the P5L-VM 1394.

Thanks again. I know you are super busy right now with all that has been going on and I thank you for your time.