G.Skill ECO PC3-12800 7-8-7 1.35v in the house!

[jalyst]

the eco stuff is amazing, not sure if anyone else has made anything quite like it yet.

[DStealth]

yeaa

Today tried CL7 and it's not a problem at all
1.62v 7-10-8-27-88-1T
2240mhz

[PiLsY]

Wow! Subtimings at auto in bios? I haven't tested my eco's on sandybridge yet - ive got an early BBSE (bought on release) and a later PSC set (4 months older). They were good but not great on P55...

[PiLsY]

P8P67 Pro loves these sticks. Full auto timings in bios at 2133mhz gives 10-10-10-30-1T and its only taking 1.54v for prime blend. Also gained some bclk (still climbing) over hypers - previously 104.2 max using 1600mhz, currently testing 105.2 @ 2133mhz. These were nowhere near this good on P55/X58. Will tweak timings once I find the mhz limit, but already above 25k read in aida @ 2250mhz 1.6250v. Astonished .

[jalyst]

Amazing to hear they just get better on latest Intel chipsets/MB's, thanks for sharing!
You make me want to get a new MB, damn you!

[PiLsY]

2237mhz 1.60v/1.15v 9-10-9-30-1T with everything else at auto is 4 hours of blend 3200mb stable. 8-10-8 is taking more volts but could be possible, testing at 1.65v/1.175v now. The sticks are also now scaling past 1.58v unlike on P55/X58. I've still only adjusted primary timings, subs will be able to be tightened quite a bit to bring up bandwidth - will post aida when tweaked. For now its just over 25.2k read when prime stable at 2237mhz and will run aida at up to 2262mhz same timings and volts. Suicide stability bandwidth should be good .

Bclk topped out at 105.2 stable and 106.1 boot. That's up 1mhz stable and almost 2mhz boot over my hyper sticks(104.2 stable/104.3 no post). So much for it being all down to cpu!

[Edit : 8-10-8 no go, 9-10-8 fine, struggling with 9-10-7 but its close. Looking forward to trying 6-8-6 @ 1866 tomorrow now ]

[BeepBeep2]

They're not record breaking sticks, but they're pretty good for what they cost.

I'm thinking of getting a Crosshair IV Formula or Extreme board. From what I've seen they can get the most out of RAM/HTT.

M4A89TD and M4A89GTD also clock high ram.

I run hypers, they clock worse IMC wise (not as high of IMC ceiling as PSC/BBSE etc) and I've had my chips over 2000 CL7 benching

[PiLsY]

Well I couldnt stabilise TRP 7 or 8 in the end at 2237mhz, got 20 mins of blend with 7 and about 2 hours with 8. 0x1a Memory Management bsod which doesnt seem to go away even if vccio, vccsa and vddr are raised. Used to be QPI/VTT on p55/x58, may be subtimings on P67. Will investigate later, just looking for rough targets at the moment.

Tried 6-8-6 @ 1866mhz 1.6v - various BSOD's booting. 6-9-6 has been priming 10 mins now, looks like its got some headroom in it.

TRCD 9 is a no go at 2133 for blend stability. Will do various benches up to 2200 though. At a guess 9-10-8 should be stable around 2200 and 9-10-7 around 2150.

These sticks are fast for the timings too - higher bandwidth than hypers by about 1.5k read at the same timings.

[jalyst]

Teh awesome keep probing, I'd love to see the limits of these sticks!
They're already doing very well, considering they're not really geared for massive OC's.

[PiLsY]

They wouldnt do 1600mhz 6-7-6 blend stable, only 1575mhz. Any higher than that and its a memory management bsod in prime. 6-8-6 went to 1685mhz no problem then im bclk limited, so for sandybridge and eco's at least that timing set is a little pointless.

9-10-9 at high mhz is going to be the fastest overall timing set, its a clear 2k mb/s read ahead of everything else.

Got 6-9-6-24-1T up to 1925mhz for 23.5k mb/s read in aida.

So, to sum up blend 4 hour stability:

6-7-6-24-1T approx 1575mhz
6-9-6-24-1T approx 1925mhz
7-9-7-24-1T approx 2050mhz
7-10-7-24-1T approx 2070mhz
8-9-8-24-1T approx 2080mhz
8-10-8-24-1T approx 2100mhz
9-10-9-27-1T 2233mhz+

And voltage point testing:
9-10-9-27-1T 2237mhz @ 1.62v
9-10-9-27-1T 2188mhz @ 1.58v
10-10-10-30-1T 2237mhz @ 1.56v
10-10-10-30-1T 2133mhz @ 1.5v
10-11-10-30-1T 2133mhz @ 1.42v
11-11-11-30-1T 2133mhz @ 1.38v
11-12-11-36-1T 2133mhz @ 1.35v



Now I dont know if this is my PSC or BBSE kit. Going to test the other against the results from this one tomorrow (need a break now!).

[jalyst]

thanks for sharing that mate.

[dejanbinladen]

thx pilsy for results here is mine

7-10-7-27 1T 2133MHz 1.64V

http://www.pohrani.com/?2r/wc/3x7HLv...27-1t-164v.jpg

[Farmer Boe]

Nice timings pilsy! Hopefully I can improve mine Eco's performance like that.

[PiLsY]

Just discovered that the mem slots on the P8P67 Pro react differently to timings. Pair A (closest to cpu) are more stable at lower mhz / lower timings and pair B give you more stable mhz with higher latency. Switched over to pair A and Ive got 6-7-6-24-1T stable at 1612mhz 1.63v. Below TRCD 8 you want the Pair A, above TRCD 8 you want Pair B. At TRCD 8 its all the same. Worth bearing in mind if you try to run 4 x 2gb on this board.

6-7-6-24-1T 1.62v 2 hr prime stable and 22.9k (same subtimings as 2237mhz top run, no tweaking yet) in aida64.

Sandybridge definitely prefers that 21.33x multiplier. I'd guess its linked to whatever passes for uncore speed these days and thats where the extra bandwidth is coming from.

Enjoying playing about with this stuff and discovering new ground. Farmer Boe you look to have better sticks than me as you can do cas 7 2133mhz - hopefully you can get some good results from yours .

[Farmer Boe]

Well I found mine are not prime stable at cas 7 for more than half an hour...it was just a quick tune for the cas 7 at first. Now i'm going to try get the 6-7-6-24 at 1600-1700mhz settings like you did on your kit.

[PiLsY]

Yeah I've noticed its back to the days of using Prime Blend for ultimate stability even though Linx gives me much higher temps I can run all mem Linx for 50 loops a good 50mhz cpu and 30mhz memory clock higher than I can do a 4 hour blend at.

Let me know if you can confirm my findings with the ram slots or if I just have something weird going on!

2237mhz 9-10-9-1T @ 1.62v turned out to be unstable - memory management BSOD late last night. I think that last couple of mhz is just a bridge too far, backed it down to 2233mhz at 1.60v and its been priming now for 12 hours. Finding slots 1 and 3 worked better at low latencys than 2 and 4 put me back a few hours yesterday, so going to tweak subtimings now and see if that leads me anywhere. After that i'll swap the other sticks in and see if they go further on the same profiles, then its a heatspreader lift as I cant remember which set of sticks are PSC and which are BBSE - lol.

[Farmer Boe]

Well last night I got my sticks to 6-8-6-21 1T @ 1696MHz prime and ibt stable. I think mine have PSC ic's as I cannot get them down to 6-7-6-X on any setting besides dropping them to 1333MHz, and who wants to do that? I have the black heatspreaders but I'm not sure if it confirms the ic's either way. I'm going to try for CL8 stability at 2133MHz tonight and see how it goes and perhaps I'll try swapping slots in my P8P67 Pro to see if that helps my tune.

[PiLsY]

See what 6-9-6 will do for you man. Remember you can drop bclk below 100 so you could run say 21.33x94 bclk for around 2000mhz.

I still seem to be struggling with the odd thread drop in prime. Coupled with the memory management bsod I wonder if I have subtimings off somewhere. No idea where to start though coming from a gigabyte background as not much has the same name.

[skuldarin]

anyone tried doing 12gb of this stuff?

i have 6 sticks of this that i originally had before going to my Dominators. i even upgraded their cooling with Xigmatek heat spreaders, of which i modifidied... well more or less stripped the thin metal sideways fins off because they weren't made for down-wards air cooling, but side to side cooling. so now that have a bunch of towering alluminum points. works like a charm though.

EDIT: damn, had a look. both my dominators and my ecos have PSC chips. YAY for tRCD nightmares!

[PiLsY]

Spreaders lifted, they were the PSC sticks. Release day BBSE sticks going in now.

These have a chance at 8-9-8 2133mhz - they do 2080mhz 8-9-8 1.58v on the P55-UD5. Fingers crossed!

[jalyst]

Thanks for sharing dude,
Sounds like the BBSE may do better for raw, unadulterated, OC'ing.
Might lift my spreaders to see what "exactly" I've got!
About to finally use all 4x 2GB sticks that I bought about 1yr ago for my P7H57D-V EVO!

[Farmer Boe]

I'm tempted to lift the heat spreaders off mine now to confirm my suspicions. I'm 90% sure I have PSC ic's on mine.

[PiLsY]

BBSE dont work well at all. Gave up. I had windows stability up to 2200mhz, prime just dropped a thread within 3 minutes.

PSC are the way to go.

Also my P8P67 Pro now needs to go back. I put AI Suite on (after multiple patches to get it installed) including fan xpert. As I rebooted after the install all my fans turned off. A full cmos clear has got the cpu and mem fan headers back (but q-fan no longer works, full speed only) but the chassis fan headers are not working at all. Coupled with my hypers not being stable and the lack of general stability I think I may try and switch for the UD4. Most problems I've come across seem to be bios related on Sandybridge - I cant help but wonder if UEFI isnt still a little immature. I guess only getting the Giggy board would answer that really.

[jalyst]

Duly noted...

TBH I'm a little paranoid about prying the heat-spreaders off mine.
I'm no hard-core overclocker, so no "biggy" if I've got BBSE

I'm at least 6mth off upgrading to SB

[PiLsY]

The trick is to not use much pressure - let the tape slowly peel itself off. Use the tips of your thumbs along the bottom (gold finger) edge. Find a side thats lifting easier than the others and go for it. If they're all stuck down well just forget it .