S2E - Q9550 - 2x2GB DDR3 7-7-7-20 1600Mhz - Trying to get stable 3.4Ghz OC

[Bobly]

UPDATE IN POST #11



Hey Guys,


I've been trying to run stable on 3.4Ghz for a couple of days now and I just can't seem to find what's missing



Specs and info:
First attempt at building my own rig (went fine), water cooling it (also going very nicely ) and overclocking it (touch and go :P).

Specs:
Vista Ultimate 64Bit
Asus Striker 2 Extreme
Intel Q9550 2.83Ghz
2x2GB Corsair DDR3 XMS3 DHX PC12800 7-7-7-20
EVGA GTX 280
WD 300GB Velociraptor
Zalman ZM-1000HP

Cooling (H20):
DDC3.2
TFC 480 w/ Zalman ZM-F3 fans (idle set to ~rpm1000, load to ~1800rpm)
D-Tek Fuzion V2
EK-FC280
EK Res 250

Pictures of rig:



Temps and voltages on load (3.2Ghz - Stablish):



Bios Settings:
CPU FSB: 1505
Vcore: 1.35
Vnb: 1.5
Vsb 1.5 (Auto)
MEM FSB: 1505
Timings: 7-7-7-20
Voltage: 1.84V
(Anything not mentionned it on Auto, speedstep disabled, loadline calibration enabled)


Now I'm nearly stable on those settings at 3.2Ghz (60% pass on 10 Linpack runs). A bit of fine tuning and it should be stable.



Anyone have any tips or suggestions to push this chip a littler further?
I'm still not familiar with what the voltage settings other than Vcore, NB, SB and Mem do so I'm open to suggestions



One addition to my system I'm considering is getting an extra 40mm fan to place on the top mosfet heatsink on the Asus Fusion Block system to help keep that NB under control:




Edit: Stable on 3.2 by raising NB Voltage to 1.54, NB temp now at 59, I don't think I can push it much further without added cooling, I remember reading it gets less stable above 60C :/

[Bobly]

Okay clean sheet of paper, starting from scratch, 20Mhz at a time.
Bios settings on auto got me from stock 1333 to 1460.
1480 - Had to raise the NB to 1.6 (Everest, 1.46 in BIOS)
1500 - Working on stability, 80% IntelBurnTest pass with CPU Voltage to 1.312 (CPU-Z, 1.35 in BIOS) and NB to 1.46 (Everest, 1.46 in BIOS) CPU 60C NB 58C

Anyone have any tips as to what might increase stability? Ram is running at it's stock settings (not auto, set them in the BIOS).

If it's any help I just tried Prime and cores 3 and 4 both failed within a minute of each other (wasn't looking exactly when it happened).

[dan7777]

your temps seem a little high for those voltages with watercooling.... i wud guess you need anthor notch to get to 3.4 you shud easily be able to get 3.8-4ghz most people with those quads get 4ghz on air cooling 2 mate.

[Bobly]

Stable FSB 1500 by raising NB to 1.5V, memory to 1.9V from 1.86 (Everest was saying they're getting 1.9V but I'm supposing it's voltage ratings aren't very accurate >.< It was stable at stock CPU speed but heh) and dropping CPU back down to 1.3V in bios (CPU-Z 1.264).

I think my temps were high on such a high clock because my voltage is currently a lot higher than it needs to be >.< It was the bloody memory, gonna reboot and start attacking 1520 ^^

Stressing on 1.264V and temps are maxing at 52C.

[dan7777]

that stil seems high man temps ive got my vcore set on mine 1.56vcore nb 1.50 memory 2.2v my max temps are 65c 24/7

[Bobly]

Arf it's definitely a NB/RAM issue, dropped back down to 1400 FSB after a night of 0 achievements and decided to let Prime running, so far it's running fine on Blend with memory set to 1.9V (Everest, Bios 1.86V) 7-7-7-20 (manufacturer rating) - Btw temps now maxing at 46 +/-1C on Vcore 1.264V and fans set to 1000rpm.

So basically here's the current situation:

Stock: CPU and Ram work fine. Prime blend mode no errors.
Overclock: CPU Prime stress test with small FFTs always seems to pass but with a FSB of 1500 or more I can't get stable on blend mode which seems to suggest a problem with the NB/Ram.

I've tried raising the NB Voltage up to 1.6V (62C stock cooler) and increasing the ram timings to 9-9-9-24 to relieve stress on the NB but still no go :/

Going to prime overnight with FSB 1400, everything else stock except ram settings set to manufacturer specifications and see how that goes... At least that should rule out faulty ram...



Any tips for getting that NB stable?

[barfastic]

bobly, im having similar difficulties with my S2E. We have the same ram, but i ahve a E8600, which i cant get above 4ghz. Ive tried pushing the Vcore up to 1.45, even 1.5 at some points and it still wasnt stable... tried increasing both NB and SB voltages, still unstable... im kinda stuck in a predicament, trying to figure out what has gotten me stuck, but i just cant figure it out. Im thinking it might be my ram... on stock cpu settings when i input the timings manually it runs, but if i raise the cpu settings even by 1fsb (from 333 to 334) it becomes unstable... Should i try loosening the ram timings?

[Bobly]

I'm afraid I have no idea >.< This is my first attempt at overclocking and I'm getting nowhere :/

[barfastic]

lol... i have overclocked before, but ti wasnt this ahrd. i had a Q6600 at 3.6 stable on water, with a P5E3 Deluxe board. This one has been busting my chops...

[Bobly]

Urgh...


Faulty ram? ... >.<

[Bobly]

Okay so there's been a MASSIVE jump in progress!

I found out one of the ram modules is faulty, so I'm going to have to RMA those tomorrow and go a couple of days without a PC... Actually I might RMA them after the weekend :P
In the mean time I've been playing with 2GB and came across something interesting...

I could not, for the life of me, get stable at 1500FSB... Or 1600FSB... so I eventually thought sod it and jumped to 1733FSB, Vcc was already at 1.375 at this point so I figured it might just work and guess what, Stable!
Following the same stroke I jumped up to 1866FSB and once again, booted and ran stable for 20mins, much longer than it had previously...

Here's where I got a bit greedy and tried 1920FSB, it booted and ran stable for a couple of minutes so I rebooted again at 2000FSB, this however prooved to be the reasonable limit at this point. Vcc set to 1.425 and while it would POST it would usually crash during the Vista loading screen.

I decided to drop down for a bit, and while I could test on 1920 I preferred to drop down to 1866 and find what Vcc and Vnb I could use while still remaining stable.
Currently 10h Prime95 stable on 1866FSB Vcc 1.360V (CPU-Z) and NB at 1.5V. Haven't started touching the other voltages yet but I think I might have to start giving them a bit of a fiddle soon.
I don't think 1920FSB would be too much out of reach, who knows, with better NB cooling and remounting the CPU I might even be able to reach 2000FSB (Okay I'm getting ahead of myself quite a bit here :P)