Memory compatibility on P67 seems to be a joke...
I traded ZenEffect my STT 2200 C8 for GSkill 2133 7-10-7 as I wanted a set of PSC to play with on AMD...
My STT 2200 C8 did only 1700 6-6-6-18 on AMD, while my STT 2000 C7 does 1800+...and GTX2 gets up in 1850-1900 land.
I was worried he'd have major problems with it on Sandy, however he has gotten the ram to play along fine at 2133 7-8-7. It's rediculous.
my STT 2200c8 wouldn't even run 2133 at all, same with Dominator GT 2000c8, what board is Zen using? Giga, EVGA?Quote:
Originally Posted by BeepBeep2
Not sure, this is all he said:
IMO it was a lower bin kit. 1700 6-6-6 1.7v is not impressive, 1800 6-6-6 needed 1.82v whereas my 2000 C7 does 1800 at 1.71vQuote:
also, I have been able to get the sst kit to cooperate at 2133 7-8-7-20 ;) yes alot of sticks wont make the grade, however its all about how you finness it in the end.
Sticks worked for:
Gigabyte Z68XP-UD4 *really diggin' this board
Gigabyte P67A-UD7 B3
the sticks just wouldnt cooperate on the maximus or p8p67-pro. :(
just poped them back in to run off a validaiton.
http://valid.canardpc.com/show_oc.php?id=2018419
My old hyper Dom GT Ver. 2.1-2000 C8 scale ok to 2133ish 7-8-7-24 but with Vdimm 1.8+ and high vtt on UD7-B3
to be completely honest, at 7-8-7-20 2133 (it wont 32m a any higher.) is slower than bbse @ 6-9-6-20@ 2220+ on maximus with these sticks. its all in the subs, and the stt isnt quite making the grade. they jog, not run. ;). Im more in a 24/7 config, and the UD4 does exactly what i need it to... and right now its testing some 2133 7-10-7 Pi's right now. It took a trip around the US and ended up right back on my front door beaten to death. they seem fine and no physical damage.
BBSE seems to be the way to go, I wish I would have bought some of the 2200 9-9-9 Ripjaw X, as those are one of the higher bins and they should be BBSE like the 2133 8-9-8 kits.Quote:
Originally Posted by ZenEffect
i only have a couple of the 8-9-8 kits, but they all perform very well. :D
I'm looking for another set of 2133 C8 ripjawsX to make it 4X2GB...gotta prepare for the new cpu:)
This one....http://www.newegg.com/Product/Produc...82E16820231434
probly will be back soon for the encore before eol
those sticks are c9 at 1.5vQuote:
Originally Posted by Dumo
do you mean the 2133 8-9-8 kit at 1.65v?
I have two of these and they will not run at rated speeds in a 4X2 config, at least on P67
Yep, 4X2 for SB-E
Memory compability not only is different from sort of board but also from every special board, only the chances are higher to get hyper to run on M4E, UD5 and 7 and one or two other boards. Of my remaining 14 hyper sticks at least 10 run flawless on 1:8 on ud5 and maximus z gene, but even sticks which are of the same triple kit two run and one not, despite running all the same on other platforms, 1000 7-7-7-20 at 1,6 for example on 1366. SB is not a platform but a crime in terms of compability and the thousands of bugs implemeted by purpose or random.
Moar voltage for 7-7-7?Quote:
Originally Posted by ZenEffect
Prolly better sticks required. I have 3 GTX2s dimms that don't do 2133 7-7-7-21, not even at 1.76... Other 3 do 2250 7-8-7 1.74
Random, I have two triple kits of Corsair 1866, 4 sticks run, one sometimes, other doesnīt. run at same level on other platforms, whereas all my remaining 2000 C8 STT run on SB, sold two of my 8 sticks a while ago, they go 1100 7-7-6-20 on UD7 at buyer at 1,76-1,78V, and run as well on my UD5 and M4Gene. All 32M and I donīt see any differences at the Corsair GT for exapmle. Sadly I sold a lot of Hypers that I havenīt tested on SB, but I can see no logical reason why of three sticks with same batch, production date and from the same kit that run at same level on 775, 1156 and 1366 two sticks run on SB and one doesnītQuote:
Originally Posted by Leeghoofd
i just don't get why the behavior of hypers is so damn variable on 1155. Is it the bios? Is it the imc itself? what is it :shrug:
i admit i haven't tested any of my hyper kits passed 1.7v, i wish i could find one that worked at even it's rated timing with the 2133 divider. (sp 32m)
btw same as loud... i could not get 4x2gb bbse kit (2000c9 trident) to work @2133 on a evga p67 ftw board. I could get 1866c8 though.. iirc that was the best i could do.
Any tips for running 4 modules @ high speed? just more vdimm? and how high can you put vccio? getting another set of gskill pi 2200c7 soon :)
I can only get them running with all slots occupied with 2T. 1T will post reboot loop infinately with 2133 divider on both asus and gigabyte.
That is normal, 1T is very difficult with full slots.
Thank you
GSKILL TECH
It is simply not possible with timings like 7-9-7, with straight timings like 9-9-9 I have no problems running 2133 1T with 1,53V for example. Board matter or the IMC of the super SB, not a ram problem
I agree with websmile, in that its IMC or board related. For example, i cant 9-9-9- 1T AT ALL with all slots occupied no matter what kit im running. 1T is the killer for me regardless of board across 3 different ICs. If you CAN do it, you must be luckier or more skilled than myself because ive wasted about 12 hours total digging into this issue with no positive results... and truthfully, its 12 hours wasted because gaming, i dont care about 1T vs 2T as the performance differance is not noticable, benching im only using 2 sticks anyways due to the frequency and timing headroom.
This is nothing about skill, as you said it's simply IMC and/or board. No tweaking, just putting in and running. As simple as the fact that no matter what voltage or mems, at 7-9-7 for example same configuration won't even boot, both kits run til bclk limit single at 7-9-7 with 1,65v and less
http://www.abload.de/thumb/8gb32mnf25.jpg
I have been able to run 4X2GB on the 2133 divider but not when the cpu was heavily oc (i.e. 5ghz)Quote:
Originally Posted by websmile
I never even tried a lower clock, regardless that is sick websmile, glad to see it is possible:up:
quick headsup guys... it's back up on egg
http://www.newegg.com/Product/Produc...82E16820231322
well here is my findings after digging into it for the last few days.Quote:
Originally Posted by websmile
Z68XP-UD4 - the data and address VRef settings appear to do nothing but break. The bios is indicating to me that the voltage should be .75 by gigabyte set default. This setting is very useful in getting the memory too cooperate at high clocks. I usually gain a touch by dropping the voltage .01v. in this scenario i would set to .74, however it when set to .75 as indicated or .74, or any other setting it wont post. This was tried on F2 bios and I know its not the latest so im going to re-test a little later.
Maximus IV Extreme-Z - the data and address vref settings work correctly. once corrected I was able to run 8gb 8-9-8-20-1T 2133 @ 1.65v. So for dram data and ctrl ref voltage its .490. I havent tried lower than 1.65v as this is stock voltage for these kits.
pictures are a different gskill kit but so far this method has worked for bbse and bdbg/psc. i have yet to test bclk with all 4 occupied.
http://i48.photobucket.com/albums/f2...976/1ttest.jpg
*edit*
resolved the issue with vref on the gigabyte. turns out, that i was at fault here. vref is defined as vddq/2 so its listing it correctly when using 1.5v vdimm and i was not taking into account the rising dram voltage when inputting values. the methodology on asus is different in that it makes you input a ratio. hope this tidbit of info helps.
Very good work and interesting result- I haven't made further tests on mini aus yet or on UD5 which I got today, with celeron 530 will be difficult anyway as low frequency is not interesting. The 2133 9-9-9 run with about 1,5V no matter which cpu frequency, but as for benching even 8-9-8 is too slow I will rather test some hyper if 8gigs run at good speed and timings, seems to be the easier way