High end 8GB DDR3 kits..where are you?

[twicksisted]

i was under the impression that with sandybridge the ram speed/timings diddnt make a "real world" difference

 

[PaulieG]

so get a smaller cooler, isn't like that chip needs a 300W+ cooler to keep it under control

LOL. I could run it with a single fan in pull, but I'm kinda fond of that Venomous X.

i was under the impression that with sandybridge the ram speed/timings diddnt make a "real world" difference

That is partially true. Though we are not talking real world here. I want a set that I can push if I want for benching.

 

[sneekypeet]

so run it passive with a desk fan blowing in its general direction

 

[PaulieG]

so run it passive with a desk fan blowing in its general direction

LOL. Funny thing is, I can run it passive if I wasn't crunching 24/7. Gonna go look at some 2x2GB kits, and fill the board up.

 

[GSquadron]

I am surprised none mentioned the dominators:
CORSAIR DOMINATOR GT 8GB (2 x 4GB) 240-Pin DDR3 SD...

 

[cadaveca]

too expensive. I got 2133 MHz 2GB Dominator GT DIMMs, 4 of them. The Mushkins are better.

 

[cadaveca]

Like this clocks? Just two sticks in while I'm tweaking the latencies, which I got perfected last night, will pop in the second set and see how it goes. They passed 8hr prime overnight, so they are good to go. 2133 MHz 9-10-9-28 @ 1.54v(haven't tested lower)

BIOS shot so you can see the subtimings I used(swapped in 1T after testing a bit, 1T is fine wit h2 sticks, but 4 will need 2T):

AIDA64 bandwidth:

 

[PaulieG]

Like this clocks? Just two sticks in while I'm tweaking the latencies, which I got perfected last night, will pop in the second set and see how it goes. They passed 8hr prime overnight, so they are good to go. 2133 MHz 9-10-9-28 @ 1.54v(haven't tested lower)

BIOS shot so you can see the subtimings I used(swapped in 1T after testing a bit, 1T is fine wit h2 sticks, but 4 will need 2T):

http://www.techpowerup.com/forums/attachment.php?attachmentid=42869&stc=1&d=1310230312


AIDA64 bandwidth:

http://www.techpowerup.com/forums/attachment.php?attachmentid=42868&stc=1&d=1310230100

Nice. might have to go hunting for those sticks.

 

[cadaveca]

They are pretty flexible, but so is just about every PSC stick is, too. I think G.Skill 6-8-6-24 1.5v kits might be nicer, but are still a bit expensive for my blood right now. I'll be keeping an eye on them though.

There's realyl not alot fo tweaking to be had on P67, but this is what I have come up with to entertain myself. Clearly, with 2133 @ 1.54, there's some overhead left here, but I've been tweaking for 24/7 on the rig I've been posting pics of the past day or so. The 2133 G.Skills I got are for the review rig aren't quite as good. At the smae time though, G.Skill does sell a set with similar timings to what I'm using here, same voltage, too(1.5v in BIOS).

Once I get all four sticks going, I'll post up some 2200 and such for ya.


Oh, and TRAS is key with PSC, low volts, and high clocks.

 

[PaulieG]

Damn. I thought that I had the Mushkins stable at 1866. With the cpu at 4.0 they passed a 10x LinX run at 1866. Then I went back to my 24/7 STABLE 4.6 overclock, and with the Redlines at 1866 I FAIL the same run. Went back into the bios at put the ram back to 1600, and once again I passed LinX. WTF?
I'm wondering if the higher cpu overclock is causing the ram to need more VCCIO? Hmm...

 

[cadaveca]

Most likely, yes, a bit more VCCIO may be needed. With the higher multi, both L3 and memory get hammered a bit more, so a bit of vDIMM might be needed to.

Don't forget to play with TRAS.

 

[PaulieG]

Most likely, yes, a bit more VCCIO may be needed. With the higher multi, both L3 and memory get hammered a bit more, so a bit of vDIMM might be needed to.

Don't forget to play with TRAS.

Well, VCCIO bumps were no help, even up to 1.15v. vdimm is already at 1.65v, which I don't want to push with the SB platform. TRAS adustments were no help either. I tried running 24,27, and 28. No go for stability at 1866. What's really weird is that I can get into Windows at 2133 and even do a Maxxmem run at 9-11-10-28. Surely there has to be a way to get them stable at 1866. I'm perplexed.

 

[cadaveca]

With 4 2GB sticks, I run 1.15 as a minimum.

Perhaps it's other timings causing issue? Wanna plug a FAT32 drive in while in BIOS, hit F12, and post some screenies(assuming you'r on the ASUS board)? Let's see what ya got.

Sadly I do not have 4GB/DIMM. Soon, though.

 

[PaulieG]

With 4 2GB sticks, I run 1.15 as a minimum.

Perhaps it's other timings causing issue? Wanna plug a FAT32 drive in while in BIOS, hit F12, and post some screenies(assuming you'r on the ASUS board)? Let's see what ya got.

Sadly I do not have 4GB/DIMM. Soon, though.

Dave, please take a look.



Uploaded with ImageShack.us



Uploaded with ImageShack.us



Uploaded with ImageShack.us

 

[cadaveca]

trrd and twr are very low.

try (for 1866)

7-auto-auto-14-7-24-7-6-8 secondaries, leave tertiary on auto. Does it fail in memtest, or pass? if it fails in memtest, how does it fail...which test, and how many errors?

If it masses in memtest, how are you testing windows? Spi32M? then prime? how does it fail?


Bump SA voltage to 0.935.

let me know how it fails, I've got the P8P67 PRO in the review station, so I could fire off some screenies and do some drive testing, but again, I won't have any 4GB/stick until later this week in the earliest. I knew I shouldn't have waited...been wanting to suss out the 8GB weirdness long ago!

 

[PaulieG]

trrd and twr are very low.

try (for 1866)

7-auto-auto-14-7-24-7-6-8 secondaries, leave tertiary on auto. Does it fail in memtest, or pass? if it fails in memtest, how does it fail...which test, and how many errors?

If it masses in memtest, how are you testing windows? Spi32M? then prime? how does it fail?


Bump SA voltage to 0.935.

They both pass memtest in single config, but fail in dual config. Multiple errors, and typical error codes for memory related issues. Using LinX, IBT and prime for testing in Windows. Typically failing at 3rd or 4th cycle in LinX and IBT. Prime will fail in just one thread after about 5 minutes. Wow, I'm rusty with secondary timings. Didn't have to mess with them much to get the speeds I wanted in x58 and p55. Actually, don't think I've touched them much since DDR2. LOL. Gonna try those settings after lunch.

 

[cadaveca]

Let me know WHAT test in memtest it fails, and that'll help me tell ya what timing to change. I'm gonna say it's a controller issue...so timings are whcky, maybe timings we don't have access to.

 

[PaulieG]

Let me know WHAT test in memtest it fails, and that'll help me tell ya what timing to change. I'm gonna say it's a controller issue...so timings are whcky, maybe timings we don't have access to.

Memory controller is what I was afraid of. I hope we are wrong. Thanks for the settings. I'll run memtest later and post the screenies.

 

[LAN_deRf_HA]

If I recall from your review wasn't the pro bad at memory clocking? In that it took notably more volts to be stable? 1.7 where other did the same with 1.65.

 

[cadaveca]

If I recall from your review wasn't the pro bad at memory clocking? In that it took notably more volts to be stable? 1.7 where other did the same with 1.65.

Yeah, that was with 8GB of 2GB/stock Dominator GTs.

I also posted in the comment thread that the most recent BIOS seems to have fixed that issue....I couldn't get 4 sticks to boot with CAS6, and I gots LOTS of mems here.


Both problems were solved shortly after my review. Again, OEMs listen to real feedback!

 

[PaulieG]

Well, I misplaced my memtest disc, and i'm out of blanks. Gonna have to run out and grab some later. However, I did do a quick LinX run with the suggested settings, and I failed on the 6th cycle. On a positive note, I decided to play around with some tight timings at 1600 to compensate for the lack of speed. I'm LinX and IBT stable at 1600 7-8-7-18 1T with the Redlines on 1.64v. Not too shabby.

 

[Hotobu]

Don't know much about OC'ing memory, but wouldn't it be better to try something with a lower voltage? All the memory I've seen talked about thus far has been 1.65V. What I've read has put me under the impression that these are from older production runs as memory manufacturers are putting out newer sticks that run on lower voltages (for various reasons). Also while 1.65 V memory will obviously work 1.5 is recommended for SB, and I think 1.65 is near the ceiling for voltage tolerance on the memory if you try to change those values.