Can't stable "Elpida Hyper" With i7-2600K

Quote:

---

Originally Posted by dinos22

i cant even boot past 1.24vtt eeek

---

On my UD5 going over 1.28vtt makes ET vcore go up, I think it may be a bug or by design as I can boot more vtt with more vcore.

1.28vtt max here. No boot at 1.3vtt

Quote:

---

Originally Posted by OC Maximus

1.28vtt max here. No boot at 1.3vtt

---

Edited

forget that I read it wrong edited it & thought brolox it was right the first time.

More vtt needs more vcore, looks like about .14~.25 is where my UD5 likes it (with my kit anyway)

Quote:

---

Originally Posted by stasio

Same as me.
Loaded "Optimized Defaults",set memory to 2133,8-8-8-24,CR2,1.66V and 32M STOP on loop 2.:mad:
Btw,what is the difference in Auto/Normal for QPI/Vtt,SAV,PCH and PLL?
I didn't see in ET6 any changes(always on stock voltage!).
Above 5GHz,I passing now with DDR3 1600 and 6-6-6-18.
Up to 5GHZ 1876 very unstable,2133 no way.:shrug:

---

Same for me with hypers,2133 9-9-9-CR2 gives errors in s-pi32M.With CR1 it would instantly freeze...:rolleyes:

OCM, try bumping vtt up a bit.. Sounds bit off tbh..

Quote:

---

Originally Posted by Johnny87au

OCM, try bumping vtt up a bit.. Sounds bit off tbh..

---

what do you suggest for vtt?

oh dam i didnt see you tryed 1.3, little puzzed .. Maby bit more vtt and Dimm voltage? Only thing i can possibly think off.. That or 32m is having problems with max frequency of 2133 on the giga boards.. ?

Quote:

---

Originally Posted by dinos22

i cant even boot past 1.24vtt eeek

---

Me neither 1.2 max boot, this other chip i'm working on is severly voltage challenged even more so vcore and vtt wise.

In fact i think i figured out some UD4 users issues. It probably there chip.

I could run 1.6+ Vtt on my personal chip with eztune once in windows.

Not that it helped much instead of a crash i got a freeze near the end.

For me it was a matter of fine tuning subs per set of ram, one set of GT's liked totally diff subs than another.

Tuning them in at 1866 6-6-6 first then taking what I knew passed there was the ticket to getting them smoking at 2133.

Have you compared 3Dmark performance Brian between the Gigas and other boards... quite some interesting things there... wonder if it's due the non EFI bios...

Quote:

---

Originally Posted by Leeghoofd

Have you compared 3Dmark performance Brian between the Gigas and other boards... quite some interesting things there... wonder if it's due the non EFI bios...

---

Seems pointless with a 5gig limited chip atm, give it a go when I find the right chip.

my OCZ mnh are running pretty well on L042B074 and UD5/f6d :up:

1067mhz 8-8-8 tested an hour Prime 2.64 3.5gb @ 1.05v/1.58v
7-7-6 also ok 32M 1067mhz 1.76v which is ~0.03v higher than same clock on my 920/e760

here's a vista 32m 8-8-8 1097mhz

vtt 1.05v set (1.08v cpuz 156)
vdimm 1.60v set

i set trfc 60, all other subs auto which do look slack for hyper


http://i447.photobucket.com/albums/q...04_OCZ_F6D.png

looking very much forward to your super pi runs :up::up::up::up:

Quote:

---

Originally Posted by cheapseats

my OCZ mnh are running pretty well on L042B074 and UD5/f6d :up:

1067mhz 8-8-8 tested an hour Prime 2.64 3.5gb @ 1.05v/1.58v
7-7-6 also ok 32M 1067mhz 1.76v which is ~0.03v higher than same clock on my 920/e760

here's a vista 32m 8-8-8 1097mhz

vtt 1.05v set (1.08v cpuz 156)
vdimm 1.60v set

i set trfc 60, all other subs auto which do look slack for hyper

---

V nice :up:

Can the chip run faster & still do DDR 2200 CL8 at decent voltages?

Dino..cheers mate :D aiming for sub 6mins ofc .. not sure of highest multi yet.

Quote:

---

Originally Posted by C-N

V nice :up:

Can the chip run faster & still do DDR 2200 CL8 at decent voltages?

---

thanks C-N :D yes for 32m and the vtt is the same, can't say about prime etc. but i've given up on stability tests atm re: LLC1

Just for kicks.

Here's an average set of hypers.

http://chew.ln2cooling.com/P67A-UD7/hypers.jpg

Here's a properly tuned set of psc.

http://chew.ln2cooling.com/P67A-UD7/4%20cores.jpg

Here's a way above average set of hypers tuned properly ( keep in mind i have like 40 gigs of this crap )

http://chew.ln2cooling.com/P67A-UD7/Hypers%20tuned.jpg

Here's cas 6 which needs more testing with a better chip, I can do 1900+ but not at the max of my chip, bclk limits when at max of chip.

http://chew.ln2cooling.com/P67A-UD7/hypers%20c6.jpg

Bottom line unless you got a killer set of hypers you will be no faster than a correctly tuned set of PSC.

Excellent testing,i had this question for a long time,which is faster 6-9-6 or 7-7-7 :) GJ

seems PSC is finally competitive but id still stick to good hypers for absolute best runs, time will only tell as it will depend on max clocks you can get out of ram as SB tends to like high ram clocks at expense of latency, when you compare ratios 16 18.66 21.33 runs get faster with higher clocks even if you drop latency from say 8-8-8- to 9-9-9-

Makes me wonder if a 24.00 multiplier will do wonders with the current arhitecture :)

Quote:

---

Originally Posted by dinos22

seems PSC is finally competitive but id still stick to good hypers for absolute best runs, time will only tell as it will depend on max clocks you can get out of ram as SB tends to like high ram clocks at expense of latency, when you compare ratios 16 18.66 21.33 runs get faster with higher clocks even if you drop latency from say 8-8-8- to 9-9-9-

---

Yah noticed that to dino, still curious as to what is needed 6-6-6 wise, I wasn't fully dialed in for that run.

Really need a chip that does 5.1 for some 5 gig high speed 6-6-6 runs so i have bclk flexibility.

I did get a run off at like 1940 4.8 but to keep my performance table in line it needs to be run at 5 gig.

I think 50% of the fast hyper run came from ability to run 58 TRFC versus the PSC topped out at 68..... the rest was in the subs.

Quote:

---

Originally Posted by Alex-Ro

Makes me wonder if a 24.00 multiplier will do wonders with the current arhitecture :)

---

lol if anyone could get it to even run heheh

my money is on hypers kicking arse on LN2 with about 1100+ clocks at 7-7-6- timings


what people do not seem to grasp is that SB is a lot more efficient than previous gen and ram will not clock as high as previous gen due to extra pressure on IMC

Quote:

---

Originally Posted by dinos22

what people do not seem to grasp is that SB is a lot more efficient than previous gen and ram will not clock as high as previous gen due to extra pressure on IMC

---

Extra pressure on DIMMs you mean in comparison to running the same DIMMs on previous architectures.

sorry yes :up:

Will catch ya on MSN at some point Dino - when we're both awake at the same time...lol

Quote:

---

Originally Posted by dinos22

seems PSC is finally competitive but id still stick to good hypers for absolute best runs, time will only tell as it will depend on max clocks you can get out of ram as SB tends to like high ram clocks at expense of latency, when you compare ratios 16 18.66 21.33 runs get faster with higher clocks even if you drop latency from say 8-8-8- to 9-9-9-

---

That is indeed true. As this run shows.

http://hwbot.org/signature.img?iid=5...se&iehack=.jpg

It was quite a bit quicker than my 7-7-7-21 Run at 2133 MHz

It left me scratching my head for quite a while.

Thats what i've also noticed, loose timings really dont matter with sandybridge compared to X58 where they play more of a important role, Even cas 9-10 @ 2k+ results in fabulous scores/latency..