P8Z77-I Deluxe and F3-1600C9D-16GXM... compatibility issues?

[ZeroOne]

Please help... this one is driving me crazy. I have a P8Z77-I deluxe and the F3-1600C9D-16GXM kit (3570K cpu). It fails LinX at stock speeds within the first 1-3 runs at the rated 1600.

Swapping in another vendors set of memory (same size / density, rated at 1866) and running at either 1866 or 1600, the system passes. 20+ runs every single time, tested 5 or 6 times over a couple of weeks.

Here's where it gets tough... the GSkill set DOES pass in a Maximus V Extreme, same exact settings, 20+ runs, multiple times.

I'm not sure why this is not stable in the board it was intended to go into (P8Z77-I).

Using Mem TweakIt, the settings appear exactly the same. The only slight difference is on the Timings #2 page, Round Trip Latency (A) and (B) appear flip-flopped. Same with DRAM I/O L (A) and (B). (Can post screenshots if needed.)

[Tradesman]

Might try simply raising the DRAM or the MC (Memory Controller) (CPU/VTT I think) voltage + 0.05 to start. Might be that DRAM and mobo just don't mesh as well as they should via BIOS

[ZeroOne]

I've tried increasing the voltage with no luck.... another bios is out to "increase system stability" or something like that. They just had one a month before that, same type of thing, and I think "memory compatibility" was in one of them. Seems like a board issue... just can't get past it. Memory runs fine at 1333 though. I know its not the memory's fault. Just not sure what to do. Need to play around with the voltages, but there aren't many. I believe just the vcore, vdimm, and vccsa. Will check tonight. Didn't see vccio as a separate option.

[Tradesman]

If not (no CPU/VTT) then it's prob the VDIMM (might even be back to VCCIO or tis may be one they combined into the VCCSA). Might post your DRAM timings both base and advanced and we'll see if anything looks off

[ZeroOne]

Bumped the vdimm up to 1.55, 1.58, 1.60, 1.62, and it actually seemed to get less stable each time. Was able to fail LinX during the first run at 1.62, within minutes. Curiously, lowering the voltage to 1.47, it ran for 11 runs at like 1.5hrs... something seems strange here...

[Tradesman]

Try the VCCSA up, and let me see if I can get a manual and check it

[ZeroOne]

Well..... 1.485 (if i remember correctly from last night)... it ran LinX for 20 passes with no error. On it's second time around now and at pass 19. Looks like it will complete again. I have no idea why this worked this time, and at a lower voltage.

I did try that other bios, before lowering or changing voltages, with no luck. I then raised the voltage (the test mentioned earlier) with a faster fail rate. Lowering seemed to work better. Before I increased to 1.48x this time I did unplug the computer to move it to a work bench for fitting some lights into it. Bringing it back, reconnecting, changing the voltage as originally planned, and testing, yielded the passing result. I'm wondering if, after the bios upgrade (and subsequent clearing), if it really did just need a power off for a period of time to really reset something...? We'll see...

[Tradesman]

Never can tell, keep us posted

[ZeroOne]

1.485 vdimm passes LinX all day long. Testing now back on auto / 1.5v. It just failed at 19min in, run 2 or 3 (restarted quickly and forget exactly).

Any reason why auto @ 1.5v would fail and 1.485 manual would pass 24/7??

[Tradesman]

Could be the mobo isn't regulating and/or reporting the voltages correctly - it's good it's running, but sort of scary that it won't with a simple 0.015 increase

[ZeroOne]

The plot thickens..... after passing LinX for about 6 21-run passes at 1.485, it failed when resetting back to auto vdimm voltage that set itself to 1.5v. Setting it BACK to 1.485, it is now unstable again, failing after around 1.5hrs, and 10 runs. Ran all day, changed, failed, changed back, still fails. This bios must be complete $#!t.

[ZeroOne]

I see yet another bios is out on 12/4 (0801). They're putting one out every couple of weeks now. I'm on the 11/21 bios (0706) and will update again and test.

For now... not stable on 0706. It wouldn't be as frustrating if it were consistent in any way.

[ZeroOne]

New BIOS is a no-go... failed after 4 runs. This build is quickly losing its fun...

[Tradesman]

Can email and keep emailing them, looks like they have multiple problem they are trying to fix via the BIOS....sadly getting the DRAM end right tends to take a lower precedence than many other mobo issues

[ZeroOne]

Was hoping for a fun little board to do some overclocking... not get things stable at stock. Seems counter productive to have a 4.8 - 5Ghz Ivy and 1333 ram. /rant

Here is what they have to say about these 8GB dimms... any opinion on this? Maybe this is why the other brand that does work in this board which costs around $150 for 2 sticks instead of $70-80 may work? Different chips? Nothing against these, but if the board is THAT sensitive... it may be a reason.

Quoting Asus forum, motherboard section, on RAM:

8GB RAM sticks are rather new, and the avail selection has not been tested extensively by ASUS for qualification. They are a special case - they come with higher latency because of the big drams themselves. Some vendors call out 9-9-9-24 for a single or twin pair pak, others list in the 10 ranges. Whether or not you can actually run your 8GB stick at listed latency is debatable. More so for 2 or 4 sticks.

10-10-10-30 2N, 4 X 8GB
http://www.gskill.com/products.php?index=431
10-10-10-27 2N' 1 X 8GB (single stick)
http://www.gskill.com/products.php?index=510
9-9-9-?, 2 X 8GB
http://www.gskill.com/products.php?index=539
Lower latency means different dram vendors and higher prices - Hynix Samsung Micron/Crucial SpecTek Apacer Elpida etc.
GSkill itself makes no drams.
You want 9-9-9-24 @ 1.5V with 8GB modules - it will cost you.
http://www.crucial.com/store/partspe...8G3D1609DS1S00