Hey guys, i'm hoping someone on this board may have an insight into the following. The rig is "parkbench" below. Delidded 3770k, 4x4g sticks of hyperx 2133, water cooled... Etc. i get an easy 45x with 5/23 mV to offset and turbo, p95 4h, IBT, stable... 1.198 load vcore LLC3. No whea, nothing. Clean OC. At 46 with 10/53 mV LLC3, same rock stable. So i go to 47, and no matter what i'm plagued with 124s. IPLL on or auto: Raise vcore, 124 even faster, lower vcore, can get10 min or so of p95 or IBT then 124... Lower vcore further and whea pop up. Ive use enthusiate current, auto current, set 500 for max, tried pwn frequency at auto, 500. Lowered PLL, relaxed mem timings, used "compatability mode for timingss, and increased dram V one notch or two. Tried VTT1 and VTT2 increases... Etc.
If i run p95 with 1600 mb ram... Maybe a minute or two longer vs 8192 mb before a 124! No whea unless i lower offset and then, trying to tune it with PLL or vcore ... 124.
So i switch to fixed OC. 49 at 1.300-1.315v LLC 1, iPPL on, 500 for currents, or "Enthusiast" - Boots no problem, superPI 1M is clean. IBT standard or with 12288mb ram... Stable! Temps never above 63C. Browse, open files, run kolance controller software... All good. Oh yeah, all along reading vcore from the VT1 it is consistently 8-10mV lower than open hardware monitor, occt, or cpuz. So that's workable. And reading the capacitors directly = software!
Start p95... And boom, a 124! No whea, just bsod.
Some new adata 2133 (2x8G) arriving today just on the outside chance the hyperx are funky. But they las overnight memtest, AND are not a problem at 46x or 45x after 8 and 10 hours prime at 8192 mb use. Down clocking the ram to 1600 or removing any two sticks makes no difference with stability.
I'm thinking next i pull the titan, and use the iGPU to take that out of the question.
Nick's canned 49 OC -> bsod 9c.
Edited by Jpmboy - 4/16/13 at 8:27am