Overclock.net banner
Cancel
Create thread New Forums

i7 920 @ 3.3 with turbo stabilizing help

Jump to Latest
1 - 20 of 39 Posts
A

Aspiranten

· Registered
Joined
·
18 Posts
Discussion starter · #1 ·
Hi guys, new to this forum and been reading alot here the past couple of days for guides on overclocking since its my first time.

Trying to clock my 920 to a sustainable 24\\7 clock of modest 3.3 ghz, got a vcore of 1.1500 and running prime small fft stable for 3 hours, large fft stable for 1 hour, no errors. Tried LinX now and it gave me an error after 3 passes (41 mins) of 37177 problemsize... No error report what so ever... Left Prime95 to run overnight last night but it gave an error after 2 hours, and one after 3 hours, so i increased Vcore just a notch earlier today. Give it another go today perhaps?

QPI is set at 1.25 (too high? from http://www.overclock.net/intel-cpus/...920-930-a.html )
And PLL is set at 1.82

Havent touched ram timings, theyre set @ stock and 1330 mhz. Uncore is 2x ram. QPI is left at auto.

Temps are 35 idle, 63-65 load (core temps, cpu is 5 degree lower)

Any suggestions?
 
#5 ·
Yeah, leave the pll at 1.8 I haven't had to touch the pll on mine for 4.5Ghz, so you shouldn't have to either.
Next i'd up the vcore to 1.25 in the bios, and disable LLC if its on.
Next, increase the DDR multiplier to give you the full 1600 Mhz ram, reduce the QPI rate to the lowest setting, not auto. Uncore is fine at 2x, Although, you may want to increase the uncore voltage to 1.25v. uncore is labelled as QPI/DRAM core voltage on asus boards.

Once you are sure that is stable, you can if you desire, start lowering the cpu vcore until it crashes when starting linx. Then its best to up it a couple of notches and do a long test on LinX and p95. If it passes say, 30-50 runs of LinX max memory, 5 hours of small fft and 12 hours of blend, then you can say you are stable. Unless you want to do more testing for your own satisfaction.
 
#7 ·
I've had LinX pass 20 runs of 5Gb, but fail p95 in 30 mins. I know, its weird. I wouldn't suggest both if I thought it was pointless.
 
Discussion starter · #9 ·
You sure i need to bump Vcore a full .1 Weidass? Sounds abit high for such a low overclock from what ive read here other people are using.

I bumped it once this morning, after that it didnt fail Prime95 for 3 hours, so i did LinX and it failed after 41 minutes, so i rebumped it, now it did 20 passes of 25000 problem size according to the "stable i7 920 club" thread successfully.

Trying max memory now 20 passes allthough it takes so long for each pass now it`ll take the rest of the evening... (12gigs of ram)

Uncore is at 1.25v, but is it safe to run the rams at 1600 without any cooling on them? I would hate to see all 12 gigs fail on me due to overheating. (first time doing it so going safe and happy)
 
Discussion starter · #13 ·
Quote:


Originally Posted by HunT3R.!
View Post

You Should be fine with 3 hours of p95. it's not like you're ever gonna get 100% running for such a long time...

True, but im also clocking this low for a long cpu life in terms of overclocking, so im thinking its better to be safe than sorry, making sure theres no faults slightly wearing on the die over time.
 
Discussion starter · #14 ·
20 passes of max mem test successfully in 4,4 hrs.

Doing 1600 and LLC off now.

Everst Ultimate is reporting my rams in each DIMM as DDR3 1333 DDR3 SDRAM, whereas the rams boxes and my receits for the rams all show PC3 12800 and 1600... Wich one is correct? Weird looking at those reports...
 
Discussion starter · #16 ·
Ah, sweet. Had to bump VTT to 1.30 so far with the rams at 1600 due to BSOD 124`s at the first test.

Test 3 and going now (blend). LLC off and Vcore 1.200 (1.144 load atm) Uncore 2x and QPI lowest setting (8:9 Uncore)

By the way, anyone else had this feeling when you first started with OC`ing that getting stable is going to be somewhat of a downer, since its so fun getting there? This is goddamn addictive
Image
 
#17 ·
The reason i suggested such a bump in vcore, was that it would pretty much ensure that you would be stable. Its just the way i prefer to work. ie, work down from a slightly higher known stable voltage.
 
#19 ·
1.35v is intel's safe max for the uncore voltage. So, yeah, 1.31 is fine. and vcore, as you say, go with what works
Image
 
Discussion starter · #20 ·
1+ hour in p95 and no probs, 5-10 mins with LinX and BSOD 124... Keep increasing VTT a notch or two? Getting dangerously high i think... Will the NB passive cooler keep up? Only cooling i have on NB is a 120mm Noctua mounted on the side chassis. Half of this airflow reaches NB i hope.

BCLK 166
Ratio 20 (21)
DRAM 1664mhz
UCLK x2
QPI 8:9

Vcore 1.2125
PLL 1.80
QPI 1.31250 (failed)
LLC off
DRAM Bus Voltage 1.64
Spread Spectrum CPU & PCIE disabled
C-STATE disabled
 
1 - 20 of 39 Posts
About this Discussion
38 Replies
8 Participants
Last post:
iNZoW
Overclock.net
posts
28M
members
562K
Since
2004
The premier forum for overclocking experts and enthusiasts. Discuss hardware optimization, custom builds, benchmarking, cooling solutions, and pushing the boundaries of computing performance. From beginner guides to extreme overclocking, join our technical community to master system tuning.
Full Forum Listing
Explore Our Forums
Intel CPUs Hardware News NVIDIA PC Gaming Water Cooling