Overclock.net - An Overclocking Community - View Single Post - Asus Z9PE-D8 Owner's thread
View Single Post
post #11 of (permalink) Old 09-08-2012, 12:58 PM
BlackenedTush
New to Overclock.net
 
BlackenedTush's Avatar
 
Join Date: Jul 2012
Location: Southwest US
Posts: 83
Rep: 3 (Unique: 3)
Hey ChrisSibbald,

Coolers too tight...?
I mentioned to PhxTriode that my coolers are Dynatron R17s. I had never installed a Sandy Brdge CPU cooler before and I got a little enthusiastic with the spring loaded mounting screws - I had torked them a tad too tight. As a result I was having problems getting any video signals out of my EVGA GeForce 670 GPU. On a hunch Chris at EVGA told me to try loosening the CPU cooler screws. I did and Voila!, everything worked. Evidently the CPU seats on many motherboards are susceptible to very slight warpage under these conditions... leading to just enough distortion to cause marginal CPU Pin connection in the seat, and many potential symptoms. Just a thot...


Mixed RAM kits?
Sorry, not sure what RAM timings you have. But I do have some experience to share that may or may not generally help. Like you, I opted for 64GB, but probably unlike you, I was naive. When my search didn't turn up any single 64GB kit, I simply purchased two 32GB kits. My assumption was pretty basic... that since all of the RAM in the two kits had been tested at 9-9-9-24, 1T, 1.5v @1600MHZ, then all of the RAM in the two kits will work at that speed. But according to Corsair tech support, my assumption excluded the probability that sticks not tested and validated TOGETHER at a particular speed very likely won't work well together in a computer at that speed. I guess there can easily be incompatibility between kit members that's somehow due to two different integrated memory controllers in the CPUs plus mobo memory controllers next to the DIMM slots - some confusion can happen when they are overwhelmed with the shear number of modules being used (8x8).

As it happens, my system has suffered from instability, corrupted OS, and corrupted apps from the very first successful post. And early on, BIOS indicated that one stick that wasn't registering at all. So I swapped it with an adjacent one and because BIOS now picked it up, I thought things were fixed. But other vague problems, frezing, etc. persisted. Eventually after a month of use that stick acted up again by only registering half of it's capacity. Again swapping helped the BIOS see it, but Speccy and some other apps didn't... go figure.

As I'm sure you know the Auto mode in BIOS picks up several (but not all) of the XMP ratings from the sticks, and it in turn sets the motherboard to work with the RAM at those settings. But only latency, timing, and voltage are honored in that way. Frequency is not. As in my case, BIOS often arbitrarily defaults it to a lower value than rated (e.g., my 1600 RAM was set to work at 1333MHz). But you also know we are provided a way to force the frequency to a handful of common settings. I naively thought that many of my instability issues were due to my RAM not being set exactly at rated speed. Haha. So I forced mine up to 1600. But my problems with instability and corruption only increased (What the heck?!...). I had very confusing symptoms. For instance, occasionally my video driver would fail for a few seconds and then recover. Evidently my assumptions about kits were wrong. Speed rating is a little more particular. And a healthy base of RAM is necessary for a healthy system.

Now not many restarts after forcing my frequency to 1600, it would not boot into Windows, and then after a few more force quits, it wouldn't even post. I went back in and yanked out half of the RAM, making sure to include those sticks that I had swapped earlier. Then it posted and Windows loaded! Testing with Memtest86+ then helped me validate the health of each stick one at a time. That testing showed one big culprit to be the stick I had previously did the swapping with. But I also found one more bad one in that kit, plus a more mildly deficient stick in the other kit (that had remained in the computer).

So evidently my mixing of kits amped up an already bad RAM problem. Corsair tech support told me that if my problems had been simply due to mixing kits, that I could fix it. But I would most likely need to leave my notions that all of the RAM would just get along together at the rated speed. Instead I would need to manually specify some lower settings that all could work together at. They suggested for starters that I go back to 1333 and also move the supply voltage upward a little to stabilize and open things up.

So these days the stakes are higher. DD3 RAM ratings are strictly tied to the kit idea, where groupings of sticks are validated together at a particular speed. Now if a kit works well together at the tested speed, one might get lucky and have a group that could also work well together at a higher clocking. But the more RAM you have, the more “Crucial” the kit philosophy is (no pun intended... )

If Mom is too slow repairing the hole in your jeans then, "sverte ræva og gå naken"
BlackenedTush is offline