Popped 2 kits of HYKO and 2600 is np ,didn't touched tertiary's just yet biggrin.gif
Are those good scores? I really can't tell what I'm looking for, I'm looking at your copy/read/write, and then your 'benched in' and latency times, and they don't seem too good (ofc tertiaries can change everything, but I only got like ~+6K from messing with the subs - big, but i mean i was still at 30k+ copy).
Belial - Yeah, tweak the sub timings but the auto adjustments are important. There's about 40 to 50 different timings. One not set correctly can make it so you can't even POST. I noticed the difference coming from an ASRock Z77 Extreme6 to the ASUS Maximus V Formula. Same RAM, they ran at the same speeds more or less but it took hours to get there on the Extreme6. A lot of tweaking in the BIOS and having to start all over again when I would fail to POST and have to clear CMOS. The MVF would change stuff for me, even if it was tweaking it to be faster. My scores were a little better, there are a lot more options to chose from in the BIOS, even some preset profiles that are a great starting point. And yes, with my MVF using MEMtweakIT, I could change every timing within the OS except CAS Latency.
Oh okay, so it's really ease of use for ram overclocking. Is this something that's like brand vs brand or board vs board? I mean wouldn't you think that the same ease of use that occurs in some Asus high end board would occur on the low end board? Wouldn't they be like 'okay guys we know that TREF causes a lot of problems so we know to put it at X on this board when it sense Y IC, but let's also apply this change to our lower end boards too'?'
The reason all the boards don't have the same options in BIOS is the same reason every Toyota Camry doesn't have leather seats and a premium sound system. Not everyone is willing to pay for all the bells and whistles. Both cars will get you to work and back, while one may be more enjoyable.
But I mean the toyota camry still has like the same odometer... i dont know a good analogy but I mean if its' down to a firmware/software issue, you'd think they'd just put the awesome firmware/software on their lower end boards since it's not a hardware issue. I guess that's kinda lame but I understand you gotta make the more expensive product appealing somehow.
So basically for someone like me who's insane and only playing with one kit of RAM and tweaking every single timing, RAM overclockability on one board or another doesn't matter... but it does make things a bit easier to figure out and POST. So really the difference in motherboards for someone like me - tweaking every single sub timing, only playing with 1-2 kits of RAM - is the VRM quality. And sure, some ln2/overclock features like buttons and stuff, and then ther's some boards like the Impact which is crazy because it has 2 dimm slots only and less tracing (so really any micro board with 2 dimms is best for ram overclocking?)?
As for OS changing, is that an Asus thing or an Asus-high-end-board thing? That's quite interesting, I do recall that changing timings in the OS really made things easier on Z77X-UD3H using memtweakit, because I could change the timings in the OS and keep tightening until the system would crash or fail p95 in a 10 second quick test, so I was able to dial down the optimal timings pretty quickly, whereas on this system and this time around I had a really hard time figuring out tWTP was my enemy.
I still am, just had a HW error in p95 after the 8th hour on tWTP=20. And just a week ago I was operating under the assumption that tWTP=12 was perfectly stable, really took me a long time to figure out that was what was preventing me from being p95 rock stable. I guess the sweet spot is 21+, still trying to dial it in...
So I have a simple question guys - does changing the slots you put the RAM possibly change RAM performance? Because check this out, I swapped my RAM from the recommended slots (going by 1-2-3-4) from 2-4 to 1-3 and my copy went up 1000mb/s, my write went up 1000mb/s, my latency dropped by 0.5ms, and my 'benched in' time seemed to drop an average of 0.200s and is hitting regularly what was before only 'my best timing ever only reached once a dozen runs'. Check this out:is this an awesome score? it's awesome right? Am I doing it right? This is 24/7 stability settings, not even benching settings
Compared to like:
Is this just because of like somehow clearing the RAM when I unplug/plugged it, or because of switching RAM slots? I mean I moved them from
the recommended slots, it's pretty crazy to hit such a huge bump in performance just from switching the RAM slots. It must be something to do with just like clearing the RAM, right? I'm consistently
hitting these higher scores, before they were consistently at what is shown in the 2nd pic (and that 2nd pic is actually with tighter tWTP too).
edit: i switched it back and i guess its just the pulling of the ram... still, what a weird results. here's my superpi too btw:
edit: omg 2 things happened that might make all my testing irrelevant.
1. friend is letting me play with same ram kit i have (i recommended it to him after all) and it might
be better binned. not sure, its close either way.
2. At 1.85v i can not only boot into CL11, it seems CL11-14-12 is actually prime95 stable (so far, i was on 1.7v CL12-15-13 because i thought it didnt scale with voltage, ie 1.75v wasnt more stable). As I understand, 1.85v is a good max 24/7 voltage so... man i just assumed 1.85v wouldnt do anything.Edited by Belial - 11/28/13 at 7:52pm