^ I don't know what you mean by 3.9ns, etc. I understand it's something like mhz x timing = ns, but i dont really understand it. I just read that higher is faster on refresh on the asus site, but i always found on overclocking psc's that like ~5k or 6K would be terribly unstable but ~8k+ was totally stable. I would raise refix9 too, so 5k/50, 6k/ 60, etc sort of ratio.
I mean I'm setting high vdimm, i guess, I'm using 1.8v, but i mean 2 sticks of ram isn't ever going to get near 70C, they run at like 40-50C. 4 sticks might get more like 60s on 1.8v but still, staying cool.
There's a ton of wild looking tertiary timings on my z87x-ud3h F6 bios, i think some are haswell related and some are gigabyte, if anyone could shed any light on it, I'd appreciate it:
First off there's a bunch of timings with _dr and _dd attached. I assume that stands for Different Rank and Different DIMM (saw those words on hwinfo in the memory section). I don't knwo what that really means, do I set them the same or what as each pair or what. Then, more confusingly, there's the original value. Like, RDRD_dr, RDRD_dd, and RDRD. It's also like this for WRWR, RDWR, WRRD.
And there's CPDED, PRPDEN, WRPDEN, and XPDLL, XP, TAONPD.
Then, there's confusing timings like Round Trip Latency and IoLATR0D0, but like x4 times.
Finally, what's tCKE for? I read asus' description about sleep, is that the sleep I think it is? So a stress tester wouldnt actually test tcke for you, you'd need to repeatedly set your pc to sleep, unsleep to test it right? So... set it as tight as humanly possible, and it doesnt affect bench scores at all?
Those Gskill Hynix CFR's that I bought, I've been going one by one on each timing in the bios and tightening as much as possible. I've just started working on the tertiaries, my results so far:
2723mhz (would use 2800 but using 47x102mhz = 4.8ghz cpu) 1.8v (there was some secondary that needed 1.8 isntead of 1.75v).
I ran maxmemm on 2666CL11 and 2800CL13 (i played around and got all 4 timings as tight as could be on those speeds) and 2800CL12 was by far the best result.
tWR 24/A (changes wont stick)
tWL 8/A (tighter wont stick)
tRFC 105 (i know 102 is unstable, so gives a bit of headroom)
tRTP 5 (3 was very unstable, so i just set to 5)
tFAW 14 (12 was unstable, i read on Asus site this should be tRRD x 4, is this true? Because that would be 4x6=24, which i think was the auto setting)
CR 1T (duh, as i understand CR only affects mhz and since mhz isnt the problem here, timings are)
tRDRD_dr 7 (set as pair, 6 unstable)
tRDRD_dd 7 (set as pair, 6 unstable)
all else Auto via XMP (which goes very loose on 2723mhz).
Using H264 benchmark as my stress test. This will likely be near my final results. I think h264 is not good for ram testing, this fails in 5 min of p95 so i'm going to find a better 24/7 stress tester.
My 5ghz Ivy Cinebenched at 10.16, so big improvement!Edited by Belial - 7/27/13 at 4:48pm